EE05822B1 - Moisture meter and method for measuring the moisture content of wood above the fibre saturation point of wood with the electric charging effect. - Google Patents

Abstract

The invention belongs to the field of research and analysis of materials, in particular to electrical methods for determining the moisture content of wood, and is applicable to moisture contents higher than the saturation point of the wood fiber. The humidity meter for measuring the moisture content of wood with an electric charge effect above the saturation point of the wood fiber includes two measuring electrodes, a stable reference voltage source, an electromagnetic input preamplifier, an analog-to-digital converter, to control the measurement process and an LCD display with mathematical algorithms embedded in the microcomputer to fully automate the measurement process. A method for measuring the moisture content of a wood above the saturation point of the wood by measuring the electrical resistance of the wood, finding the real-time electrical capacity of the wood at different time points and determining the wood load number to create and validate a calibration model.

Description

TECHNICAL FIELD

The present invention belongs to the field of research and analysis of materials properties, methods and devices for determining wood moisture, more specifically electrical conductivity and impedance methods and devices.

STATE OF THE ART

The idea of determining the moisture content of wood based on the measurement of the electrical conductivity of wood was first proposed by Stamm in 1927. Since then, wood moisture meters operating on the principle of the electrical resistance or electrical conductivity of wood (which is the inverse of electrical resistance) have been manufactured industrially in series and are widely used in practice today. The problematic measuring range for the well-known resistance-type wood moisture meters used in practice starts at 30% wood moisture content (relative to dry weight) and increases from there to about 150% moisture content as the moisture content increases. This moisture content range is called the region above the saturation point of the wood fiber, because then free water begins to accumulate in the capillaries of the wood.

The electrical resistance of wood varies within very wide limits, depending on the moisture content of the wood, the temperature, and to some extent also the type of wood. Laboratory measurements of the electrical resistance of wood confirmed a rough estimate that, for example, at room temperature of 20 degrees Celsius, the electrical resistance of wood with an 8% moisture content is about 10 giga-ohms, with an 18% moisture content about 10 mega-ohms, with a 47% moisture content about 100 kilo-ohms, with a 150% moisture content about 10 kilo-ohms. Consequently, it is necessary to calibrate the moisture meter to obtain a reliable measurement result.

Based on the literature (Tamme, V., Muiste, P., Padari, A., Tamme, H. 2014. Modelling of Resistance-Type Wood Moisture Meters for Three Deciduos Tree Species (Black Alder, Birch, Aspen) in Moisture Contents Above Fibre Saturation Point. Baltic Forestry, 20 (1): 157-166), an extensive study was conducted to clarify the reliability of three wood moisture meters in practical use and production.

at variable moisture contents greater than the saturation point, i.e. over 30% of wood. The study found that at a 95% confidence level, it was possible to provide new calibration models for a known moisture meter and three wood species, based on which it was possible to eliminate the initially found large systematic errors (up to 75% moisture content (NS)) in the measurement results.

The electrical resistance of wood has been measured in known inventions using either the direct current or alternating current method. Both variants are in use.

According to patent US4408128 (A), a known method for determining the moisture content of wood is essentially determining the alternating current electrical resistance of wood. In the case of the known technical solution, the measurement range of the moisture content of wood is given in the description of the invention from 10% to 40%. For this moisture content range, the patent description also gives a calibration curve that relates the actual moisture content of wood to the alternating current electrical resistance.

According to patent ES2566775 B2 (A1), a known method for determining the moisture content of wood is essentially determining the alternating current electrical resistance of wood. In the known technical solution, the measuring range of the moisture content of wood is given as 7% to 20% moisture content. For the said measuring range, the patent description also provides a calibration curve that relates the measured alternating current electrical resistance of wood to the actual moisture content of the corresponding wood. The calibration curves are given for two types of wood.

According to patent US4259633 (A), a known method for determining the moisture content of wood is essentially the determination of the direct current electrical resistance of wood. The temporal change in the electrical resistance, or drift, is compensated for by analog correction of the supply voltage to obtain a stable reading. There is no publicly available data on the calibration model in the description of the known technical solution. The general measurement range of the known technical solution is declared in the patent description to be 10% to 90% moisture content.

According to patent US67008555 (A), a calibration method is known, the characteristic feature of which is that the measured quantity, the electrical capacitance C of wood, is associated in the calibration method with the moisture content of the wood.

The closest technical solution is the determination of the moisture content of wood by the DC electrical resistance method known according to patent US4621229 (A). The reading of the known moisture meter is essentially the moisture content of wood calculated or predicted by the internal data processing algorithm or calibration function of the moisture meter. A characteristic feature of the known technical solution is that the measurable quantity, the electrical resistance of wood R, is associated with the moisture content of the wood in the calibration method. A disadvantage of the known method is that the measurable quantity, the electrical resistance of wood, is associated with the moisture content of the wood in the calibration method, which does not allow for reliable measurement results at variable moisture contents greater than the saturation point of the wood fiber, i.e. over 30% of the wood.

