WO2020043030A1 - Data credibility evaluation and calibration method for air pollution monitoring device - Google Patents

Abstract

The present invention introduces a credibility weight factor for evaluating environmental sensor data, evaluates the credibility thereof, and then calibrates same after evaluation. The credibility weight factor is influenced by the distance between a reference site and a calibrated station or a geographic position center, the geographic position of the reference site, the evaluation situation of another site on the reference site, the stability of the reference site and other influence factors, such that urban pollution monitoring data is more reliable, and the calibration of the monitoring site is also more accurate.

Description

Data credibility evaluation and calibration method for air pollution monitoring equipment Technical field

The invention relates to an air pollution monitoring equipment data credibility evaluation and calibration method, and belongs to the field of environmental monitoring.

Background technique

The monitoring indicators of atmospheric pollutants in environmental monitoring are sulfur dioxide, nitrogen oxides, ozone, carbon monoxide, PM ₁ (particles with aerodynamic particle size less than 1 micron), PM _2.5 (particles with aerodynamic particle size less than 2.5 micron) in the atmosphere ), PM ₁₀ (particles with aerodynamic particle size less than 10 microns), PM ₁₀₀ (particles with aerodynamic particle size less than 100 microns), and VOCs (volatile organic compounds) or TVOC (total volatile organic compounds). The atmospheric environment monitoring system can collect and process the monitored data, and timely and accurately reflect the regional ambient air quality status and changes.

At present, the atmospheric environment monitoring equipment mainly includes fixed monitoring stations and mobile monitoring equipment. The current fixed monitoring stations are mainly divided into large fixed monitoring stations and small stations. Mobile monitoring equipment mainly includes special atmospheric environmental monitoring vehicles, drones and handheld devices. The aforementioned small monitoring stations and handheld devices all use air quality sensors to measure pollutants in the atmosphere. Small sensors have the characteristics of low cost, miniaturization and online monitoring, and can be used on a large scale. The air quality sensor itself may cause errors due to inconsistencies between the measured values and the true values for various reasons. Compared with large-scale precision instruments or manual monitoring methods, air quality sensors also have lower accuracy, poor stability, large errors, and require frequent calibration.

The laser scattering method for air pollution particulate matter sensors has a broad market prospect because of its low cost and portability. However, the portable analysis device using the scattering method has disadvantages such as poor measurement consistency, large noise, and low measurement accuracy. The core device is easily affected by various environmental factors and fluctuates, which easily causes misjudgment.

When the sensor data changes suddenly and sharply, if the monitoring instrument can intelligently determine whether the change is caused by a sensor failure or sudden pollution, it will greatly improve the reliability of the data and is of great value for ensuring the quality of environmental monitoring data. When the equipment is inaccurate, it can be automatically calibrated to improve the accuracy of monitoring and the online rate of data can be greatly improved. It is of great value for continuous monitoring required for haze control. At the same time, it can save manpower and material resources in equipment maintenance and reduce waste of social resources.

In the field of atmospheric environment monitoring, high-precision monitoring equipment is expensive and huge. At present, for the monitoring of atmospheric environment at the city or large area scale, the state-controlled stations set by the country belong to such high-precision stations, but their number is small, the coverage is limited, the site geographical location is limited, and the state-controlled The operation of the station is greatly affected by equipment conditions and maintenance conditions, and data output is unstable. At present, the environmental data of cities or large areas published by the state are obtained based on the average of the monitoring data of these state-controlled sites or through simple calculations. However, the atmospheric environmental data monitored by these state-controlled sites cannot represent the atmospheric environmental data of the entire city or region.

Maintaining good equipment is a necessary condition for obtaining accurate and reliable monitoring data. Calibration of the equipment is the key to obtaining reliable and accurate data. At present, data from national control stations or super stations are often used as calibration reference data.

The data from the national control site is used as the benchmark data, and the technology for calibrating micro monitoring stations or mobile monitoring stations is a common method adopted by enterprises. However, using only a small number of national control stations as a reference cannot represent the accurate value of the city or region. Further use of data from these national control stations to average or simple calculations for calibration of other micro monitoring stations or mobile monitoring stations is even more unreliable. These micro-monitoring sites or mobile monitoring site equipment may be more inaccurate after data calibration at the national control site, making the results of the micro-monitoring site or mobile monitoring site low or too high. The state control station is not stable. Due to equipment reliability, operation and maintenance conditions, power supply and other reasons, the state control station cannot output normal data.

Summary of the Invention

Earlier application

PCT / IB2019 / 051243

PCT / IB2019 / 051244

Explanation of terms

α data set: monitoring data of the base station (national control station, city control station, separately set calibration station); α1 indicates the data or data group at the reference station at T = 1; A1 indicates a national control station at T = Data at 1 time.

β data set: the monitoring data of the fixed monitoring station, β1 indicates the data or monitoring data set of the fixed atmospheric meshing microstation at T = 1; B1 indicates the data of the fixed atmospheric meshing microstation at T = 1.

γ data set: monitoring data from mobile monitoring stations, γ1 represents the data or monitoring data set of the fixed atmospheric gridding microstation at T = 1; Y1 represents the data of the fixed atmospheric gridding microstation at T = 1.

Fixed monitoring station: A station with the ability to monitor the atmospheric environment can be a national control station, a calibration station, or a grid-based microstation.

Mobile monitoring station: A monitoring station equipped with atmospheric environmental monitoring equipment and capable of moving. It can be a social vehicle equipped with miniature monitoring equipment, or it can be a professional atmospheric environment monitoring vehicle.

Contrast coefficient: A quantity that indicates the degree of linear correlation between variables, usually expressed by the letter η.

