Reciprocating compressor fault diagnosis method and system based on p-v diagram characteristic parameters

Abstract

A reciprocating compressor fault diagnosis method and system based on p-V diagram characteristic parameters, the method comprising the following steps: extracting the characteristic parameters of the p-V diagram under normal operating conditions; taking the suction and discharge pressure ratio, equivalent area and pressure centroid coordinates of the p-V diagram as input parameters, and taking the expansion process comprehensive index of the p-V diagram as output parameters, a BP neural network model is built; the characteristic parameters of the p-V diagram under normal operating conditions are used to train the BP neural network model and determine the system safety threshold; the characteristic parameters of the p-V diagram under the to-be-identified operating conditions are extracted, and the average deviation of multiple groups of actual output parameters and expected output results is calculated by substituting the trained BP neural network model; the average deviation is compared with the system safety threshold to diagnose whether a fault occurs and locate the fault type. The fault signal extraction characteristics of the present application are more obvious, and the high cost of a large number of fault data training samples required by traditional artificial intelligence technology can be avoided.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of reciprocating compressors, and particularly relates to a reciprocating compressor fault diagnosis method and system based on p-V diagram characteristic parameters. BACKGROUND

[0002] Reciprocating compressors are widely used in the industrial fields of petroleum, mining, metallurgy, chemical industry, food engineering, natural gas transportation, etc. due to their advantages of high compression pressure, stable adjustment of exhaust pressure, etc. Unplanned shutdown caused by compressor faults often brings great economic losses to related enterprises, and may even pose a danger to the life safety of on-site operators. Therefore, it is of great significance to maintain the reliable and safe operation of compressor units by applying fault diagnosis technology to ensure economic benefits and avoid safety accidents.

[0003] In recent years, with the vigorous development of automation technology and computer technology, real-time condition monitoring technology has been widely applied to reciprocating compressors, and fault diagnosis methods combining artificial intelligence technology with traditional monitoring parameters such as vibration parameters, acoustic emission signals and thermal parameters have been developed. Among them, the fault diagnosis method based on mechanical vibration signals and acoustic emission signals is prone to be disturbed by noise signals, etc. due to the complex information contained in mechanical vibration and acoustic signals, and the signal preprocessing and feature extraction process is relatively complex. Moreover, the current fault diagnosis method requires a large amount of compressor fault operation data as training samples for artificial intelligence algorithms. However, due to the requirements of compressor operation economy and reliability, it is difficult to obtain sufficient sample data. SUMMARY

[0004] The application aims to solve the problems in the prior art, and provides a reciprocating compressor fault diagnosis method and system based on p-V diagram characteristic parameters, which combines BP neural network, performs fault diagnosis on compressor vulnerable parts according to the changes of p-V diagram characteristic parameters, and reduces the engineering difficulty of applying artificial intelligence technology to compressor fault diagnosis. Moreover, using p-V diagram data under normal operating conditions as training samples can avoid the high cost of training a large number of fault data samples for traditional artificial intelligence technology.

[0005] In order to achieve the above-mentioned purpose, the application has the following technical solutions:

[0006] A reciprocating compressor fault diagnosis method based on p-V diagram characteristic parameters, comprising the following steps:

[0007] Extracting characteristic parameters of the p-V diagram under normal operating conditions, wherein the characteristic parameters include the suction and exhaust pressure ratio of the p-V diagram, the equivalent area, the pressure term centroid coordinates and the expansion process comprehensive index;

[0008] The BP neural network model is built with the suction and exhaust pressure ratio, equivalent area, and pressure term barycenter coordinate of the p-V diagram as input parameters and the expansion process comprehensive index of the p-V diagram as output parameter;

[0009] The BP neural network model is trained with the characteristic parameters of the p-V diagram under normal operating conditions and the system safety threshold is determined;

[0010] The characteristic parameters of the p-V diagram under the to-be-identified operating conditions are extracted and substituted into the trained BP neural network model to calculate the average deviation of multiple groups of actual output parameters and expected output results;

[0011] The average deviation is compared with the system safety threshold to diagnose whether a fault occurs and locate the fault type.