The main disadvantage of the known method is that at wood moisture contents above the fiber saturation point, the dispersion of electrical resistance measurement data increases progressively with increasing wood moisture content, causing serious problems in calibrating resistance-type wood moisture meters, which in turn is the reason for insufficient accuracy in determining wood moisture content at wood moisture contents above 30%.

The closest technical solution is, according to patent US4623229 (A), a known moisture meter for measuring moisture content above the saturation point of wood by the effect of electrical charging of wood, which includes measuring electrodes, a reference voltage source, a preamplifier with an electrometric input and an analog-to-digital converter with a moisture meter, the construction of which is illustrated in figure 1 of patent US4621229 (A).

The disadvantages of the known moisture meter include insufficient accuracy in determining the moisture content of wood, since the measured quantity, the electrical resistance of wood, is associated with the moisture content of the wood.

The literature (V. Tamme. Development of electrical resistance-type control methods in wood dryers. Dissertation for the degree of Doctor of Philosophy. Tartu, 2016, 135 pages) has analyzed the possibility of so-called individual calibration of measuring electrodes when determining the moisture content of wood fiber above 30%. It has been found that by means of individual calibration of measuring electrodes, it is practically possible to exclude the occurrence of random measurement errors when determining the moisture content of wood fiber above 30%.

The disadvantage of individual calibration is that an individual calibration performed for the electrodes at a specific location loses its validity if the measuring electrodes (needle or screw electrodes) are installed at the next randomly selected location in the wood. Thus, individual calibration is actually a one-time procedure that is only justified in laboratory conditions.

One important conclusion can be drawn from the experience of individual calibration of measuring electrodes: the real cause of large random measurement errors when determining the moisture content of wood fiber above 30% is most likely the resistance of the electrical contact between the wood and the measuring electrode, the value of which is individual and unique at each point where the measuring electrodes are inserted into the wood due to the heterogeneous and variable natural properties of the wood (density, moisture content, so-called water pockets, resin content, etc.).

In known technical solutions, wood moisture meters are actually wood electrical resistance meters that are calibrated using the dry weight method (e.g. standard ISO 3130:1975) to measure wood moisture content. The corresponding calibration models are usually constructed as linear regression models.

Based on the literature (see V. Tamme. Development of electrical resistance-type control methods in wood drying. Dissertation for the degree of Doctor of Philosophy. Tartu 2016, 135 pages), it is known that at wood fiber moisture contents above 30%, the electrical resistance of the measuring electrode-wood contact is also accompanied by the formation of electrical capacitance of the electrode/wood contact, whereby both - electrical resistance and electrical capacitance - are time-dependent quantities.

Practice has shown that corrosion of the original metal measuring electrodes (type M19) from the well-known company Gann (www.gann.de), which owns the invention, can significantly disrupt the recording of the time dependences of electrical resistance and electrical capacitance.

The aim of the invention was to develop a new generation of wood moisture meter and a method for calibrating it, for reliable determination of wood moisture content above the saturation point of wood fibers, more precisely above 30%.

ESSENCE OF THE INVENTION

The essence of the present invention is to create a moisture meter different from the previously known technical solution for determining the moisture content of wood at wood moisture contents greater than the wedge loading point of the wood fiber, more precisely, at wood moisture contents greater than 30%, and a method for calibrating the moisture meter.

The moisture meter for measuring the moisture content of wood above the saturation point of the wood fiber by the effect of electrical charging includes two measuring electrodes, a stable reference voltage source, a preamplifier with an electrometric input, an analog-to-digital converter, a precision resistor in the feedback circuit of the preamplifier with an electrometric input, a block of electronic active filters, which includes filters for filtering low-frequency and high-frequency measurement noise using an operational amplifier and a capacitor, a microcomputer for controlling the measurement process, and LCD displays, whereby mathematical algorithms have been entered into the microcomputer for the purpose of fully automating the measurement process.

The measuring electrodes of the moisture meter are corrosion-free insulated carbon fiber needle electrodes. They can also be made of stainless steel or aluminum alloy.

According to the method for calibrating a moisture meter with an electric charging effect on wood, the measurement process includes measuring the electrical resistance and charging current of the wood as a function of time, with a pre-selected measurement interval, at a constant charging voltage for a measurement cycle of a prescribed length, whereby the electrical capacitance of the wood at different points in time is found in real time based on the time course of the measured charging current and the charging number of the wood is determined for the preparation and validation of an optimal calibration model, whereby the measurement process is controlled digitally.

According to the method of electrical charging of wood, the electrical resistance of wood, according to the charging number of wood, is inversely proportional to the electrical capacitance of wood at all different actual moisture contents of wood in the calibration model, whereby the deviations of electrical resistances are combined in the calibration model by integrating a variable dependent on electrical capacitance into the calibration model. The variable dependent on electrical capacitance is the electrical resistance of wood through the charging number and the moisture content.

