CN107895194A - A kind of nuclear power plant's main coolant system method for diagnosing faults - Google Patents

Abstract

The invention relates to a method for fault diagnosis of the main coolant system of a nuclear power plant. The model building module, the BPA generator, the evidence fusion module and the decision-making diagnostic device are connected in sequence, and the data obtained by the model constructed by the modules through the calculation of the BPA generator is used as evidence fusion The input of the module, and the results fused by the evidence fusion module are sent to the decision-making diagnostic device, and finally the decision-making diagnostic device outputs the judgment result. The fault symptom is used as the evidence for identifying the target, and the fusion result is obtained to judge the fault type by the method of evidence synthesis. The present invention analyzes and expresses the uncertainty of fault diagnosis from the perspective of evidence, and continuously narrows the hypothesis set by relying on the accumulation of evidence, and "Don'tknow" and "uncertain" are distinguished, so that the diagnosis result is more objective and accurate. The triangular fuzzy number is used to generate BPA, and the specific value obtained through quantitative calculation is convenient for analysis, and the calculation is relatively simple; secondly, it has stronger flexibility and is helpful for analysts to make decisions.

Description

A Fault Diagnosis Method for Main Coolant System of Nuclear Power Plant

technical field

The invention relates to a fault diagnosis technology, in particular to a fault diagnosis method for the main coolant system of a nuclear power plant.

Background technique

Nuclear energy is safe, clean and efficient energy. In order to meet the power demand, optimize the energy structure, and promote sustainable economic development, my country is vigorously developing the nuclear power industry. At present, there are 24 nuclear power units under construction, with an installed capacity of 26.549 million kilowatts, ranking first in the world in terms of scale under construction, accounting for 37.77% of the installed capacity of nuclear power units under construction in the world. Nuclear power safety has always been the focus of attention. The safety of nuclear power plants should be able to effectively and reliably guarantee the safety of surrounding residents and nuclear power workers, so the possibility of potential nuclear power accidents must be minimized. Researching and perfecting nuclear power fault diagnosis system is the need to improve the operational reliability of nuclear power plants, and has an important role and significance for nuclear power safety and economic operation. In the nuclear power system, the main coolant system is one of the important components. It is also called the reactor coolant system. Several parallel closed cooling loops. Its main function is to cool the core and transfer the heat generated by the core to the steam generator to generate steam. At the same time, as a second barrier, it can effectively prevent the leakage of radioactive materials.

The fault diagnosis system is an important system to ensure the safe and stable operation of nuclear power plants. It is a process of identifying multiple fault modes of equipment, which contains a lot of uncertainty. The traditional fault diagnosis technology has a mature theoretical basis and some practical operating experience, but in the main coolant system of nuclear power plants, there are not only various types of equipment faults but also various causes of faults. The paper "Research on Distributed Fault Diagnosis Technology of Main Coolant System of Nuclear Power Plant" proposed a distributed fault diagnosis method based on feature extraction neural network, but this method cannot accurately describe the difference between the diagnosis results when the symptom parameters take different values , so it can't reflect the impact of the change of symptom parameters on the diagnosis results, which is not conducive to the operator's decision support. This kind of problem exists widely in the actual main coolant system fault diagnosis and needs to be solved urgently.

Contents of the invention

The invention aims at the problem that in the main coolant system of a nuclear power plant, there are not only various types of equipment failures but also different causes of failures, which makes the fault diagnosis difficult to be accurate, and proposes a fault diagnosis method for the main coolant system of a nuclear power plant, which can be more accurate It can accurately describe the change of fault diagnosis confidence with symptom parameter value, effectively reflect the relationship between fault and symptom, and make accurate judgment and decision.