[0012] Preferably, the expansion process comprehensive index of the p-V diagram is calculated according to the following formula:

[0013] pV k =const

[0014] In the formula, k is the process index; and const is a constant.

[0015] The logarithmic processing is performed on both sides of the equation to obtain the following result:

[0016] lnp+klnV=const

[0017] The p-V diagram is converted into an lnp-lnV diagram, and then the linear fitting is performed on the expansion process curve of the diagram to obtain the approximate value of the expansion process index, which is defined as the expansion process comprehensive index.

[0018] Preferably, the suction and exhaust pressure ratio of the p-V diagram is calculated according to the following formula:

[0019]

[0020] In the formula, p is the average suction pressure; is the average exhaust pressure;

[0021] The calculation formulae of the two parameters are as follows:

[0022]

[0023]

[0024] In the formula, N is the number of pressure points collected in the working process; p icorresponding to pressure; a represents the end time of the exhaust process; b represents the start time of the intake process; c represents the end time of the intake process; d represents the start time of the exhaust process; subscript s represents the intake process; and subscript d represents the exhaust process.

[0025] As preferred, the equivalent area of the p-V diagram is calculated according to the following formula:

[0026]

[0027] As preferred, the pressure term centroid coordinate of the p-V diagram is calculated according to the following formula:

[0028]

[0029] As preferred, when training the BP neural network model, the characteristic parameters of the p-V diagram under the normal operation condition of the compressor are substituted into the model for iterative training until the training error is less than a specified value or the preset number of iterations is reached, and the training is completed.

[0030] As preferred, the system safety threshold is characterized by the average absolute error between the predicted value and the actual value of the expansion process comprehensive index obtained by the BP neural network model, and the calculation expression is as follows:

[0031]

[0032] In the formula, e is the system error; m is the actual expansion process comprehensive index of the training sample; is the predicted expansion process comprehensive index of the training sample; and N is the total number of training samples.

[0033] As preferred, when the characteristic parameters of the p-V diagram under the to-be-identified operation condition are substituted into the trained BP neural network model to calculate the average deviation of multiple sets of actual output parameters and expected output results, the average deviation is calculated according to the following formula:

[0034]

[0035] In the formula, e' is the average deviation between the predicted expansion process comprehensive index and the actual expansion process comprehensive index of the to-be-identified sample; m' is the actual expansion process comprehensive index of the to-be-identified sample; is the predicted expansion process comprehensive index of the to-be-identified sample; and M is the total number of to-be-identified samples.

[0036] The calculated average deviation e' and the system error e are compared, if |e'| is less than the system error |e|, it is diagnosed that the compressor is in normal operation and no fault occurs; if |e'| is greater than the system error |e|, it is diagnosed that the compressor has an in-cylinder wear part fault.

[0037] As preferred, the average deviation is compared with the system safety threshold, and the fault type is located according to the comparison result according to the following table:

[0038]

[0039] A reciprocating compressor fault diagnosis system based on p-V diagram characteristic parameters, comprising:

[0040] A normal operation condition parameter extraction module is configured to extract characteristic parameters of a p-V diagram under a normal operation condition, wherein the characteristic parameters include a suction and discharge pressure ratio of the p-V diagram, an equivalent area, a pressure term barycenter coordinate, and an expansion process comprehensive index.

[0041] A BP neural network model building module is configured to build a BP neural network model by taking the suction and discharge pressure ratio of the p-V diagram, the equivalent area, and the pressure term barycenter coordinate as input parameters and taking the expansion process comprehensive index of the p-V diagram as an output parameter.

[0042] A safety threshold determination module is configured to train the BP neural network model by using the characteristic parameters of the p-V diagram under the normal operation condition and determine a system safety threshold.

[0043] An operation condition parameter deviation calculation module is configured to extract the characteristic parameters of a p-V diagram under a to-be-identified operation condition, and calculate an average deviation between a plurality of actual output parameters and an expected output result by substituting the characteristic parameters into the trained BP neural network model.

[0044] A fault diagnosis positioning module is configured to compare the average deviation with the system safety threshold, diagnose whether a fault occurs, and locate the fault type.

