CN109813358A

CN109813358A - A sensor fault detection method based on redundant coupled signal diagnosis

Info

Publication number: CN109813358A
Application number: CN201910001620.XA
Authority: CN
Inventors: 王东; 许曈; 何倩; 曹斌; 张为公
Original assignee: Southeast University
Current assignee: Southeast University
Priority date: 2019-01-02
Filing date: 2019-01-02
Publication date: 2019-05-28
Anticipated expiration: 2039-01-02
Also published as: CN109813358B

Abstract

The invention discloses a kind of Transducer-fault Detecting Methods based on the diagnosis of redundancy coupled signal, include the following steps: 1, establish the mathematical model between fault-tolerant matrix, measurement value sensor and output vector；2, the initialization assignment of each parameter and matrix；3, judgement may problematic sensor；4, the sensor that confirmation is broken down.The inventive structure is simple, and flexibility is high, and low cost can fast implement the fault detection of sensor.

Description

A kind of Transducer-fault Detecting Method based on the diagnosis of redundancy coupled signal

Technical field

The invention belongs to observation and control technology field, in particular to a kind of sensor fault inspection based on the diagnosis of redundancy coupled signal Survey method.

Background technique

Transient fault is to lead to the major failure form of systematic failures, and transient fault is restored to be the important of guarantee system safety Means, when system jam, the intervention of redundant configuration component can undertake the work of trouble unit, reduce the event of system Downtime increases the reliability of system, therefore carries out redundant component configuration sometimes for the physical quantity of required detection.For The redundant component of configuration, such as two (kinds) or multiple (kind) sensors, the reliability of each sensor self performance also need in advance Realize effectively detection.

In each field such as communication, software, machinery, electronics, all there is signal coupling phenomenon, some force snesors, outputs Signal is related with each power/moment components, it is therefore necessary to and signal to be detected could be obtained by carrying out decoupling to each road output signal, this Increase the difficulty that the sensor that breaks down is detected in engineer application.

Therefore the fault detection method of sensor can be fast implemented with important by redundancy coupled signal by designing one kind Meaning.

Summary of the invention

Goal of the invention: in view of the foregoing drawbacks, it is simple that the present invention provides a kind of structure, and flexibility is high, and low cost can be quick Middle and small scale redundancy coupled signal is handled, realizes the sensor event based on the diagnosis of redundancy coupled signal of the fault detection of sensor Hinder detection method.

Technical solution: the present invention proposes a kind of Transducer-fault Detecting Method based on the diagnosis of redundancy coupled signal, including Following steps:

(1) mathematical model between fault-tolerant matrix, measurement value sensor and output vector is established；

(2) initialization assignment is carried out to each parameter and matrix；

(3) judgement is carried out to each sensor and selects possible problematic sensor；

(4) sensor that confirmation is broken down.

Further, the mathematical modulo in the step (1) between the fault-tolerant matrix of building, measurement value sensor and output vector Specific step is as follows for type:

Ax=b

Wherein, fault-tolerant matrix A is that the m × n being arranged according to engineering practical experience ties up input matrix；Measurement value sensor x is The dimensional vector of n × 1 of the practical composition that gets parms of multiple or multiple sensors, each or every kind of sensor read a ginseng every time Number；B is the dimension output column vector of m × 1；M is the number of probes of input, and n is the sensor signal measured value of input.

Further, each parameter and matrix that initialization assignment is carried out in the step (2) include fault-tolerant matrix A, sensing Device measured value x and fault-tolerant threshold value δ.

Further, judgement is carried out to each sensor in the step (3) and selects possible problematic sensor Specific step is as follows:

(3.1) compare the size relation between the fault-tolerant threshold value δ of the fixation of each element and setting in output vector b；

(3.2) corresponding possible problematic sensor number is found out.

Further, the fault-tolerant threshold value of the fixation of each element and setting in output vector b is compared in the step (3.1) Specific step is as follows for size relation between δ: when the row element of output column vector b is less than the fault-tolerant threshold value δ of fixation of setting When assign the output column vector b row element value be 0；If element is greater than the fault-tolerant threshold value δ of fixation of setting in column vector b, do not do Processing.