Calibration coefficient: In the present invention, the calibration coefficient refers to a correction coefficient used to calibrate and correct the deviation of the data set of the sensor.

Particulate matter measured by light scattering method is susceptible to environmental factors, such as humidity and other factors. There are also many ways to improve sensor accuracy.

The current monitoring station calibration method mainly uses regular manual maintenance. Staff go to the site to clean up and maintain the equipment, and carry standard equipment and standard gas to manually calibrate the sensors on site. Or simply make coefficient corrections to the monitoring equipment. These calibration methods have different levels of problems such as inaccuracy, complex calibration and high cost.

In view of the above-mentioned deficiencies, the present invention provides a method for evaluating and calibrating the reliability of data of air pollution monitoring equipment, introducing a reliability weighting factor to evaluate its reliability, and then performing calibration after the evaluation. The credibility weight factor is affected by the distance between the reference site and the calibrated station or the geographical center, the geographic position of the reference site, the evaluation of the reference site by other sites, the stability of the reference site, and other influencing factors. Makes the urban pollution monitoring data more reliable, and also makes the monitoring station calibration more accurate. At the same time, multiple data are used for mutual calibration and comparison between sensors to achieve complementary data deviations and mutual verification to improve the reliability, consistency, accuracy and life of the sensors.

For the monitoring data of the monitoring site, a credibility weight factor is introduced to evaluate its credibility, and then the calibration is performed after the evaluation. The credibility weight factor is affected by the distance between the reference site and the calibrated station or the geographical center, the geographic position of the reference site, the evaluation of the reference site by other sites, the stability of the reference site, and other influencing factors. The credibility weight factor is positively related to the distance factor, geographical location factor, other site evaluation factors, and stability factor.

F _c ∝f _d , F _c ∝f _l , F _c ∝f _e , F _c ∝f _s

Calculation method of credibility weight factor:

F _c = f (f _d , f _l , f _e , f _s )

A credibility weight factor calculation formula is:

F _c = f _d × f _l × f _e × f _s

Credibility weight factor: F _c (factor credibility)

Distance factor: f _d (factor distance)

Geographical factor: f _l (factor location)

Other site evaluation factors: f _e (factor evaluated)

Stability factor: f _s (factor stability)

After calculating the credibility of the base station, you can use the credibility to adjust the data of the base station and use it for calibration calculation of other monitoring stations; or to rank or set the credibility of the base station and exclude it after ranking Base station data with lower credibility or excluded base stations with a certain credibility, and then use the filtered base station data to perform calibration calculations on other monitoring stations.

Distance factor

The credibility of the data monitored by a monitoring site decreases with increasing distance from the monitored area to the monitoring site, and the same applies to the monitoring data of the high-precision site. When evaluating the air quality in a certain area, the data fusion results of multiple sites in the spatial range are considered. The present invention proposes an evaluation describing the effective area of a monitoring site. The method includes several weighting factors to indicate the spatial impact weights of the data collected by the data site when real monitoring is performed in an area in the city, and then describes the data impact range of the site. Or data valid range.

In the calibration process, the distance factor between the reference station and the calibrated station can also be considered, and a distance factor is introduced. The distance factor f _{d is} used to consider the influence of the distance factor between the monitored point and the monitoring station on the reliability of the monitoring data. The distance factor can be normalized by the inverse ratio of the distance from the geometric center point of the area to each monitoring point. The distance factor can be obtained by the data obtained by monitoring stations in a certain area. The other embodiment of the distance factor is a specific location. The pollution data is composed of monitoring data from several monitoring stations that are close to each other. These monitoring data can have different weights for pollution data at a specific location. The weight is normalized by the inverse ratio of the distance from the specific location to each monitoring point. The weight is the distance factor.

In the calculation of f _d , d represents the distance from the geometric center point of the area to each station in the area or the distance from the specific position to each monitoring point, and the set value of the distance is represented by A. Within the set distance A, the distance factor is 1. After exceeding the set distance A, the farther the distance is, the smaller the weight of the monitoring station data is, and the closer the distance is, the greater the weight of the monitoring station data is.

The distance factor calculation formula is:

d represents the distance from the geometric center point of the area to each station in the area or the distance from the specific location to each monitoring point.

The parameter κ is a distance weight parameter. In general, κ = A.

Geographical Factor

In the actual monitoring process, there may be pollution source factors around the monitoring site and it will affect the monitoring results. Therefore, such sites need to be given geographical location impact evaluation factors. Factors are related to factors such as the distance to the pollution source and pollutant emissions. Therefore, in the process of calculating f _l, a factor f _le for assessing the pollution degree of the pollution source around the monitoring site will be involved, and a factor f _{ld for} assessing the distance between the monitoring site and the surrounding pollution sources (see geographical location pollution degree factor and geographical location pollution distance factor for details). The relationship table, the set value in the relationship table can also be adjusted according to the actual pollution situation), evaluate the factor f _{lo of} other influencing factors of the pollution sources around the monitoring station (see the other influencing factors table of pollution sources for details).

Geographical factor calculation method:

f _l = g (f _le , f _ld , f _lo )

A formula for calculating the geographical location factor:

f _l = f _le × f _ld × f _lo

Geographical pollution factor: f _le (factor location emission)

Geographic location pollution distance factor: f _ld (factor location distance)

Other factors of location: f _lo (factor location other)

Without considering other factors, the relationship between the pollution degree factor of geographical location and the pollution distance factor of geographical location is as follows:

Other factors that need to be considered in geographical factors:

Building occlusion: If there is a large building construction occlusion within a certain range around the base station, the smaller the factor factor is.

There are places in the surrounding environment that can affect the diffusion of particulate pollutants: there are forests, parks and other places that may affect the spread of pollutants within a certain range around the base station, and facilities that can reduce the concentration of pollutants. The influencing factors should be appropriately reduced. .