[0045] Compared with the prior art, the present application has at least the following beneficial effects:

[0046] By extracting the characteristic parameters of the p-V diagram, building the BP neural network model by using the characteristic parameters of the p-V diagram, training the BP neural network model by using the characteristic parameters extracted under the normal operation condition and determining the system safety threshold, and comparing the changes of the p-V diagram characteristic parameters under the to-be-identified operation condition with the system safety threshold, whether the reciprocating compressor vulnerable parts have a fault can be diagnosed, and the fault type can be located by comparing the average deviation with the system safety threshold. The present application solves the problems that the fault diagnosis signal processing and feature extraction process are complex, and a large amount of fault data is required for training samples. The fault signal extraction feature is more obvious, the engineering difficulty of applying artificial intelligence technology to compressor fault diagnosis is reduced, and the p-V diagram data under the normal operation condition is used as the training sample, thereby avoiding the high cost of a large amount of fault data training samples required by traditional artificial intelligence technology. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1is a flow chart of a reciprocating compressor fault diagnosis method based on p-V diagram characteristic parameters according to an embodiment of the present application;

[0048] Figure 2 is a p-V diagram of a reciprocating compressor according to an embodiment of the present application;

[0049] Figure 3 is an lnp-lnV diagram of a reciprocating compressor according to an embodiment of the present application;

[0050] Figure 4 is a BP neural network structure diagram of a parameter prediction model according to an embodiment of the present application. DETAILED DESCRIPTION

[0051] The present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0052] As shown in the drawings, Figure 1 a reciprocating compressor fault diagnosis method based on p-V diagram characteristic parameters according to an embodiment of the present application, the fault diagnosis process is divided into two parts, namely the training part and the diagnosis part. The training part refers to the construction of a process index prediction model based on a BP neural network model. The p-V diagram data under normal working conditions is used as the training sample to calculate the error value e between the actual output and the expected output of the model, which is used as the system error of the model training. The role of the diagnosis part is to determine whether the input p-V diagram data of the to-be-diagnosed working condition has a fault and the specific fault type. The implementation method is to extract the characteristic parameters of the p-V diagram of the to-be-diagnosed working condition, use the prediction model constructed by training to solve the comprehensive process index, then calculate the average deviation e' between the calculated output value and the actual measured value at this time, and compare it with the system error e. If |e'| is less than the system error |e|, the compressor is diagnosed as normal operation without fault; if |e'| is greater than the system error |e|, the compressor is diagnosed as having an in-cylinder wear part fault. According to the different error performance results, the fault position of the compressor can be preliminarily located.

[0053] Specifically, the reciprocating compressor fault diagnosis method based on p-V diagram characteristic parameters according to an embodiment of the present application includes the following steps:

[0054] S1. Extracting the characteristic parameters of the p-V diagram under normal running conditions, the characteristic parameters including the suction and discharge pressure ratio of the p-V diagram, the equivalent area, the pressure item centroid coordinates and the expansion process comprehensive index;

[0055] S2. Building a BP neural network model with the suction and discharge pressure ratio of the p-V diagram, the equivalent area and the pressure item centroid coordinates as input parameters, and the expansion process comprehensive index of the p-V diagram as output parameters;

[0056] S3. Train the BP neural network model using the feature parameters of the pV diagram under normal operating conditions and determine the system safety threshold;

[0057] S4. Extract the feature parameters of the pV diagram under the operating condition to be identified, and substitute them into the trained BP neural network model to calculate the average deviation between multiple sets of actual output parameters and expected output results;

[0058] S5. Compare the average deviation with the system safety threshold to diagnose whether a fault has occurred and to locate the type of fault.