Further, corresponding possible problematic sensor number is found out in the step (3.2), and specific step is as follows: The row matrix that the non-zero element of output vector corresponds to fault-tolerant matrix A is found, the corresponding non-zero element value of the row is possible problematic Sensor number.

Further, specific step is as follows for the sensor that confirmation is broken down in the step (4):

(4.1) row matrix that possible problematic sensor corresponds to fault-tolerant matrix A is found out；

(4.2) these correspondence rows for taking out output vector, reconfigure column vector；

(4.3) judge whether possible problematic sensor breaks down, and finally determine the sensor of existing failure.

Further, the sensor of existing failure is finally determined in the step (4.3), and specific step is as follows: if step (4.2) the column vector full rank constructed in, the problematic sensor of possibility break down the simultaneously final output failure really Sensor number, otherwise the problematic sensor of possibility be not present failure do not export the sensor number then.

The present invention by adopting the above technical scheme, has the advantages that

A kind of Transducer-fault Detecting Method based on the diagnosis of redundancy coupled signal of the present invention is a kind of and utilizes computer The data collection and transmission of hardware resource itself and Development of Software Platform, structure is simple, and flexibility is high, low cost, energy Enough quickly processing middle and small scale redundancy coupled signals, realize the fault detection of sensor.

Detailed description of the invention

Fig. 1 is flow chart of the invention.

Specific embodiment

Combined with specific embodiments below, the present invention is furture elucidated, it should be understood that these embodiments are merely to illustrate the present invention Rather than limit the scope of the invention, after the present invention has been read, those skilled in the art are to various equivalences of the invention The modification of form falls within the application range as defined in the appended claims.

The embodiment of the present invention includes following steps as shown in Figure 1::

Step 1: the mathematical model between fault-tolerant matrix, measurement value sensor and output vector is established；

Mathematical relationship between fault-tolerant matrix, measurement value sensor and output vector meets:

Ax=b (1)

Wherein, fault-tolerant matrix A is that the m × n being arranged according to engineering practical experience ties up input matrix；Measurement value sensor x is The dimensional vector of n × 1 of the practical composition that gets parms of multiple or multiple sensors, each or every kind of sensor read a ginseng every time Number；B is the dimension output column vector of m × 1；M is the number of probes of input, and n is the sensor signal measured value of input.To any 1 ≤ k≤m, a_k1x₁+a_k2x₂++a_knx_n=b_kFormula (1) linear correlation need to be met, then parameter must meet in the fault-tolerant matrix A being arranged a_k1 ²+a_k2 ²++a_kn ²≠0。

Step 2: the initialization assignment of each parameter and matrix；

Fault-tolerant matrix A, measurement value sensor x and fault-tolerant threshold value δ are respectively as follows:

X=[7 4312 2]^T (3)

δ=0.2 (4)

There is coupled relation each other in No. 1 sensor, No. 3 sensors and No. 5 sensors known to formula (2) and formula (3), and 2 There is coupled relation each other in number sensor, No. 4 sensors and No. 6 sensors.

Step 3: judgement may problematic sensor；

Fault-tolerant matrix A is tieed up by the m × n set and the dimensional vector of n × 1 x can get the column vector b of m × 1, column vector b is each Row element is compared with the fault-tolerant threshold value δ of the fixation of setting respectively, if the column vector b row element is less than the fault-tolerant threshold of fixation of setting Value δ, assigning the column vector b row element value is 0；If element is greater than the fault-tolerant threshold value δ of fixation of setting in column vector b, without place Reason.

It calculates and obtains shown in column vector b such as formula (5):

B=[1 010 0]^T (5)

Due to fault-tolerant threshold value δ=0.2, there is no the nonzero values for being less than fixed fault-tolerant threshold value, therefore to the column vector b calculated It does not deal with.

The corresponding possible problematic sensor number, that is, fault-tolerant matrix A of each row nonzero element of column vector b corresponds to the row coefficient It is not 0 serial number.Be the first row and the third line known to formula (5) by the column vector b for being more than fault-tolerant threshold value δ, from formula (2) can pivot column to Measuring the corresponding possible problematic sensor number of b the first row is No. 1 and No. 5；Column vector b the third line correspondence known to formula (2) can The problematic sensor number of energy is No. 1 and No. 3.