The technical solution of the present invention is: a fault diagnosis method for the main coolant system of a nuclear power plant, which specifically includes the following steps:

1) The model building module establishes the corresponding relationship between symptoms and faults: set the fault type and its corresponding symptoms, establish a fault space model, and obtain the corresponding relationship between symptoms and faults;

2) The BPA is generated by the basic probability assignment function BPA generator: first, the triangular fuzzy function is established by using the minimum value, the optimum value, and the maximum value of the fault symptom parameters, and on this basis, the specific measured value of the input symptom is compared to Yes, the basic probability assignment function of each fault type is generated according to the BPA generation algorithm;

3) Use the evidence fusion module to fuse the BPA of each fault symptom generated by the BPA generator according to the symptom measurement value to obtain a fused BPA, and convert this BPA into a probability distribution for subsequent decision-making;

4) The decision-making diagnostic device judges the fault type and outputs the reliability value: the probability distribution obtained by the evidence fusion module is input into the decision-making diagnostic device, the one with the highest probability is the identified fault type, and the reliability of the conclusion is given , which is the probability value of the largest failure mode.

Described step 2) in, BPA generator is made up of following device:

(1) Historical database input device, input the maximum, minimum and best normal values of symptom parameters;

(2) symptom parameter input device, input the measured value of each symptom parameter, i.e. real-time data;

(3) Membership function generation device generates a suitable membership function according to the model constructed by the module, and calculates the distribution of the basic probability function;

The BPA generator generates BPA as follows:

Establish triangular fuzzy numbers: the vertical axis μ _A represents the symptom measurement value, and the three points X _ij1 , X _ij0 , and X _ij2 on the abscissa respectively represent the minimum, best and maximum normal values corresponding to a certain fault symptom parameter, where the symptom parameter The best normal value X _ij0 corresponding to the vertical axis symptom measurement value is 1, which is the apex of the triangle, and this fault symptom triangular fuzzy number function is established, and all fault symptom triangular fuzzy number functions are established in this way;

For example, the fault type identification framework Θ={F ₁ , F ₂ , F ₃ , F ₄ , N}, where F ₁ , F ₂ , F ₃ , and F ₄ are four types of faults, N is normal operation, and a certain symptom parameter is set The measured value is X _i , and this parameter is a typical symptom of fault F _i . The degree _to which Xi belongs to target F _i is expressed as μ _i , the degree to which it belongs to the other three faults is μ _i ', and the degree to which it belongs to normal condition N The degree is μ _N , where the value of μ _N is the membership degree value y _N obtained after substituting the value of Xi _i into the triangular fuzzy numbers X _ij1 , X _ij0 , and X _ij2 , and each membership degree value is calculated by the following three formulas:

Then normalize the above three membership degrees, and establish the BPA of this measured value as follows:

Among them, F _a , F _b and F _c are three kinds of faults except F _i respectively.

The beneficial effect of the present invention is that: the fault diagnosis method of the main coolant system of the nuclear power plant of the present invention uses the fault symptom as the evidence for identifying the target, and uses the method of evidence synthesis to obtain the fusion result to judge the fault type. The present invention analyzes the fault diagnosis from the perspective of evidence Uncertainty is analyzed and expressed, relying on the accumulation of evidence to continuously narrow the hypothesis set, and to distinguish "don't know" from "uncertain", so that the diagnosis result is more objective and accurate. The triangular fuzzy number is used to generate BPA, and the specific value obtained through quantitative calculation is convenient for analysis, and the calculation is relatively simple; secondly, it has stronger flexibility and is helpful for analysts to make decisions.

Description of drawings

Fig. 1 is the flow chart of fault diagnosis method for main coolant system of nuclear power plant of the present invention;

Fig. 2 is the structural diagram of BPA generator;

Fig. 3 is a triangular ambiguity function diagram used in the present invention.

Detailed ways

As shown in Figure 1, a flow chart of a fault diagnosis method for the main coolant system of a nuclear power plant includes the following steps:

1. The model building module is to establish the corresponding relationship between symptoms and faults: set the fault type and its corresponding symptoms, establish a fault space model, and obtain the corresponding relationship between symptoms and faults;

2. Generate BPA by BPA (Basic Probability Assignment Function) generator: First, use the minimum value, optimum value, and maximum value of the fault symptom parameters to establish a triangular fuzzy function, and on this basis, the specific measurement value of the input symptom Carry out the comparison, and generate the basic probability assignment function of each fault type according to the BPA generation algorithm;

3. Use the evidence fusion module to fuse the BPA of each fault symptom generated by the BPA generator according to the symptom measurement value to obtain a fused BPA, and convert this BPA into a probability distribution for subsequent decision-making;

4. The decision-making diagnostic device judges the fault type and outputs the reliability value. The probability distribution obtained by the evidence fusion module is input into the decision-making diagnostic device, the one with the highest probability is the identified fault type, and the credibility of the conclusion is given, which is the probability value of the largest fault mode.