[0059] like Figure 2 As shown, Figure 2 The diagram shown is the pV diagram of a reciprocating compressor. Characteristic parameters of the pV diagram under normal operating conditions are extracted based on the pV diagram. In this embodiment of the invention, the suction-discharge pressure ratio, equivalent area, centroid coordinates of the pressure term, and comprehensive expansion process index of the pV diagram are selected as characteristic parameters for fault diagnosis. The extraction methods for these characteristic parameters are as follows:

[0060] a. Extract the comprehensive index of the pV diagram expansion process based on the sample data;

[0061] Based on the principle of reciprocating compressors, the formula for calculating process indices is:

[0062] pV k =const

[0063] In the formula, k is the process exponent; const is a constant;

[0064] Taking the logarithm of both sides of the equation, we get the following result:

[0065] lnp + klnV = const

[0066] It can be observed that the formula after logarithmic processing transforms into a linear relationship between lnp and lnV, with the process exponent being the slope. For example... Figure 3 As shown, for ease of calculation, the pV diagram can be converted into an lnp-lnV diagram. Then, by linearly fitting the expansion process curve of the diagram, an approximate value of the expansion process index can be obtained, which is defined as the comprehensive expansion process index.

[0067] b. Calculate the intake and exhaust pressure ratio ε based on the sample data;

[0068]

[0069] In the formula, The average inspiratory pressure; The average exhaust pressure;

[0070] The formulas for calculating the two parameters are as follows:

[0071]

[0072]

[0073] In the formula: N is the number of pressure points collected in the working process; p i is the corresponding pressure; a represents the end time of the exhaust process; b represents the start time of the intake process; c represents the end time of the intake process; d represents the start time of the exhaust process; the subscript s represents the intake process; and the subscript d represents the exhaust process.

[0074] c. Extracting the equivalent area A of the p-V diagram according to the sample data;

[0075] The equivalent area of the p-V diagram reflects the indicated power of the compressor. The failure of the cylinder internal wearing parts of the compressor will affect the indicated power of the compressor, which will cause a change in the equivalent area of the p-V diagram. The calculation formula of the equivalent area is as follows:

[0076]

[0077] d. Extracting the p-V diagram pressure item centroid coordinate C p

[0078] Since the shape of the p-V diagram is irregular, the centroid coordinate needs to be obtained by integration, and the calculation formula is as follows:

[0079]

[0080] Figure 4 The BP neural network structure of the parameter prediction model is as shown in the figure, and the BP neural network is built according to the structure. The input parameters of the BP neural network are the suction and exhaust pressure ratio ε of the p-V diagram, the equivalent area A of the diagram, and the centroid coordinate C of the pressure item diagram p . The output parameter is the expansion process comprehensive index m. In the embodiment of the present application, the number of hidden layer nodes is determined according to the following formula when the number of network hidden layers is 1:

[0081]

[0082] In step S3, when training the BP neural network model, the feature parameter samples extracted from the p-V diagram under the normal operating condition of the compressor are put into the model for iterative training until the training error is less than a specified value or the preset number of iterations is reached, and the training is completed.

[0083] The average absolute error between the predicted value and the actual value of the expansion process comprehensive index obtained by the calculation model represents the system error, that is, the safety threshold for judging the operating state of the compressor.

[0084] The calculation expression is as follows:

[0085]

[0086] In the formula: e is a system error; m is an actual expansion process comprehensive index of a training sample; is a predicted expansion process comprehensive index of the training sample; N is a total number of the training samples.

[0087] The fault state recognition process first extracts feature parameters of the input p-V diagram data under the running working condition to be recognized, including a suction and exhaust pressure ratio, an equivalent area of a p-V diagram, a pressure term barycenter coordinate (model input), and an expansion process comprehensive index (expected output). Deviation of the predicted expansion process comprehensive index of the p-V diagram sample to be recognized from the actual process comprehensive index is calculated through the following relation formula. The embodiment of the application considers the randomness influence of a single group of data, which may lead to an incorrect recognition result. Therefore, the fault recognition method of the application adopts a mode of calculating average values after calculating deviations of multiple groups of samples to be recognized, so as to obtain average deviation of multiple groups of actual outputs and expected outputs, as a final deviation for fault recognition.

[0088]

[0089] In the formula: e' is average deviation of the predicted expansion process comprehensive index of the sample to be recognized from the actual expansion process comprehensive index; m' is the actual expansion process comprehensive index of the sample to be recognized; is the predicted expansion process comprehensive index of the sample to be recognized; M is a total number of the samples to be recognized.

[0090] The calculated average deviation e' is compared with the system error e. If |e'| is less than the system error |e|, it is diagnosed that the compressor is in normal operation and no fault occurs. If |e'| is greater than the system error |e|, it is diagnosed that the compressor has an in-cylinder wear part fault.