The variables characterizing the electrical charging of wood, electrical resistance and electrical capacitance, are measured with the best possible accuracy at a constant charging voltage in potentiostatic charging mode. During one measurement cycle, a number of test data are collected, the obtained test data are processed, mathematical relationships are used for this, and as a result, the quality of the calibration model is improved. The collected data is processed using a microcomputer and the time courses of electrical resistance and electrical capacitance are recorded.

LIST OF DRAWINGS

Fig. 1 shows the analog part of an electronic device that implements the principle of electrical charging of wood.

Fig. 2 shows the basic diagram of a moisture meter using the effect of electrical charging of wood.

Fig. 3 shows typical dependences of the electrical resistance of wood Rx = Rwood + Rcontact

of time during a 20 second measurement cycle.

Figure 4 shows the dependences of the electrical capacitance C of wood on time during a 20-second measurement cycle.

Figure 5 shows a graph of the logarithm of the wood loading number, which corresponds to formula (6).

Figure 6 shows a LogR1n-dependent calibration model for predicting wood moisture content based on wood electrical resistivity measurements.

Figure 7 shows a comparison model for comparing actual wood moisture content and wood moisture content predicted based on R1n.

Figure 8 shows a comparison model for comparing the actual moisture content of wood and the prediction of wood moisture content based on R1n averaged over 10 replicate measurements.

Figure 9 shows a calibration model for predicting wood moisture content based on the combined variable (R1n+ C1n).

Figure 10 shows a comparison model for comparing the actual moisture content of wood and the moisture content of wood predicted based on the variable (R1n + C1n)c, if the computational systematic error resulting from the compression of random errors at the actual moisture content of wood of 105% NS has been corrected.

Fig. 11 shows a calibration model dependent on the combined variable Log(R1+R20) for predicting the actual moisture content of wood above 30%.

Figure 12 shows a comparison model for comparing the actual moisture content of wood and the moisture content predicted based on the variable Log(R1n + R20n).

Figure 13 shows a comparison model for comparing the actual moisture content of wood and the moisture content predicted based on the variable Log(R1n+ R20n) when 10 replicate measurements have been made.

Figure 14 shows the calibration model with a combined independent x-variable R + 0.00005 ·C3, which is referred to hereafter as the R+Par5 variable.

Figure 15 shows a comparison model for comparing the actual moisture content and the moisture contents predicted using the calibration function shown in Figure 14.

Figure 16 shows the R+Par5 calibration model when k=10 replicate measurements have been made.

Figure 17 shows a comparison model for comparing the actual moisture content of wood and the moisture content of wood predicted by the calibration model shown in Figure 19, if k=10 replicate measurements have been made.

Figure 18 shows the tolerance intervals for different models.

EXAMPLE

The invention will now be described more fully with reference to the accompanying drawings, in which a preferred embodiment of the drawings is shown.

Fig. 1 shows the analog part of an electronic device that implements the principle of electric charging of wood. Electric charging of wood occurs at a constant voltage, for example, U0= 1.024 volts, which is taken from a reference voltage source. A precision resistor R0 is connected to the feedback circuit of the electrometer E, or more precisely, an operational amplifier with an electrometric input. The measured electrical resistance of the wood Rx is connected to the input of the electrometer. The charging current Ilaad flows through the measured electrical resistance of the wood. The output voltage of the electrometer UVäij changes within the prescribed limits.

Equivalent diagram of the electrical charging of wood (Fig. 2), where the resistance of the wood/measuring electrode electrical contact depends logarithmically on time:

R(t) = a ln(t) + b, (1)

and the electrical capacitance of the wood/electrode contact depends linearly on time:

C(t) = а1 · t + b1, (2)

where a, b, a1 and b1 are constants for a given specific location of the measuring electrodes in the wood. If the location of the measuring electrodes in the wood is changed, the constants acquire new values, but the general shape of the functions (logarithmic (1) - linear (2)) remains the same.

The principle diagram shown in Fig. 1 has been used to calculate the time dependences of electrical resistance and electrical capacitance, where the voltage that electrically charges (polarizes) the wood is taken from a precision reference voltage source U0=1.024 V and the time-dependent electrical resistance of the wood is calculated from the formula

where Rx is the electrical resistance of the wood in ohms, R is the resistance of the electrometer's feedback precision resistor in ohms (100 megohms), and Uoutput is the measured analog output voltage in volts.

The wood charging current in amperes is calculated from the formula:

and the electrical capacitance of wood is calculated from the formula:

where Cx is the electrical capacitance of the wood in farads, Q is the electrical charge of the wood in coulombs, the specific value of which is determined by numerical integration of the charging current Ilaad, which in turn is carried out in real time by a microcomputer controlling the operation of the device.