Long-term influence on wind direction: The base station is located in a long-term fixed wind direction area, which may cause the fixed station to fail to represent the air quality in the measurement area. The factor factor should be appropriately reduced.

Pollutants discharged from surrounding pollution sources are not the main pollutants monitored by the monitoring station: there are pollution sources around the monitoring station, but the pollutants emitted by the pollution source are not the primary pollutants monitored by the monitoring station.

Other site evaluation factors

During the long-term monitoring process, the data obtained by the Chinese control station will also be affected by factors such as equipment aging and the reliability will change. Therefore, it is necessary to use data from other nearby stations (national control stations, fixed micro stations, vehicles) to analyze this. One point of the national control station to evaluate, evaluate its accuracy, and give weight.

For a single national control station, the reliability of the data generated by the site at a certain moment cannot be determined, and only extreme anomalies can be ruled out. When the data change trend of the site from other surrounding stations of the same level is significantly different, one reason may There is a pollution source nearby. Another reason may be that the monitoring equipment at the site is abnormal. At this time, it is necessary to verify that the data of the site belongs to the former case or the latter case by using other equipment densely arranged near the site. If other devices nearby such as a fixed micro station or a mobile monitoring station have similar data change trends to the site, the site data is credible, and conversely, the site data credibility is reduced. Therefore, other site evaluation factors f _e need to be set to give weight to the monitoring station data for this influencing factor.

Other site evaluation factor calculation formulas:

与该监测站点监测值M进行如下运算得到比值ε，则ε的大小即可表示附近的其他设备如固定微站或移动监测站与该站点所得到数据是否有相近的数据变化趋势，从而根据ε以及上述f _e(ε)关系式可得到该站点针对此影响因素的权重。 The specific method is: take data from several monitoring stations within a certain distance around the monitoring station, which can be within 10 kilometers, and average them. And average

Perform the following calculation with the monitoring value M of the monitoring station to obtain the ratio ε, then the magnitude of ε can indicate whether other nearby devices, such as a fixed micro station or a mobile monitoring station, have a similar data change trend with the data obtained by the station. And the above f _e (ε) relationship can get the weight of the site for this influencing factor.

代表该监测站点周边10公里范围内的若干监测站点数据的平均值，什么平均值。 formula

Represents the average value of data from several monitoring stations within 10 kilometers around the monitoring station, and what is the average value.

M in the formula represents the monitoring value of the monitoring station.

Stability factor

There are two main extreme situations in the current state-controlled site operation. One is that the site data is significantly abnormal. For example, the PM10 data is less than the PM2.5 data. At this time, the PM10 data at the site will be manually checked and screened. There is no data upload after power on or just after power on; in addition to the above two cases, there may be other reasons such as network abnormalities that cause data abnormalities. When the above extreme situations occur, no matter the operation and maintenance reasons or the equipment failure, the reliability of the equipment will decrease during this period of time. You can set a reliability factor, which can be expressed by the cumulative number of abnormalities in a period of time, such as Counted monthly, the initial value is 1, and the reliability factor decreases by 0.1 each time. During the calibration, whether the national control station participates in the calibration can be determined according to the reliability weight of the national control station.

Calculation formula of stability factor:

n is the number of times that the site has abnormal data within a certain period of time. The abnormal situation and judgment are as follows. A period of time can be one month, one week, one day, and other time periods.

abnormal situation determination No data output due to power failure No output for more than 1 hour No data output due to maintenance reasons No output for more than 1 hour Data failure caused by network failure No output for more than 30 minutes, Data exception: PM _2.5 data is greater than PM ₁₀ data Data abnormal time exceeds 1 hour

Weight factor method for judging credibility

Utilize the introduced credibility weight factor to evaluate the credibility of its site or data. The evaluation method can be a way of directly ranking the credibility weight factor, or a threshold limit.

The direct ranking method is to arrange the credibility weight factors from large to small. The closer the credibility weight factor is to 1, the higher the ranking, the more credible the site or data. Select the top 10%, 20%, or a certain percentage of the credibility weighting factors, or exclude the bottom 10%, 20%, or a certain percentage of the credibility weighting factors, the selected or the remaining after screening The site or data corresponding to the credibility weighting factor can be used for calibration calculation and is valid data.

The way to limit the threshold is to set a certain threshold (thresholds can be 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, etc.), select a credibility weight factor that exceeds this threshold, or exclude below this A threshold credibility weight factor, the selected station or data corresponding to the credibility weight factor remaining after screening can be used for calibration calculation and is valid data.

Calibration calculation method

Contrast coefficient calculation formula:

或

or

Base station correction calculation formula:

y _c = F _c × y

Apply the normalized base station correction calculation formula to modify the base station data, and the normalization calculation formula:

Calibration formula:

y _c is the baseline data after correction or screening

y is the benchmark data. The benchmark data may be unprocessed data or revised or filtered benchmark data (y _c )

x is the data to be calibrated

x ′ is the data after calibration

η is the contrast coefficient

η _c is the corrected contrast coefficient

c is a calibration coefficient, and c may be η, η _c, or a contrast coefficient after other mathematical operations.

In the case where there is only one reference station that meets the standard, the correction of the reference data is calculated as follows:

y _c = yF _c × (xy)

How to obtain the stability coefficient λ:

A) The stability factor λ is the ratio of the number of base station data to the total number of base station data in the set interval. If λ is greater than the set percentage (the set percentage can be 80%, 90%, and other percentages), the base station data set is considered stable. A higher λ indicates a more stable data set.

The setting interval is the range given to the reference data within the set T time range. The mathematical expression of the setting interval is (Yu × Y, Y + u × Y). Y can be the average value of the base station data within the T time range, Median, mode and other statistical methods, u is the interval coefficient.