The model building module, BPA generator, evidence fusion module, and decision-making diagnostic device are connected in sequence. The data calculated by the model built by the modules is used as the input of the evidence fusion module, and the result is fused by the evidence fusion module and sent to the decision-making process. The diagnostic device finally outputs the judgment result from the decision diagnostic device.

The BPA generator is shown in Figure 2 and consists of the following devices:

1) Historical database input device for inputting the maximum, minimum and best normal values of symptom parameters;

2) The symptom parameter input device, which inputs the measured value of each symptom parameter, that is, real-time data;

3) The membership function generation device generates a suitable membership function according to the model constructed by the modules, and calculates the distribution of the basic probability function.

When a symptom parameter value is closer to its best normal value, the corresponding fault is less likely to occur, and the system is more likely to operate normally; on the contrary, the corresponding fault is more likely to occur, and the system The less likely it is to function properly. Based on this idea, BPA is generated according to the following steps:

Establish triangular fuzzy numbers (X _ij1 , X _ij0 , X _ij2 ) as shown in Figure 3, the vertical axis μ _A represents the symptom measurement value, and the three points X _ij1 , X _ij0 , and X _ij2 on the abscissa respectively represent a certain fault The minimum, best and maximum normal values of the symptom parameters, among which the best normal value X _ij0 of the symptom parameters corresponds to the vertical axis symptom measurement value of 1, which is the apex of the triangle, and the triangular fuzzy number function of the fault symptom is established, and all faults are established in this way Symptoms of triangular fuzzy number functions. It is known that the system fault type identification framework of the present invention Θ={F ₁ , F ₂ , F ₃ , F ₄ , N}, where F ₁ , F ₂ , F ₃ , and F ₄ are four fault types, and N is normal operation. Assume that the _measured value of a certain symptom _parameter is _Xi , _and this _parameter is a typical symptom of the fault F _i . Normally the degree of N is μ _N . Among them, the value of μ _N is the membership value y _N obtained after substituting the value of _Xi into the triangular fuzzy numbers (X _ij1 , _Xij0 , _Xij2 ) (see Figure 3). Each membership degree value is calculated by the following three formulas:

Then normalize the above three membership degrees, and establish the BPA of this measured value as follows:

Among them, F _a , F _b and F _c are three kinds of faults except F _i respectively.

The generated BPA is then input to the evidence fusion module for fusion through evidence combination rules, and output to the decision-making diagnostic device to convert it into a probability distribution through the gambling probability formula for decision-making judgment.

The present invention will be further described below in conjunction with accompanying drawing:

First establish the fault space model, F={F ₁ , F ₂ , F ₃ , F ₄ , N} (F ₁ —left loop steam generator heat transfer tube rupture accident, F ₂ —right loop steam generator heat transfer pipe rupture accident, F ₃ —the rupture accident of the main steam pipe of the left loop in the containment, F ₄ —the rupture accident of the main steam pipe of the left loop, and N—normal operation), each symptom parameter has a normal value when the system is in normal operation. The value range, as shown in Table 1, is the fault symptom parameter and normal range table. Then perform feature extraction for each symptom parameter, use the characteristic value 1 to represent abnormal conditions, and 0 to represent normal conditions, and convert the measured values of each parameter into characteristic parameters of {0,1}, from which several faults and corresponding typical symptoms can be found The corresponding relationship between fault monitoring parameters and feature extraction list shown in Table 2 and the corresponding relationship between symptoms and faults shown in Table 3.