The cycle of electrical charging of wood at a constant charging voltage lasts, for example, 20 seconds, with the measurement interval being pre-selected between 0.1 and 1.0 seconds. Thus, in the given example, during one measurement cycle, 200 to 20 time-dependent measurements of the electrical resistance Rx of the wood and the charging current Ilaad are recorded.

To highlight the essence of the invention, the empirical charge number of wood has been defined based on the analysis of experimental data. If we form the products of the measured electrical resistances and the calculated electrical capacitances during each subsequent measurement interval, it turned out that these products are practically constant (with very small variability, see Fig. 3), and also do not depend on the moisture content of the wood in the region of wood moisture contents above 30%. This empirical constant with very small variability was called the charge number of wood. By taking the logarithms of the charge number of wood, we obtain a simple mathematical relationship, which is convenient to present as a graph (Fig. 3):

log(LA) = log R1n + log C1n, (6)

where LA is the charge number of the wood, R1n are the electrical resistances of the wood and C1n are the electrical capacitances of the wood in the first second of the nth measurement cycle, respectively.

The comments on the empirical characteristic of wood, the load capacity, are as follows.

1. Since the wood loading number itself does not depend on the moisture content of the wood, it is not suitable as a variable in a calibration model, the purpose of which is to predict the moisture content of wood above the wood fiber saturation point, that is, at a moisture content of wood above 30%.

2. The wood charge number mathematically represents the empirical characteristic that the electrical resistance of wood is inversely proportional to the electrical capacitance of wood at all actual moisture contents of wood present in the calibration model. This fact can theoretically be used to compress large deviations in electrical resistance by integrating a new variable dependent on electrical capacitance into the calibration model. This procedure can be called improving the signal-to-noise ratio of the calibration model, or also increasing the quality of the calibration model.

Fig. 2 shows a schematic diagram of a moisture meter with the effect of electric charging of wood, with the functional elements of the moisture meter shown as a block diagram. The moisture meter includes measuring electrodes 1 and 2, a stabilized reference voltage source 3, an analog measuring circuit block 4, an analog-digital converter 5, a microcomputer 6 and LCD displays 7. The analog measuring circuit block 4 includes a preamplifier A1 with an electrometric input, a block of electronic active filters 8 and a circuit 9 (R2 and R3) for summing the output voltage of the analog amplifier and the zeroing voltage. The measuring electrodes 1 and 2 are intended to be pressed into wood 11 to measure the moisture content (NS).

In Fig. 2, the functional elements of the device are shown schematically as a block diagram. An analog measuring circuit block 4, which includes a preamplifier with an electrometric input A1 as a signal preamplifier, a feedback precision resistor 10 (R0), a potentiometric circuit R1 for setting the amplifier to zero, a circuit 9 (for example, resistors R2 and R3) for summing the amplifier output voltage and the zeroing voltage, a block of electronic active filters 8 (in Fig. 2, marked as the block "Filter"), which includes filters for filtering low-frequency and high-frequency measurement noise using an operational amplifier A2 and a capacitor C1. A stabilized reference voltage source 3 outputs a stabilized DC voltage with a value of U0 to the measuring electrode 2, which voltage on the measuring electrode 2 drops during the measurement process at the equivalent resistance Rx of wood 11 or wood material to a potential of zero volts on the measuring electrode 1, which is

connected to the electrometric input of the preamplifier A1 with the point О indicating the zero potential of the input in the analog measuring circuit block 4. The general purpose of the analog measuring circuit block 4 is to convert the value of the electrical resistance Rx measured at the input of the preamplifier A1 as linearly and noise-free as possible into the analog output voltage Uout, thereby creating the prerequisites for converting the analog output signal into a digital signal using the analog-digital converter 5. Further digital processing of the measurement result is carried out by a microcomputer 8, the processing result is numerically output on the LCD display 7. The power supply circuits of the electronic elements 3, 5, 6, 7, 8, 9 of the moisture meter are not shown in Figure 2, nor is the sound source (buzzer) or light source (light emitting diode) marking the measurement cycle shown.

Fig. 3 shows typical time dependences of the electrical resistance of wood 11 Rx=Rwood +Rcontact during a 20-second measurement cycle. The measurements were made in 21 randomly selected locations, in the case of one tree species, and at a depth of 1/3 of the wood thickness. The tree species was alder, with an actual average moisture content of 105%, which was determined by the dry weight method according to the ISO3130:1975 standard and at a temperature of 50 degrees Celsius, the electrical resistance was measured in the wood at a depth of 1/3 of the material thickness from the surface. One graph corresponds to each measurement cycle. A trend line is added to the graph showing the maximum electrical resistance measurement results of wood and to the graph showing the minimum measurement results in rafters. It can be visually identified from the graphs that the smallest dispersion is in the electrical resistance of wood at the beginning of the measurement at point R1n. Comparative descriptive statistics at measurement points R1n and at measurement points after 20 seconds R20n confirm this conclusion. The calculated corresponding coefficients of variation (σ/average)· 100% are respectively: (R1n=9.7% and R20n=15.9%), i.e. the coefficient of variation increases significantly at the 20th second from the start of the measurement.