[0091] Step S5, when positioning the fault type, determines that the compressor has an in-cylinder wear part fault when the deviation |e'| exceeds a safety threshold. According to the performance result of the deviation, in combination with the following fault logic diagnosis table, positioning of the fault can be realized.

[0092]

[0093] Another embodiment of the application further provides a reciprocating compressor fault diagnosis system based on p-V diagram feature parameters, comprising:

[0094] A normal running working condition parameter extraction module is used to extract feature parameters of a p-V diagram under a normal running working condition. The feature parameters include a suction and exhaust pressure ratio, an equivalent area of a p-V diagram, a pressure term barycenter coordinate, and an expansion process comprehensive index of the p-V diagram.

[0095] The BP neural network model building module is configured to build a BP neural network model by taking the suction and exhaust pressure ratio, equivalent area and pressure centroid coordinate of the p-V diagram as input parameters and taking the expansion process comprehensive index of the p-V diagram as output parameters.

[0096] The safety threshold determination module is configured to train the BP neural network model by using the characteristic parameters of the p-V diagram under normal operation conditions and determine a system safety threshold.

[0097] The operation condition parameter deviation calculation module is configured to extract the characteristic parameters of the p-V diagram under the to-be-identified operation condition, and input the characteristic parameters into the trained BP neural network model to calculate the average deviation of multiple groups of actual output parameters from expected output results.

[0098] The fault diagnosis positioning module is configured to compare the average deviation with the system safety threshold, diagnose whether a fault occurs and locate the fault type.

[0099] The present application makes the fault signal extraction feature more obvious, reduces the engineering difficulty of applying artificial intelligence technology to compressor fault diagnosis, and uses the P-V diagram data under normal operation conditions as training samples to avoid the high cost of a large number of fault data training samples required by traditional artificial intelligence technology.

[0100] It should be noted that the information interaction and execution process between the above module units are based on the same concept as the method embodiments, and the specific functions and technical effects brought by them can be referred to the method embodiment part, which will not be repeated here.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of functional units and modules is taken as an example for illustration, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit or module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The above integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit or module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0102] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium at least includes any entity or device capable of carrying the computer program code to the photographing device / terminal equipment, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. For example, U disk, mobile hard disk, magnetic disk or optical disk, etc.

[0103] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in a certain embodiment can be referred to the relevant description of other embodiments.

[0104] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A reciprocating compressor fault diagnosis method based on p-V diagram characteristic parameters, characterized in that, The method comprises the following steps: extracting characteristic parameters of the p-V diagram under normal operating conditions, the characteristic parameters comprising a suction and exhaust pressure ratio, an equivalent area, a pressure term barycenter coordinate, and an expansion process comprehensive index of the p-V diagram; building a BP neural network model with the suction and exhaust pressure ratio, the equivalent area, and the pressure term barycenter coordinate of the p-V diagram as input parameters and the expansion process comprehensive index of the p-V diagram as an output parameter; training the BP neural network model with the characteristic parameters of the p-V diagram under normal operating conditions and determining a system safety threshold value; extracting the characteristic parameters of the p-V diagram under the to-be-identified operating conditions, substituting the characteristic parameters into the trained BP neural network model to calculate average deviations of multiple groups of actual output parameters from expected output results; comparing the average deviations with the system safety threshold value to diagnose whether a fault has occurred and locate a fault type; the expansion process comprehensive index of the p-V diagram is calculated according to the following formula: wherein k is the process index; const is a constant; logarithmic processing is performed on both sides of the equation to obtain the following result: the p-V diagram is converted into an lnp-lnV diagram, and then a linear fitting is performed on an expansion process curve of the diagram to obtain an approximate value of the expansion process index, which is defined as the expansion process comprehensive index; the system safety threshold value is characterized by an average absolute error between a predicted value and an actual value of the expansion process comprehensive index obtained by the BP neural network model, and the calculation expression is as follows: In the formula: e is the system error; m is the actual integrated index of the expansion process of the training sample; is the predicted integrated index of the expansion process of the training sample; N is the total number of training samples.