Table 1

Table 2

table 3

Obviously, when the value of a symptom parameter is closer to its optimal normal value, the possibility of the corresponding failure is lower, and the possibility of the system running normally is higher; otherwise, the possibility of the corresponding failure is greater. The less likely the system is to function properly. On this basis, the BPA generation model of this paper is established as follows:

Establish triangular fuzzy numbers (X _ij1 , X _ij0 , X _ij2 ) as shown in Figure 3, where X _ij1 , X _ij0 , and X _ij2 represent the minimum, best, and maximum normal values of symptom parameters, respectively. Given the system identification framework of the present invention Θ={F ₁ , F ₂ , F ₃ , F ₄ , N}, let the measured value of a certain symptom parameter be X _i , and this parameter is a typical symptom of fault F _i , Xi _i belongs to The degree of belonging to the target F _i is expressed as μ _i , the degree of belonging to the other three faults is μ _i ', and the degree of belonging to the normal situation N is μ _N . Among them, the value of μ _N is the membership value y _N obtained after substituting the value of _Xi into the triangular fuzzy numbers (X _ij1 , _Xij0 , _Xij2 ) (see Figure 3). Each membership degree value is calculated by the following three formulas:

Then normalize the above three membership degrees, and establish the BPA of this measured value as follows:

Among them, F _a , F _b and F _c are three kinds of faults except F _i respectively.

Then use the Dempster combination method to fuse the above BPA to get the fusion result. The Dempster combination formula is as follows:

in

Among them, A _i , B _j , C _l are focal elements in the identification framework of evidence theory, m ₁ (A _i ) is the basic probability assignment assignment of focal element A _i in the first BPA, m ₂ (B _j ) is the The basic probability assignment of focal element B _j in the two BPAs, m ₃ (C _l ) is the basic probability assignment assignment of focal element C _l in the third BPA.

Finally, the fault type is identified and judged by the decision-making diagnostic device. Assuming that m is the BPA function on Θ, its corresponding gambling probability conversion formula BetP _m :Θ→[0,1] is defined as follows:

Among them, |A| is the potential of the set A (that is, the number of elements in A).

After obtaining the probability distribution, find out the one with the highest probability, which is the identified fault, and output the reliability value.

where w is the failure mode when the basic probability assignment (BPA) is converted into a probability distribution, that is, w belongs to F1, F2, F3, F4 and N. A is the focal element of BPA, which can be any subset of the power set of the identification framework, that is, A is any subset of the set {F1, F2, F3, F4, N}. m{A} is the basic probability assignment assignment of focal element A. m is the empty set The basic probability assignment assignment for .

It is found through verification that the method of the present invention can not only obtain correct diagnostic results, but also can better reflect the differences between the diagnostic results when the symptom parameters take different values, and it is easier to see the change of the confidence of the diagnostic results, which can be used for the future. The fault diagnosis research in the actual nuclear power system provides a certain reference and reference.

Claims (2)

1. a kind of nuclear power plant's main coolant system method for diagnosing faults, it is characterised in that specifically comprise the following steps：

1) corresponding relation that model construction module is established between sign and failure：Fault type and its corresponding sign are set, is built Vertical defective space model, obtains the corresponding relation between sign and failure；

2) BPA is generated by basic probability assignment function BPA makers：First, the minimum value of failure symptom parameter, optimal is utilized Value, maximum set up triangle ambiguity function, and on this basis, the specific measured value of sign of input is compared, according to BPA generating algorithms generate the basic probability assignment function of each fault type；

3) BPA for each failure symptom for being generated BPA makers according to sign measured value using evidence fusion module is carried out Fusion, the BPA after a fusion is obtained, and this BPA is converted into probability distribution, in order to subsequently carry out decision-making；

4) decision-making diagnostor failure judgement type, certainty value is exported：The probability distribution that evidence fusion module obtains is input to certainly In plan diagnostor, maximum probability person is identified fault type, and provides the confidence level of the conclusion, as maximum failure The probable value of pattern.