Figure 4 shows typical time dependences of the electrical capacitance C of wood during a 20-second measurement cycle. The measurements were made in 21 randomly selected locations in wood (for the sake of clarity of the figure, only the min, max and 4 intermediate graphs of the dependence of capacitance on time from the 21 actually determined graphs are shown), for one wood species, and at a depth of 1/3 of the wood thickness. The wood species given in the example is alder, with an actual average moisture content of 105%, which was determined by the dry weight method according to the ISO 3130:1975 standard and at a temperature of 50 degrees Celsius, the electrical capacitance, as well as the electrical resistance, were measured in the wood at a depth of 1/3 of the material thickness from the surface. One graph corresponds to each measurement cycle. Maximum measurement results of the electrical capacitance of wood

A trend line has been added to the graph showing the minimum measurement results and the graph showing the minimum measurement results in the raft. It can be visually identified from the graphs that the smallest dispersion is in the electrical capacitance of wood at the beginning of the measurement at point C1n. Comparative descriptive statistics at measurement points C1n and C20n confirm this conclusion. The calculated corresponding coefficients of variation (i.e. standard deviation divided by the arithmetic mean in percentages: (σ/average)· 100% are (C1n=10.4% and C20n=14%), so 20 seconds after the start of the measurement, the coefficient of variation quantitatively characterizing the dispersion increased significantly.

Fig. 5 shows a graph of the logarithm of the wood loading number, which corresponds to formula (6). The loading number itself does not depend on the moisture content of the wood, but its importance lies in the fact that using statistical variables based on the loading number, it is theoretically possible to compress the variance of the regression model for calibrating a wood moisture meter, especially at high moisture contents of the wood, where traditional models using the electrical resistance of the wood become progressively less accurate as the moisture content increases (i.e., they provide a less accurate prediction of the wood moisture content).

The following figures, fig. 6, fig. 7 and fig. 8, include an embodiment where the electrical resistance of the wood R1n is selected as the independent variable in the calibration model as the variable with the smallest initial dispersion (see fig. 3).

Figure 6 shows a LogR1n-dependent calibration model for predicting wood moisture content based on wood electrical resistivity measurements. The total number of measurements per model was n=63.

Figure 7 shows a comparison model for comparing the actual moisture content of wood and the moisture content predicted based on R1n. Such a comparison model allows for an assessment of the predictive potential of the calibration model given in Figure 6, as the actual moisture content of wood is determined with high accuracy by the dry weighing method according to ISO 3130:1975.

Figure 8 shows a comparison model for comparing the actual moisture content of wood and the prediction of the moisture content of wood based on the average R1n of 10 replicate measurements. In practice, 10 replicate measurements are the most accepted method for prediction.

to increase reliability, as it is not very labor-intensive. This technique is also called validating the calibration model with 10 replicate measurements.

Compared to the single measurement predicted tolerance interval of 35% NS for the actual moisture content of 90% of black alder wood, known in the literature (Tamme, V., Muiste, P., Padari, A., Tamme, H. 2014. Modelling of Resistance-Type Wood Moisture Meters for Three Deciduos Tree Species (Black Alder, Birch, Aspen) in Moisture Contents Above Fibre Saturation Point Baltic Forestry, 20 (1): 157-166), the single measurement tolerance interval for the actual moisture content (NS) of 105% predicted by the calibration model of Fig. 6 is 20% NS. Thus, the single measurement prediction improved by 15% NS. However, if we compare the single measurement prediction of the model Fig. 7 and the tolerance interval prediction made by the repeated measurement of the model Fig. 8, 10 at the same actual moisture content of 105%, then the improvement was 20% NS -8.5% NS = 11.5% NS. The improvement in the prediction of the tolerance interval for wood moisture content is probably related to several factors. First, the recalibration of previously calibrated wood moisture meters carried out in practice, while in Fig. 6 the original electrical resistance meter was directly calibrated as a wood moisture meter. Second, the known wood moisture meter had an extremely wide measurement range in the electrical resistance scale (10 gigaohms to a few tens of kilo-ohms), which probably reduced the electrical resistance measurement accuracy by more than 30% at wood moisture contents, ultimately increasing the dispersion of the measurement results.

In Fig. 9, Fig. 10 and Fig. 11, a combined variable (R1n + C1n) based on the ideology of wood loading number is used as an independent variable in the calibration model to suppress the large dispersion of the calibration model, especially at higher wood moisture contents.

Figure 9 shows a calibration model for predicting wood moisture content based on the combined variable (R1n + C1n). The total number of measurements per model is n=63.