B) The stability coefficient can also be related to the variance of the reference data in the set interval for the set T time range

If the variance of the reference data within the set T time range> the variance set value B, it does not fall into the set interval.

If the variance of the reference data within the set T time range is less than the variance set value B, it falls into the set interval.

C) The stability coefficient can also be related to the standard deviation of the reference data within the set T time range.

If the variance of the reference data within the set T time range> the standard deviation setting E, it does not fall into the set interval.

If the variance of the reference data within the time range of the setting T is less than the standard deviation setting E, it falls into the setting interval.

For mobile monitoring, it needs to have sufficient credibility for calibration and punctuality.

When the mobile monitoring station adopts a redundant multi-sensor design, the credibility of the mobile monitoring station will be greatly improved.

The steps for multiple calibrations of mobile monitoring stations based on the regional data of a single reference station are:

1) Judging the credibility weight factor, applying the method of judging the credibility weight factor, using the monitoring data of the reference station corresponding to the selected credibility weight factor as the calibration basis, or using the remaining credibility weight after screening The monitoring data of the reference station corresponding to the factor is used as the calibration basis; or the credibility weight factor of the reference station is evaluated. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid, As a basis for calibration.

2) Choose a reference station that meets the standards as a reference. A mobile monitoring device may pass through the same reference station multiple times. Each time this device passes this reference station, a comparison is performed to obtain the η value of the comparison.

3) After recording a set number of η values (the set number can be 10, 50, 100, etc.), statistically calculate all the η values obtained from the record, such as average, normal distribution value, approximation , PID and other mathematical methods. Finally get the average value after statistical calculation

和校准计算方式中的公式再对该台移动监测设备进行校准。 4) Use the final average

And the formula in the calibration calculation method to calibrate the mobile monitoring device.

The steps for multiple calibrations of mobile monitoring stations based on the regional data of multiple reference stations are:

1) Judging the credibility weight factor, applying the method of judging the credibility weight factor, using the monitoring data of the reference station corresponding to the selected credibility weight factor as the calibration basis, or using the remaining credibility weight after screening The monitoring data of the reference station corresponding to the factor is used as the calibration basis; or the credibility weight factor of the reference station is evaluated. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid, As a basis for calibration.

2) In the actual monitoring and calibration process, a mobile monitoring device may pass through the area covered by multiple reference stations multiple times. Every time this device passes this area, it is compared with the average reference data of the passing time in this area to get the corresponding value of η. The reference value of the region can be obtained by using the normalized calculation method.

3) After recording a set number of η values, perform statistical calculations on all the η values recorded, such as average, normal distribution value, approximation, PID and other mathematical methods. Finally get the average value after statistical calculation

和校准计算方式中的公式再对该台移动监测设备进行校准。 4) Use the final average

And the formula in the calibration calculation method to calibrate the mobile monitoring device.

The procedure for calibrating other fixed stations based on the area data of a single base station is as follows.

1) Judging the credibility weight factor, applying the method of judging the credibility weight factor, using the monitoring data of the reference station corresponding to the selected credibility weight factor as the calibration basis, or using the remaining credibility weight after screening The monitoring data of the reference station corresponding to the factor is used as the calibration basis; or the credibility weight factor of the reference station is evaluated. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid, As a basis for calibration.

2) The reference station that meets the standard is used as a reference, and every interval, the reference station data is used as a reference to perform a comparison with the fixed station to be calibrated to obtain the corresponding η value.

3) After recording the number of η values, perform statistical calculations on all the η values recorded, such as average, normal distribution value, approximation, PID and other mathematical methods. Finally get the average value after statistical calculation

和校准计算方式中的公式再对待校准固定站进行校准。 4) Use the final average

And the formula in the calibration calculation method, and then calibrate the fixed station to be calibrated.

The procedure for calibrating other fixed stations based on the regional data of multiple reference stations is as follows.

1) Judging the credibility weight factor, applying the method of judging the credibility weight factor, using the monitoring data of the reference station corresponding to the selected credibility weight factor as the calibration basis, or using the remaining credibility weight after screening The monitoring data of the reference station corresponding to the factor is used as the calibration basis; or the credibility weight factor of the reference station is evaluated. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid, As a basis for calibration.

2) In the actual monitoring and calibration process, a mobile monitoring device may pass through the area covered by multiple reference stations multiple times. Every time this device passes this area, it is compared with the average reference data of the passing time in this area to get the corresponding value of η. The reference value of the region can be obtained by using the normalized calculation method.

使用最终得到的平均值

再对该台移动监测设备进行校准 3) After recording a certain number of η values, perform statistical calculations on all the η values recorded, such as average, normal distribution value, approximation, PID and other mathematical methods. Finally get a statistically calculated average

Use the resulting average

Calibrate the mobile monitoring device

和校准计算方式中的公式再对待校准固定站进行校准。 4) Use the final average

And the formula in the calibration calculation method, and then calibrate the fixed station to be calibrated.

Calibration execution conditions

的计算(设定数量可以是是10次、100次等，也可以是每小时一次、每5小时一次等)。除此之外，在一个校准周期内所获取η值应尽量均匀。例如每月执行校准操作一次的话，则每天至少需要得到一个

值；同时每次η值应当在均匀分散的时间内取得，如获得了12组对比数据，这12组对比数据是每隔两小时获得1组对比数据，而不是在1个小时内集中获得的12组对比数据。 The time range of the data used for calibration needs to be limited. The process of the mobile monitoring device performing the calibration operation is periodic (for example, once a month). Between calibration periods, the device should get a set number of valid η values for

Calculation (the set number can be 10 times, 100 times, etc., or once every hour, every 5 hours, etc.). In addition, the value of η obtained in a calibration cycle should be as uniform as possible. For example, if you perform a calibration operation once a month, you need to get at least one

At the same time, each time η value should be obtained within a uniformly dispersed time. For example, if 12 sets of comparison data are obtained, these 12 sets of comparison data are obtained every 2 hours instead of 1 hour. 12 sets of comparative data.