2. nuclear power plant's main coolant system method for diagnosing faults according to claim 1, it is characterised in that in the step 2) BPA makers are made up of following device：

(1) historical data base input unit, maximum, minimum and the optimal normal value of sign parameter are inputted；

(2) sign parameter input device, each sign measured value of parameters, i.e. real time data are inputted；

(3) membership function generating means, suitable membership function is generated according to the model of module construction, and be calculated substantially general The distribution of rate function；

BPA makers generation BPA methods are as follows：

Establish Triangular Fuzzy Number：Longitudinal axis μ_ASign measured value is represented, three point X successively wherein on abscissa_ij1, X_ij0, X_ij2Table respectively Show the optimal normal value X of minimum, the optimal and maximum normal value for corresponding to some failure symptom parameter, wherein sign parameter_ij0It is right It is 1 to answer longitudinal axis sign measured value, is triangular apex, establishes this failure symptom Triangular Fuzzy Number function, establish institute in this approach Faulty sign Triangular Fuzzy Number function；

Such as fault type framework of identification Θ={ F₁, F₂, F₃, F₄, N }, wherein F₁, F₂, F₃, F₄For 4 kinds of fault types, N is normal fortune OK, if certain sign measured value of parameters is X_i, and the parameter is failure F_iTypical sign, X_iIt is under the jurisdiction of target F_iDegree represent For μ_i, the degree for being under the jurisdiction of other three kinds of failures is μ_i', the degree for being under the jurisdiction of normal condition N is μ_N, wherein μ_NValue be X_iValue Substitute into Triangular Fuzzy Number X_ij1, X_ij0, X_ij2What is obtained afterwards is subordinate to angle value y_N, calculated respectively by following three formula and be respectively subordinate to angle value：

<mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <msub> <mi>X</mi> <mi>i</mi> </msub> <mo>-</mo> <msub> <mi>X</mi> <mrow> <mi>i</mi> <mi>j</mi> <mn>0</mn> </mrow> </msub> <mo>|</mo> </mrow> <mrow> <msub> <mi>X</mi> <mrow> <mi>i</mi> <mi>j</mi> <mn>2</mn> </mrow> </msub> <mo>-</mo> <msub> <mi>X</mi> <mrow> <mi>i</mi> <mi>j</mi> <mn>1</mn> </mrow> </msub> </mrow> </mfrac> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> <mo>=</mo> <msub> <mi>y</mi> <mi>N</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msup> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>&amp;prime;</mo> </msup> <mo>=</mo> <mn>1</mn> <mo>-</mo> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> </mrow> </mtd> </mtr> </mtable> </mfenced>

It is subordinate to angle value by 3 above again to be normalized, the BPA for establishing this measured value is as follows：

<mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>m</mi> <mi>k</mi> </msub> <mo>{</mo> <msub> <mi>F</mi> <mi>i</mi> </msub> <mo>}</mo> <mo>=</mo> <mfrac> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mrow> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>+</mo> <msup> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>&amp;prime;</mo> </msup> <mo>+</mo> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> </mrow> </mfrac> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>m</mi> <mi>k</mi> </msub> <mo>{</mo> <msub> <mi>F</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>F</mi> <mi>b</mi> </msub> <mo>,</mo> <msub> <mi>F</mi> <mi>c</mi> </msub> <mo>}</mo> <mo>=</mo> <mfrac> <mrow> <msup> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>&amp;prime;</mo> </msup> </mrow> <mrow> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>+</mo> <msup> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>&amp;prime;</mo> </msup> <mo>+</mo> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> </mrow> </mfrac> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>m</mi> <mi>k</mi> </msub> <mo>{</mo> <mi>N</mi> <mo>}</mo> <mo>=</mo> <mfrac> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> <mrow> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>+</mo> <msup> <msub> <mi>&amp;mu;</mi> <mi>i</mi> </msub> <mo>&amp;prime;</mo> </msup> <mo>+</mo> <msub> <mi>&amp;mu;</mi> <mi>N</mi> </msub> </mrow> </mfrac> </mrow> </mtd> </mtr> </mtable> </mfenced>

Wherein F_a、F_bAnd F_cRespectively remove F_iOuter other three kinds of failures.