Figure 10 shows a comparison model for comparing the actual moisture content of wood and the moisture content of wood predicted based on the variable (R1n + C1n)c, if the calculated systematic error of 11.17% NS resulting from the compression of random errors at the actual moisture content of wood of 105% NS has been corrected.

Comparing the model shown in Fig. 6 (R1n) and the model shown in Fig. 9 (R1n + C1n), it can be concluded that, for example, at an actual moisture content of wood of 105%, the

the tolerance interval of a single measurement is 20% NS to 7% NS, respectively. This means that adding electrical capacitance to the calibration model significantly reduces the error estimate of a single measurement, i.e. improves the quality of the calibration model, especially for actual moisture content of wood above 100%. If, in the case of the model shown in Fig. 9, 10 repeated measurements are used to reduce the dispersion, then when predicting the actual moisture content of wood at 105% based on a single measurement (in this case, the average result of 10 repeated measurements is understood in the sense of a single measurement), instead of the tolerance interval of 7% NS, only the interval of 2.2% NS may be reached (see Fig. 9). In Fig. 10, a very high R2=0.9885 value can be observed, which indicates a very good predictive potential of the model with corrected systematic error. If, in Fig. 10, the best model has a predicted tolerance interval of 7% NS for a single measurement at an actual wood moisture content of 105%, then at an actual wood moisture content of 77% NS it is 14.5% NS and at an actual wood moisture content of 34% NS the tolerance interval of the single measurement is 12.1% NS. Thus, a decrease in the predicted tolerance interval for a single measurement can be observed at higher, preferably over 100% wood moisture contents. The result is completely contrary to the behavior of the model that depends only on the electrical resistance of the wood (see Fig. 6).

Comparing the (R1n + C1n ) model shown in Fig. 9 and the (R1n + C1n+1/(R1n + C1n)) model shown in Fig. 12, it can be seen that the use of the more complicated model does not give a special effect compared to the simpler model (Fig. 9). For example, in the case of 105% alder wood, the predicted single measurement tolerance interval improved from 7.17% NS to 7.15% NS, thus only 0.02%.

The (R1n+R20n) combined variable calibration model is also indirectly related to the ideology of the wood loading number, because the increase in the electrical resistance (R20n) of wood by the 20th second of loading is in a certain synchrony with the increase in the electrical capacitance of wood by the 20th second of loading (see Figures 3 and 4). The following Figures 14, 15 and 16 show examples of the use of (R1n+R20n) type models (for the sake of simplicity, subscripts have been omitted in the titles of the graphs).

Figure 11 shows a calibration model dependent on the combined variable Log(R1+R20) for predicting the actual moisture content of wood above 30%. The total number of measurements per model is n=63.

Figure 12 shows a comparison model for comparing the actual moisture content of wood and the moisture content predicted based on the variable Log(R1n + R20n).

Fig. 13 shows a comparison model for comparing the actual moisture content of wood and the moisture content of wood predicted based on the variable R1n + R20n) when k=10 replicate measurements have been made, i.e. the comparison model shown in the previous Fig. 15 has been validated for 10 replicate measurements to reduce dispersion.

Figure 14 shows a calibration model combined with an independent x-variable R+0.00005· C3, or hereinafter abbreviated as (R+Par5) calibration model, which no longer requires systematic error correction like the R+C model (see Figure 10), because no systematic error occurs when using the R+Par5 variable.

Figure 15 shows a comparison model for comparing the actual moisture content and the moisture contents predicted using the calibration function shown in Figure 17 and for evaluating the predictive potential of the model.

Figure 16 shows the R+Par5 calibration model when k=10 replicate measurements have been made.

Fig. 17 shows a comparison model between the actual wood moisture content and the one in Fig.

For comparing the wood moisture contents predicted using the calibration model shown in Figure 19, if k=10 replicate measurements have been made.

Figure 18 shows the tolerance intervals of different models: a) the predicted single measurement tolerance interval of the standard electrical resistance R calibration model (dotted line in the figure), b) the single measurement tolerance interval found using the combined variable R+C model (thick solid line in the figure), and c) the tolerance interval of the actual wood moisture content (dashed-dotted line in the figure) for comparison.

Fig. 18 shows graphically how the tolerance intervals of the models with different independent x-variables behave as the actual moisture content of the wood increases. The dotted-dashed line represents the tolerance interval of the actual moisture content of the wood, but because it is so narrow (less than 1% NS),

then the range is not visible in the figure, there is only one dotted-dashed line. The dotted line represents the tolerance interval predicted by the standard calibration model based on the electrical resistance R for a single measurement. This trend was first noted by Edwards in 1974. The thick solid line in the figure represents the behavior of the tolerance interval found using the combined variable R+C model. It can be seen that initially the behavior is the same as the Edwards trend, but from the actual moisture content of the wood of 78% NS the trend reverses to the Edwards trend, and the compression of the tolerance interval begins at higher actual moisture contents of the wood, which means that the ideology of the charge number begins to operate. Such compression of the tolerance interval is new compared to the prior art.