Excluding special extreme cases

Some data under special circumstances need to be excluded, such as extreme weather events (storm, snowstorm, etc.), high humidity and high temperature. If the monitoring process is affected by weather and other conditions, and the data has extreme values during monitoring, the data comparison calibration will be suspended during these time periods, because the η value in these time periods will be different from the η value obtained in most time periods. The difference causes the problem of reduced calibration accuracy.

Special extreme case condition humidity Humidity> 90% humidity Humidity <90% temperature Temperature ＞ 40 ℃ temperature Temperature <40 ℃ extreme weather Sandstorm, heavy rain, snowstorm, etc. PM _2.5 extremes PM _2.5 ＞ 500 PM _2.5 extremes PM _2.5 ＜ 20

Multi-sensor monitoring equipment calibration

When the station to be calibrated is a group of equipment containing two or more monitoring units, each monitoring unit can be calibrated separately using the reference station. You can also use the base station to calibrate one of the monitoring units first, and then calibrate the other units from the calibrated unit.

Based on α data set, calibrate β data set and γ data set

The first calibration method proposed by the present invention is to calibrate the β data set and the γ data set based on the α data set. In the case where the β data set and the γ data set meet the uniformity requirements, the base α data set is determined by analyzing the data of the α data set. Methods for analyzing alpha datasets include direct average method, average method after removing the highest and lowest values, Kalman filter, Bayesian estimation, D-S evidence reasoning, artificial neural network and other methods.

After the benchmark alpha data set is determined, the beta data set is compared with the benchmark alpha data set to obtain a calibration coefficient for the beta data set, which is used to calibrate the beta data set. Similarly, by comparing the γ data set with the benchmark α data set, a calibration coefficient of the γ data set is obtained, which is used to calibrate the γ data set. The comparison method can be linear calibration, non-linear calibration, or other calibration methods.

During the calibration process, multiple calibration coefficients are generally calculated, and calibration coefficients whose coefficients differ by less than a certain value are taken as valid calibration coefficients. The average of these valid calibration coefficients is used as the final calibration coefficient to calibrate the calibration object.

The calibration coefficient may also need to take into account the spatial distribution. The calibration coefficients of the β dataset can be weighted according to the distance from the β site to the α site. The closer the distance is, the greater the weight; if the β site is within a certain distance from the α site, the weighted average value is used as the calibration target accurate value. For the γ site, data within a certain distance from the α site is taken as valid data to participate in the calibration calculation.

The calibration coefficient of the γ data set can also be determined according to different data intervals of the data, that is, multiple calibration coefficients are set in different data intervals. In the selection of calibration coefficients in different intervals, direct average method, average method after removing the highest value and the lowest value can still be used.

Based on the gamma data set, calibrate the beta data set

Based on the calibrated gamma data set, the beta data set is calibrated. The gamma data set is considered to be a valid reference value after calibration.

When the γ mobile station passes the β fixed station within a set distance, the data set of the β site and the data set of the passing γ site are used to calculate the calibration coefficient. The calibration method can select linear calibration or non-linear calibration. The distance can be 500m, 1km, 2km, 5km.

After ranking the β data set and the γ data set with accuracy, the high accuracy device is calibrated to the low accuracy device.

The third calibration method proposed by the present invention is to calibrate the β data set and the γ data set by accuracy, and then calibrate from a device with high accuracy to a device with low accuracy.

Schematic table of beta data set and gamma data set

The β data set and the γ data set are compared with the α data set to obtain an accuracy index. The comparison method may be a correlation coefficient, a ratio average, and the like. After obtaining the accuracy index, the accuracy is ranked from high to low, and the lower-ranked data set is calibrated. The calibration method uses the first method. Recalculate accuracy after calibration to rank.

For β fixed stations, select α national control stations within a certain range for accuracy calculation. In the case where there are multiple national control stations in a certain range, the accuracy can be the average of multiple national control stations, or it can be calculated by weighted average based on the distance as a weight. In the case where there is no national control station within a certain range, Accuracy calculations are performed using the average of the alpha data set for the entire city. For the γ mobile station, when the γ mobile station moves to a certain range of the α national control station, the accuracy calculation is performed. After ranking, the higher ranked data set is compared with the lower ranked data set, the calibration coefficient is calculated, and the higher ranked data set is used to calibrate the lower ranked data set.

Method for collaborative work of monitoring stations

When the monitoring data of the fixed monitoring station is abnormal, it can communicate with the mobile monitoring station to control the working state of the sensors of the mobile station, and increase the monitoring frequency and data return frequency. The determination of the abnormal data may be that the contrast coefficient exceeds the set range, that is, the data of the site is determined to be abnormal.