2. The method according to claim 1, wherein, the suction and exhaust pressure ratio of the p-V diagram is calculated according to the following formula: wherein is the average intake pressure; is the average exhaust pressure; the calculation formulas of the two parameters are as follows: In the formula: N is the number of pressure points collected during the working process; p i is the corresponding pressure; a represents the end time of the exhaust process; b represents the start time of the intake process; c represents the end time of the intake process; d represents the start time of the exhaust process; subscript s represents the intake process; subscript d represents the exhaust process.

3. The method according to claim 2, wherein, the equivalent area of the p-V diagram is calculated according to the following formula: 。 4. The method according to claim 3, wherein, the pressure term barycenter coordinate of the p-V diagram is calculated according to the following formula: 。 5. The method according to claim 1, wherein, when training the BP neural network model, the characteristic parameters of the p-V diagram under normal operating conditions of the sufficient compressor are substituted into the model for iterative training until the training error is less than a specified value or a preset number of iterations is reached, and the training is completed.

6. The method for diagnosing a reciprocating compressor fault based on p-V diagram characteristic parameters according to claim 1, characterized in that, when extracting the characteristic parameters of the p-V diagram under the to-be-identified operating conditions, substituting the characteristic parameters into the trained BP neural network model to calculate average deviations of multiple groups of actual output parameters from expected output results, the average deviations are calculated according to the following formula: In the formula: is the average deviation of the predicted comprehensive index of the inflation process of the sample to be identified from the actual comprehensive index of the inflation process; is the actual comprehensive index of the inflation process of the sample to be identified; is the predicted comprehensive index of the inflation process of the sample to be identified; M is the total number of to-be-identified samples; The calculated average deviation is compared with the system error e If the system error is less than the system error e , it is diagnosed that the compressor is running normally without any fault; if the system error is greater than the system error e , it is diagnosed that the compressor has a cylinder wear part fault.

7. The method according to claim 6, wherein the p-V diagram characteristic parameter is a pressure ratio (PR) defined as: PR = (Pmax - Pmin) / (Pmax + Pmin) where Pmax and Pmin are the maximum and minimum pressures of the p-V diagram, respectively. comparing the average deviations with the system safety threshold value, locating a fault type according to a comparison result according to the following table: 。 8. A reciprocating compressor fault diagnosis system based on p-V diagram characteristic parameters, characterized in that, comprise: a normal operating condition parameter extraction module configured to extract characteristic parameters of a p-V diagram under normal operating conditions, the characteristic parameters comprising a suction and exhaust pressure ratio, an equivalent area, a pressure term barycenter coordinate, and an expansion process comprehensive index of the p-V diagram; a BP neural network model building module configured to build a BP neural network model with the suction and exhaust pressure ratio, the equivalent area, and the pressure term barycenter coordinate of the p-V diagram as input parameters and the expansion process comprehensive index of the p-V diagram as an output parameter; a safety threshold value determination module configured to train the BP neural network model with the characteristic parameters of the p-V diagram under normal operating conditions and determine a system safety threshold value; an operating condition parameter deviation calculation module configured to extract the characteristic parameters of the p-V diagram under the to-be-identified operating conditions, substituting the characteristic parameters into the trained BP neural network model to calculate average deviations of multiple groups of actual output parameters from expected output results; The fault diagnosis positioning module is configured to compare the average deviation with a system safety threshold, diagnose whether a fault occurs, and locate a fault type; The expansion process comprehensive index of the p-V diagram is calculated according to the following formula: wherein k is a process index; const is a constant; The logarithmic processing is performed on both sides of the equation to obtain the following result: The p-V diagram is converted into an lnp-lnV diagram, and then the linear fitting is performed on the expansion process curve of the diagram to obtain an approximate value of the expansion process index, which is defined as the expansion process comprehensive index; The system safety threshold is characterized by the average absolute error between the expansion process comprehensive index prediction value and the actual value of the BP neural network model, and the calculation expression is as follows: In the formula: e is the system error; m is the actual integrated index of the expansion process of the training sample; is the predicted integrated index of the expansion process of the training sample; N is the total number of training samples.

Images