Comparing the (R1n+C1n) model shown in Fig. 9 and the (R1n+R20n) models shown in Fig. 14 at 105% actual wood moisture content, one can observe an approximately twice larger estimate of the tolerance interval of the single measurement for the (R1+R20) model, which increased from 7%NS to 13.44%NS and from 2.2%NS to 8.1%NS for 10 replicate measurements (see Fig. 16), respectively. On the other hand, comparing the tolerance intervals of the single measurement in Fig. 16 and Fig. 10 in the remaining two moisture groups with actual wood moisture content of 78%NS (T. I=21% vs. 14.5%) and 34%NS (T. I=8.6% vs. 12.1), one can observe that the tolerance interval decreases in the model in Fig. 10 in the 78%NS humidity group and slightly increases in the 34%NS humidity group.

In this technical solution, the measuring electrodes include two symmetrical, insulated needle electrodes with a predetermined distance between them, which are inserted into the wood to a certain depth from the wood surface (usually 1/3 of the wood thickness), and between which a constant DC voltage, i.e. the wood charging voltage, is applied.

The predictive potential of the calibration model can be adjusted according to the user's wishes (for example, choosing a suitable model can guarantee better predictive potential at higher moisture contents of wood).

The calibration model for a moisture meter is essentially a statistical regression model. The calibration model is constructed based on a specific set of empirical test data, or sample. The most important element of the calibration model is the calibration function, which is presented in the form of a mathematical expression, and knowing this, it is possible to calculate each measured independent

For the x-variable (electrical resistance of wood R, electrical capacitance of wood C, or some sum of these two parameters), calculate the value of the moisture content of the wood, which is the dependent y-variable from the point of view of the calibration function.

Calibration models have been found empirically, through trial and error, and are therefore not theoretically derived mathematical expressions.

The moisture meter works as follows. When describing the operation of the moisture meter, it is assumed that the device is in working order, i.e. before starting measurements, the device has been tested with precision resistors selected according to the measurement range. Testing the working order of the moisture meter is not conventionally part of the measurement process.

The measurement process can be conditionally divided into successive steps or operations or stages.

The first step is as follows. The measuring electrodes 1 and 2 are pressed into the wood material 11 (usually a board, plank or beam with an initial moisture content of more than 30%), usually to a depth of 1/3 of the material thickness, so that the electrical resistance Rx of the wood is formed in a direction perpendicular to the longitudinal grain of the wood, usually with an electrode spacing of 30 mm (unless the repeatability conditions of the calibration model provide for a different direction or a different spacing). To avoid the current path along the surface layer of the wood, the measuring electrodes must be partially insulated, so that only the conical part of the needle electrodes 1 and 2 inserted into the wood conducts the current. The measuring electrodes are then connected to a shielded coaxial cable, so that the electrode 2 is connected to the reference voltage source 3 by the non-sensitive terminal of the moisture meter and to the block 5 with point 0 by the sensitive terminal. Care must be taken to ensure that the sensitive part of the device does not receive an electrostatic charge during the installation of the wires, which can distort the measurement results at best, and damage the device at worst. The first step of the measurement process must be done manually (unless it is done by a robot). Since the measurement process of the moisture meter is pre-programmable and controlled by the microcomputer 6, the rest is done automatically.

The second step is the actual measurement. A zero check is performed on the amplifier A1, followed by the collection of test data for 20 seconds at a previously set measurement interval. The raw measurement data is further digitally filtered for noise, and the real-time calculation of the wood

electrical resistance and electrical capacitance, the test data is saved and provided with a file name. A pause of a predetermined length follows, then the collection of measurement test data starts again for 20 seconds. The measurement cycle of 20 seconds and the pause of a predetermined length are marked with both an audio and a light signal for the convenience of using the device. The pause can be used by the microcomputer for the initial statistical processing of the test data of the previous measurement according to the program. The pause can also be used to insert the measuring electrodes into the next position on the stick.

The third step is data analysis in a microcomputer 6 and the numerical output of the wood moisture content measurement result on a display 7 with statistical confidence limits. The data analysis uses a calibration method specially developed for this device, the application of which results in a scientifically justified estimate of the wood moisture content at moisture contents above the fiber saturation point.

A moisture meter with an electric charging effect for measuring moisture content above the saturation point of wood allows real-time recording and storage in the microcomputer memory of the time course of the electrical resistance and capacitance of wood during the electric charging cycle of wood, at a pre-selected measurement speed.

After the measurements are completed, the measurement data file is processed in the device's microcomputer according to the selected calibration model and a forecast of the wood moisture content is output along with an estimate of the statistical error. A characteristic feature of the method of electrical charging of wood for determining the moisture content of wood at a moisture content higher than the saturation point of the wood fiber is that when measuring moisture above the saturation point of the wood fiber, a significant decrease in measurement uncertainty occurs as the wood moisture content increases. The decrease in measurement uncertainty, especially at high moisture contents of wood of 100% and more, has been achieved by introducing the original wood charging number and taking it into account when compiling the calibration model.