Brief description of the drawings

Figure 1 Schematic diagram for calculating the distance factor (center of geographical position);

Figure 2 Schematic diagram of calculation of distance factors (reference station and monitored points);

Figure 3 Schematic diagram of the calculation method of the pollution degree factor of the geographical location;

Figure 4 Schematic diagram of calculation methods for other site evaluation factors;

FIG. 5 is a schematic diagram of the relative positions of the mobile station E and the reference station at time T ₁ ;

FIG. 6 is a schematic diagram of the relative positions of the mobile station E and the reference station at time T ₂ ;

FIG. 7 is a schematic diagram of the relative positions of the mobile station E and the reference station at time T ₃ ;

In the figure: 101 is the base station 1, 102 is the base station 2, 103 is the base station 3, 104 is the base station 4, 100-D is the base station D, 201-B is the calibrated station B, and 201 is Small monitoring stations 1, 202 are small monitoring stations 2, 203 are small monitoring stations 3, 204 are small monitoring stations ₄ , 301-T1 are mobile monitoring stations E at T ₁ , and 301-T2 are mobile monitoring stations E at T Position ₂ at time, 301-T3 is the position of mobile monitoring station E at time T ₃ , 401-C is the pollution source small coal plant C, 101-1 is the distance from the reference station 1 to the center of the area, and 102-1 is the reference 2 The distance from the station to the regional center point, 103-1 is the distance from the reference station 3 to the regional center point, 104-1 is the distance from the fixed station 4 to the regional center point, and 101-B is the reference station 1 to the calibrated station The distance from position B, 102-B is the distance from reference station 2 to the calibrated station B, 103-B is the distance from reference station 3 to the calibrated station B, 104-B is the reference station 4 from the calibrated station The distance from station B to 201-C is the distance from reference station 1 to pollution source C, 202-C is the distance from reference station 2 to pollution source C, and 203-C is the distance from reference station 3 to pollution source C; 204- C is for base station 4 Transfection of C source distance.

detailed description

Example one

Apply the credibility weight factor calculation formula, distance factor calculation formula, and normalization method. When d represents the distance from the geometric center point of the area to each monitoring station: specify the weight of the monitoring station data when A is equal to 5km Is 1, as shown in Figure 1.

It is known that the positions of monitoring stations 1, 2, 3, and 4 to the center of the area are 8km, 6km, 7km, and 5km, respectively. Applying the distance factor calculation method, according to the normalized calculation formula, the calculation is taking into account the distance factor The way is:

2号监测站数据为PM′ _2c，

3号监测站数据为PM′ _3c，

4号监测站数据为PM′ _4c，f _d(5)＝1。 The data of No. 1 monitoring station is PM ′ _1c , the distance factor

The data of No. 2 monitoring station is PM ′ _2c ,

The data of monitoring station 3 is PM ′ _3c ,

The data of No. 4 monitoring station is PM ′ _4c , f _d (5) = 1.

PM _c = F _c × PM _C ′

Apply the credibility weight factor calculation formula, distance factor calculation formula, and normalization method. When d represents the distance from the monitored position to each monitoring point: specify that the weight of the monitoring station data when A is equal to 5km ,as shown in picture 2.

It is known that the locations of monitoring stations 1, 2, 3, and 4 to the monitored points are 9km, 8km, 7km, and 5km respectively. Applying the distance factor calculation method, according to the normalized calculation formula, the calculation takes into account the distance factor. The way is:

2号监测站数据为PM′ _2c，

3号监测站数据为PM′ _3c，

4号监测站数据为PM′ _4c，f _d(5)＝1。 The data of No. 1 monitoring station is PM ′ _1c , the distance factor

The data of No. 2 monitoring station is PM ′ _2c ,

The data of monitoring station 3 is PM ′ _3c ,

The data of No. 4 monitoring station is PM ′ _4c , f _d (5) = 1.

PM _c = F _c × PM _C ′

Example two

As shown in Figure 3, there is a small coal yard C that produces moderate pollution around the monitoring point. As shown in Figure 3, the distances from monitoring stations 1, 2, 3, and 4 are 6km, 5km, and 4km, respectively. , 3km.

Without considering other factors, according to the calculation of the geographical location factor based on the geographical location pollution factor and geographical location distance factor, the weights of the four monitoring points are 0, 0.1, 0.2, and 0.3, respectively.

Example three

为105，则平均值

与该监测站点监测值y的比值

根据其他站点评价因子计算方法： As shown in FIG. 4, it is known that there are monitoring stations 1, 2, 3, and 4 within a range of 10 km around the reference station D. The pollutant concentration values monitored by the reference station D are 100, 1, 2, 3, and 4 The concentration values monitored by the monitoring stations are 120, 120, 90, and 90, respectively, and the average value of the concentration values monitored by monitoring stations 1, 2, 3, and 4 can be calculated.

Is 105, the average

Ratio to monitoring value y of this monitoring station

Calculation method based on other site evaluation factors:

The other site evaluation factors for monitoring site D are:

f _e (1.05) = 1

Embodiment 4

During the monitoring of a monitoring site in January 2019, the number of extreme conditions such as power outages, equipment maintenance, network failures, and abnormal data was determined to be 9 in accordance with the determination rules in the table below. According to the following monitoring station stability factors, Formulas and criteria:

n is the number of times that the site has abnormal data within a certain period of time. The abnormal situation and judgment are as follows.

abnormal situation determination No data output due to power failure No output for more than 1 hour No data output due to maintenance reasons No output for more than 1 hour Data failure caused by network failure No output for more than 30 minutes, Data exception: PM2.5 data is greater than PM10 data Data abnormal time exceeds 1 hour

The stability factor of the monitoring station is:

f _s (n) = 1-9 × 0.1 ＝ 0.1

Example 5

During the monitoring activities in January 2019, mobile monitoring station E entered the coverage area of monitoring station 1, monitoring station 2 and monitoring station 10 for a total of one time. It is known that mobile monitoring equipment E entered No. 1, 2, 3 The monitoring data when the coverage area of 10km around the monitoring station is 100, 110, 120 respectively; as shown in Figure 5, Figure 6, and Figure 7.