Step 4: confirm the sensor of failure.

The corresponding possible problematic sensor number of column vector b the first row is No. 1 and No. 5, is known and No. 1 biography by formula (2) The relevant row matrix of sensor is the first row and the third line, they are expert at the column vector constituted known to formula (5)By Row matrix relevant to No. 5 sensors known to formula (2) is the first row and fifth line, knows that they are expert at by formula (5) and constitutes Column vector

The corresponding possible problematic sensor number of column vector b the third line is No. 1 and No. 3, is known and No. 1 biography by formula (2) The relevant row matrix of sensor is the first row and the third line, they are expert at the column vector constituted known to formula (5)By Row matrix relevant to No. 3 sensors known to formula (2) is the third line and fifth line, knows that they are expert at by formula (5) and constitutes Column vector

Because of the column vector full rank that row matrix relevant to No. 1 sensor is constituted, therefore there are failures for No. 1 sensor.

Claims

1. a sensor fault detection method based on redundant coupling signal diagnosis, is characterized in that, comprises the steps:

(1) Establish a mathematical model between fault-tolerant matrix, sensor measurement value and output vector;

(2) Initialize and assign each parameter and matrix;

(3) Judging each sensor and selecting the sensor that may have problems;

(4) Confirm the faulty sensor.

2. A sensor fault detection method based on redundant coupling signal diagnosis according to claim 1, characterized in that, in the step (1), a specific mathematical model of a fault-tolerant matrix, a sensor measurement value and an output vector is constructed. Proceed as follows:

Ax=b

Among them, the fault-tolerant matrix A is an m×n-dimensional input matrix set according to actual engineering experience; the sensor measurement value x is an n×1-dimensional column vector formed by the actual acquisition parameters of multiple or multiple sensors. Read a parameter; b is the m×1-dimensional output column vector; m is the number of input sensors, and n is the measured value of the input sensor signal.

3. A sensor fault detection method based on redundant coupled signal diagnosis according to claim 1 or 2, wherein the parameters and matrices that are initialized and assigned in the step (2) include a fault-tolerant matrix A, a sensor The measured value x and the fault tolerance threshold δ.

4. A sensor fault detection method based on redundant coupling signal diagnosis according to claim 3, characterized in that, in the step (3), the specific step of judging each sensor and selecting a sensor that may have problems as follows:

(3.1) Compare the magnitude relationship between each element in the output vector b and the set fixed fault tolerance threshold δ;

(3.2) Find out the sensor number corresponding to the possible problem.

5. A sensor fault detection method based on redundant coupled signal diagnosis according to claim 4, characterized in that, in the step (3.1), each element in the output vector b is compared with the set fixed fault tolerance threshold δ The specific steps of the size relationship between the two are as follows: when the element of the row of the output column vector b is less than the set fixed fault tolerance threshold δ, assign the value of the row element of the output column vector b to 0; if the element in the column vector b is greater than the set fixed fault tolerance If the threshold value is δ, no processing will be performed.

6. a kind of sensor fault detection method based on redundant coupling signal diagnosis according to claim 5, is characterized in that, in described step (3.2), find out the specific step of corresponding possibly problematic sensor number as follows: find the output The non-zero element of the vector corresponds to the matrix row of the fault-tolerant matrix A, and the non-zero element value corresponding to this row is the sensor number that may have problems.

7. a kind of sensor fault detection method based on redundant coupling signal diagnosis according to claim 3, is characterized in that, in described step (4), the concrete steps of confirming the sensor that has failed are as follows:

(4.1) Find the matrix row of the fault-tolerant matrix A corresponding to the sensor that may have a problem;

(4.2) Take out these corresponding rows of the output vector, and reconstruct the column vector;

(4.3) Judge whether the sensor that may have a problem is faulty, and finally determine the faulty sensor.

8. A sensor fault detection method based on redundant coupling signal diagnosis according to claim 7, characterized in that, in the step (4.3), the specific step of finally determining the faulty sensor is as follows: if step (4.2) The column vector constructed in is full rank, the possibly problematic sensor does fail and finally outputs the failed sensor number, otherwise the possibly problematic sensor does not have a fault and does not output the sensor number.