The measurement accuracy of moisture meters has been increased by the introduction of calibration models. The creation of calibration models for moisture meters, or essentially recalibration of moisture meters, did not eliminate large random errors in the error estimate given to individual measurements. For example, with 90% actual moisture content (actual moisture content

was determined by the dry weight method according to ISO3130: 1975) for wood, the individual measurement predicted by the model could fall within the moisture content range of 72.5% to 105.5% moisture content with 95% confidence. In the example given, the estimated moisture content tolerance interval attributable to the individual measurement was 35% moisture content, and this at the 95% confidence level. The reliability of the resulting statistical regression models (i.e. the new calibration models for the moisture meters) was checked with non-parametric Kolmogorov-Smirnov and Shapiro-Wilk tests. All the models found were found to be statistically reliable.

Another important conclusion of the modeling is the fact that the residual standard deviation of the regression models decreased surprisingly quickly with the number of repeated measurements - this means that the models were statistically well convergent. Thus, the optimally selected number of repeated measurements (10 and more) allowed to significantly improve the quality of the calibration models and suppress random measurement errors of wood moisture content in the region of moisture content above 30%.

In this technical solution, the measuring electrodes include two symmetrical insulated needle electrodes, with a distance of 30 mm between them, which are inserted into the wood to a certain depth from the wood surface (usually 1/3 of the wood thickness), and between which a constant DC voltage, i.e. the wood charging voltage, is applied.

Using a moisture meter with the effect of electrical charging of wood, it has proven possible to get rid of the progressively widening error corridor characteristic of well-known resistance-type wood moisture meters, without taking the logarithm of the electrical resistance of the wood and without performing repeated measurements.

The nature of the invention is not limited to the above embodiment, but allows for a number of further developments in the field of modeling. For example, the above embodiment does not present calibration models based on R(0,1)1n ,C(0,1)1n and their combinations for the time instant t = 0.1 seconds. The possible selection of combined variables is also not exhaustive in this embodiment.

Claims (10)

1. A moisture meter for measuring moisture content above the saturation point of wood by the effect of electrical charging of wood, comprising two measuring electrodes (1) and (2), a stable reference voltage source (3), a preamplifier with an electrometric input (A1) and an analog-to-digital converter (5), characterized in that it comprises a precision resistor (10) in the feedback circuit of the preamplifier with an electrometric input (A1), a block of electronic active filters (8), which comprises filters for filtering low-frequency and high-frequency measurement noise using an operational amplifier A2 and a capacitor C1, a microcomputer (6) for controlling the measurement process and an LCD display (7), whereby mathematical expressions are entered into the microcomputer (6), which makes the measurement process fully automated. 2. A moisture meter according to claim 1, characterized in that the measuring electrodes (1) and (2) are corrosion-free insulated carbon fiber needle electrodes. 3. A moisture meter according to claim 1, characterized in that the measuring electrodes (1) and (2) are corrosion-free insulated electrodes made of stainless steel. 4. A moisture meter according to claim 1, characterized in that the measuring electrodes (1) and (2) are corrosion-free insulated electrodes made of aluminum alloy. 5. A method for measuring the moisture content of wood above the saturation point by the effect of electrical charging of wood, according to which the electrical resistance of wood (Rx) is measured in wood, which differs in that the electrical capacitance (C) of wood is found in real time at different time points based on the time course of the measured charging current and the charging number of wood is determined for the preparation and validation of a calibration model, whereby the measurement process is controlled digitally. 6. The method according to claim 5, characterized in that the wood charge number, according to which the electrical resistance of the wood is inversely proportional to the electrical capacitance of the wood at all different actual moisture contents of the wood in the calibration model, is used to aggregate the deviations of the electrical resistances in the calibration model by integrating a variable dependent on the electrical capacitance into the calibration model, wherein a statistical regression model is used as the calibration model, which is compiled based on a set of empirical test data. 7. The method according to claims 5 and 6, characterized in that the variable dependent on the electrical capacitance is the electrical resistance of the wood through the charge number and the moisture content. 8. The method according to claims 5 to 7, characterized in that the variables characterizing the electrical charging of wood, the electrical resistance Rx and the electrical capacitance Cx, are measured with the best possible accuracy at a constant charging voltage in a potentiostatic charging mode. 9. The method according to claims 5 to 8, characterized in that a number of test data are collected during one measurement cycle, the obtained test data are processed, for which mathematical relationships are used, and as a result the quality of the calibration model is improved. 10. The method according to claims 5 to 9, characterized in that the collected test data is processed by a microcomputer, and the time courses of the electrical resistance and electrical capacitance are recorded.