Time T ₁ when the mobile monitoring device into E Stations Stations No. 1 is fixed around the 10km coverage 1, data stations and stations 2, respectively 105,115,110; T ₂ the time when the mobile monitoring device into the E The data of monitoring station 1, monitoring station 2 and monitoring station are 10, 105, 110 when it is 10km around fixed monitoring station 2; at time T _3, when mobile monitoring equipment E enters 10km around fixed monitoring station 3, monitoring station 1 The data of monitoring station 2 and monitoring station are 100, 115, 120 respectively.

At time T ₁ is known from stations 1 f _d is the weight factor of 0.9, 0.8 geographical factor f _l, f _e other sites of an evaluation factor, the stability factor f _s is 0.6, the weighting factor according to the credibility F _c when the calculated F _c = calculated Stations weights _{f d × f l × f e} × f s 1 of the weight _{F 1c = 0.9 × 0.8 × 1} × 0.6 = 0.432; distance factor weights f _d stations 2 is 0.8, geographical location factor f _l is 0.8, other site evaluation factors f _e are 1, stability factor f _s is 0.8, then according to the calculation method of the credibility weight factor F _c F _2c = f _d × f _l × f _e × f _s calculates the weight F _{2c of} monitoring station 2 = 0.8 × 0.8 × 1 × 0.8 = 0.512; the distance factor weight f _d of monitoring station 3 is 1, the geographical position factor f _l is 0.8, and the other site evaluation factors f _e are 0.9, the stability factor f _s is 1, then according to the calculation method of the confidence weight factor F _c F _3c = f _d × f _l × f _e × f _s , the weight F _{3c of the} monitoring station 3 is calculated 1 × 0.8 × 0.9 × 1 = 0.72; so the base station monitoring value after correction or screening at time T ₁ is calculated according to the normalization algorithm:

So by comparing the calculation formulas:

Similarly, at time T2 weighting factors from the known stations f _d 1 is 0.6, 0.8 geographical factor f _l, f _e other sites evaluation factor is 0.8, the stability factor f _s is 1, then according to the credibility Calculation method of weight factor F _c F _c = f _d × f _l × f _e × f _{s The} calculated weight of monitoring station 1 is F _1c = 0.6 × 0.8 × 0.8 × 1 = 0.384; distance factor weight f of monitoring station 2 _d is 1. The geographical position factor f _l is 0.8, the other site evaluation factors f _e are 0.9, and the stability factor f _s is 0.8. According to the calculation method of the credibility weight factor F _c F _2c = f _d × f _l × The weight F _{2c of} monitoring station 2 is calculated by f _e × f _s = 1 × 0.8 × 0.9 × 0.8 = 0.576; the distance factor weight f of monitoring station 3 is _d 0.7, the geographical location factor f _l is 0.9, and other station evaluation factors f _e is 0.8, f _s is a stability factor, the weighting calculation according to the credibility factor F _c when _{F 3c = f d × f l} × f e × f s calculated stations weight of 3 F _3c = 0.7 × 0.9 × 0.8 × 1 = 0.504; so calculate the base station monitoring value after correction or screening at time T ₂ according to the normalization algorithm:

So by comparing the calculation formulas:

Stations 1 at time T3 known distance weighting factor f _d is 0.5, 0.9 geographical factor f _l, f _e other sites evaluation factor is 0.7, the stability factor f _s is 0.8, according to the credibility of the weighting factor F calculation of _c F _c = calculated Stations weights _{f d × f l × f e} × f s 1 of the weight _{F 1c = 0.5 × 0.9 × 0.7} × 0.8 = 0.384; stations from the factor weights f _d 2 is 0.8 If the geographical position factor f _l is 0.7, the other site evaluation factors f _e are 0.8, and the stability factor f _s is 0.9, then according to the calculation method of the confidence weight factor F _c F _2c = f _d × f _l × f _e × f _s calculates the weight F _{2c of} monitoring station 2 = 0.8 × 0.7 × 0.8 × 0.9 = 0.403; the distance factor weight f _d of monitoring station 3 is 1, the geographical position factor f _l is 0.8, and the other station evaluation factors f _e are 0.9 If the stability factor f _s is 0.6, then according to the calculation method of the confidence weight factor F _c F _3c = f _d × f _l × f _e × f _s , the weight F _{3c of the} monitoring station 3 is calculated 1 × 0.8 × 0.9 × 0.6 = 0.432; so the corrected or screened base station monitoring value at time T ₃ is calculated according to the normalization algorithm:

So by comparing the calculation formulas:

The average of the calibration coefficients:

According to the calibration calculation formula, the output value of the calibrated mobile monitoring device E is:

Example Six

As shown in FIG. 3, the monitoring data of the fixed No. 1 calibrated station, No. 2 calibrated station, and No. 3 calibrated station in the known figure are β ₁ = 120, β ₂ = 115, β ₃ = 110, and the reference The data of the station are α = 110, and the distances of the fixed micro-stations No. 1, 2, and 3 from the national control reference station are 5km, 6km, and 7km, respectively. The station data is calibrated one by one to get the calibration coefficient, and the above data is counted in the following table:

经校准的B ₁为

After applying the calibration formula, the calibrated B ₂ is

Calibrated B ₁ is

In the above calculation of the comparison coefficients η _α-β3 , the denominator (reference data) is A, and the numerator (calibrated data) is B ₃ .

In the calculation of the above-mentioned calibration coefficients c _β3-β2 , the denominator (reference data) is B ₂ , the numerator (calibrated data) is B ₃ , and other calculations of the calibration coefficients are performed by analogy.

Claims (11)

An air pollution monitoring equipment data credibility evaluation method; the air pollution monitoring equipment credibility is evaluated by a credibility weight factor (F _c ); the credibility weight factor and a distance factor (f _d ) , Geographical location factor (f _l ), other site evaluation (f _e ), stability factor (f _s ) are positively correlated; The calculation method of the confidence weight factor (F _c ) is: F _c = f (f _d , f _l , f _e , f _s ) The calculation method of the distance factor (f _d ) is:

Among them, the parameter d is the distance from the geometric center point of the area to each station in the area or the distance from a specific location to other monitoring points; the parameter κ is a distance weight parameter, in general, κ = A The calculation method of the geographic location factor (f _l ) is: f _l = g (f _le , f _ld , f _lo ) Among them, f _le : geographical location pollution degree factor, f _ld : geographical location pollution distance factor, f _lo : other geographical location factor factors. The evaluation method according to claim 1, wherein the calculation formula of the other site evaluation factor (f _e ) is:

Among them, the parameter ε represents other nearby equipment, including a fixed micro-station or a mobile monitoring station, and whether there is a similar data change trend with the data obtained by the monitoring station; The calculation formula of the parameter ε is:

formula

Represents the average value of data from several monitoring sites within 10 kilometers around the monitoring site, where M represents the monitoring value of the monitoring site. The evaluation method according to claim 1, wherein the calculation formula of the stability factor (f _s ) is:

Among them, the parameter n is the number of data anomalies at the monitoring station within a period of time; the length of the period of time: the shortest is 1 day, the longest is 1 month. The evaluation method according to claim 3, wherein the abnormal data condition is one of the following: 1) No data is output due to power failure, and no output exceeds 1 hour; 2) No data output due to maintenance reasons, and no output exceeds 1 hour; 3) The data is abnormal due to network failure, and there is no output for more than half an hour; 4) The PM _2.5 data is greater than the PM ₁₀ data, and the abnormal time exceeds 1 hour. The method according to claim 1, wherein a method for screening the credibility weight factors is: 1) The credibility weighting factors are arranged from large to small, and a certain percentage of credibility weighting factors before the ranking is selected, and the data corresponding to the selection of a certain percentage of credibility weighting factors before the ranking is valid data ; Exclude a certain percentage of the credibility weight factor after the ranking, the data corresponding to the certain percentage of the credibility weight factor after the exclusion ranking is valid data; 2) selecting a credibility weight factor that exceeds a certain threshold, the data corresponding to the credibility weight factor that exceeds a certain threshold is valid data; excluding credibility weight factors that are below a certain threshold, the exclusion is lower than The data corresponding to a certain threshold credibility weighting factor is valid data. The method according to claim 5, wherein the certain proportion is one of 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, and 90%; the certain The threshold is one of 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9. A method for calibrating an air pollution monitoring device. The mobile monitoring device performs multiple calibrations based on the area data of a single reference station. The steps are: 1) Evaluate the credibility weight factor of the reference station. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid and is not used as a basis for calibration; 2) Select a reference station whose credibility weight factor meets the set value as a reference; if a mobile monitoring device passes the same reference station multiple times, each time this device passes the reference station, a comparison is performed to obtain the Η values of two comparisons; 3) After recording a set number of η values, the set number is between 10 and 100; all the η values obtained from the record are statistically calculated, and the calculation method is average, normal distribution value, approximation, or PID Mathematical method; finally the average value after statistical calculation

4) Use the final average

And a calibration calculation formula to calibrate the mobile monitoring device. A method for calibrating air pollution monitoring equipment. The mobile monitoring equipment performs multiple calibrations based on regional data of multiple reference stations. The steps are: 1) Evaluate the credibility weight factor of the reference station. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid and is not used as a basis for calibration; 2) In the actual monitoring and calibration process, each time a mobile monitoring device passes the area covered by a reference station, it is compared with the average reference data of the passing time in the area to obtain the corresponding η value. The average reference data of the area is Derived from the normalized calculation method; 3) After recording a set number of η values, perform statistical calculations on all the η values recorded. The calculation method is average, normal distribution value, approximation, or PID mathematical method; finally, the average value after statistical calculation is obtained

4) Use the final average

And calibration calculation formula to calibrate the mobile monitoring equipment; The normalized calculation method is:

Among them, y _c is the revised reference data; y ′ is the uncorrected reference station data; n is the number of reference stations that have reached the standard. A method for calibrating other fixed stations based on the regional data of a single base station; the steps are: 1) Judging the credibility weight factor. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid and is not used as a basis for calibration; 2) The reference station that meets the standard is used as a reference, and at intervals, the base station data is compared with the fixed station to be calibrated to obtain the corresponding η value; 3) After recording the number of η values, perform statistical calculations on all the η values recorded, such as average, normal distribution value, approximation, PID and other mathematical methods, and finally obtain the average value after statistical calculation

4) Use the final average

And calibration calculation formula to calibrate the mobile monitoring device. A method for calibrating other fixed stations based on the regional data of multiple reference stations; the steps are: 1) Judging the credibility weight factor. If the credibility weight factor of the reference station is less than the set value, the monitoring data of the reference station is invalid and is not used as a basis for calibration; 2) In the actual monitoring and calibration process, each time a mobile monitoring device passes the area covered by a reference station, it is compared with the average reference data of the passing time in the area to obtain the corresponding η value; the reference value of the area Calculated using the normalized calculation method; 3) After recording a certain number of η values, perform statistical calculations on all the η values recorded, using calculation methods including average, normal distribution value, approximation, or PID; finally, obtain a statistically calculated average value

Use the resulting average

Calibrate the mobile monitoring equipment again; 4) Use the final average

And calibration calculation formula to calibrate the mobile monitoring equipment; The normalized calculation method is:

Among them, yc is the revised reference data; y ′ is the uncorrected reference station data; n is the number of reference stations that have reached the standard. The method according to any one of claims 7 to 10, wherein the calculation formula of the η parameter is:

or

Among them, x is the calibrated data, y is the reference data, and η is the contrast coefficient.

Images