Electrolytic tank fault analysis method, device and equipment, storage medium and related product

Abstract

The invention discloses an electrolytic cell fault analysis method, device and equipment, a storage medium and a related product. The method comprises the following steps: acquiring a training data set; performing data supplementation on the training data set to generate a supplemental data set; the supplementary data set comprises a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set; performing clustering processing on the supplementary data set, and establishing a corresponding relation between each fault type and response characteristics; the response features are common feature data of multiple groups of basic information corresponding to the fault type; and on the basis of the corresponding relation between each fault type and the response feature, a fault type and response feature matching database is constructed, electrolytic cell fault diagnosis is carried out on the basis of the fault type and response feature matching database, and fault diagnosis can be carried out on the electrolytic cell on the basis of a small number of training data sets.

Description

Technical Field

[0001] The present application relates to the technical field of hydrogen production by electrolysis of water, and in particular to an electrolytic cell fault analysis method, device, equipment, storage medium and related products. Background Art

[0002] The rapid development of renewable energy sources such as wind power and photovoltaics has brought about a large number of renewable energy fluctuation power consumption problems. Hydrogen energy, as a clean and efficient energy storage method, has received more and more attention. Alkaline water electrolysis hydrogen production technology is currently a relatively mature water electrolysis hydrogen production technology. With the increase in the shipment of large alkaline electrolyzers and the high cost of maintenance of large alkaline electrolyzers, methods for online status monitoring and fault diagnosis of alkaline electrolyzers have become particularly important.

[0003] Traditional fault diagnosis methods for alkaline electrolyzers include: directly monitoring the various operating parameters of the electrolyzer and fault diagnosis based on machine learning. Directly monitoring the various operating parameters of the electrolyzer has many problems, such as multiple data collection requirements.

[0004] Although fault diagnosis based on machine learning overcomes the problem of multiple data collection requirements in directly monitoring the various operating status parameters of the electrolyzer, there are also other problems. For example, the training process of machine learning requires a large amount of training data, but in fact, for alkaline electrolyzers, especially large alkaline electrolyzers, the experimental acquisition of fault operation data is difficult, costly, and has a high risk factor. If multiple faults are artificially created in large alkaline cells, not only will the maintenance cost be very high, but many faults involve cross-talk between the hydrogen and oxygen electrodes, and the increase of oxygen content in hydrogen and hydrogen content in oxygen, and the risk factor of the experiment is high. In order to obtain a sufficient amount of data set to support machine learning methods, it is difficult and costly to conduct experiments on actual electrolyzers alone. Summary of the invention

[0005] Based on the above problems, the present application provides an electrolytic cell fault analysis method, device, equipment, storage medium and related products, the purpose of which is to diagnose electrolytic cell faults based on a small amount of training data set.

[0006] The embodiments of the present application disclose the following technical solutions:

[0007] The first aspect of the present application provides a method for analyzing electrolytic cell failure, the method comprising:

[0008] Acquire a training data set; the training data set includes a fault type set of the electrolytic cell and a set of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution, and uneven thermal field distribution; the basic information includes current, voltage, and current response obtained after applying a sinusoidal AC voltage signal;

[0009] Supplementing the training data set to generate a supplemented data set; the supplemented data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set;

[0010] Performing clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature; the response feature is common feature data of multiple groups of basic information corresponding to the fault type;

[0011] Based on the corresponding relationship between each fault type and the response feature, a fault type and response feature matching database is constructed, and electrolytic cell fault diagnosis is performed based on the fault type and response feature matching database.

[0012] Optionally, the obtaining of a training data set specifically includes:

[0013] Five faulty electrolytic cells are obtained; the fault types of the five faulty electrolytic cells correspond to the five fault types one by one;

[0014] Operate each faulty electrolytic cell separately, and collect the current and voltage of each faulty electrolytic cell;

[0015] A sinusoidal AC voltage signal is applied to each faulty electrolytic cell, and the current response of each faulty electrolytic cell is collected.

[0016] Optionally, the step of supplementing the training data set to generate a supplemented data set specifically includes:

[0017] Constructing a data supplement model based on a generative adversarial neural network; the data supplement model includes a generative neural network and a discriminative neural network;

[0018] The data supplement model is trained based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain a trained data supplement model;

[0019] Inputting a set of basic information corresponding to each fault type in the training data set into the trained data supplementation model to obtain multiple sets of basic information corresponding to each fault type;

[0020] A supplementary data set is generated based on the training data set and multiple groups of basic information corresponding to each group of fault types.

[0021] Optionally, the training of the data supplement model based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain the trained data supplement model specifically includes:

[0022] Inputting a set of basic information corresponding to each fault type in the training data set into the generative neural network to obtain an output of the generative neural network, and inputting the output of the generative neural network and a set of basic information corresponding to each fault type in the training data set into an initial discriminant neural network to obtain an output of the discriminant neural network;

[0023] Based on the output of the discriminant neural network and the generative adversarial loss function, the generative neural network is iteratively trained. Based on the output of the discriminant neural network and the discriminant adversarial loss function, the discriminant neural network is iteratively trained until the value of the generative adversarial loss function is minimized and the value of the discriminant adversarial loss function is maximized, thereby obtaining the trained data supplementation model.

[0024] Optionally, clustering the supplementary data set to establish a corresponding relationship between each fault type and a response feature specifically includes:

[0025] Performing fast Fourier transform on multiple groups of basic information corresponding to each fault type in the fault type set in the supplementary data set to obtain high-frequency harmonic amplitude and phase data, and performing normalized data preprocessing to obtain a processed supplementary data set; the processed supplementary data set includes the fault type set and high-frequency harmonic data of multiple groups of basic information corresponding to each fault type in the fault type set;

[0026] Based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set, determining the common characteristic data of the multiple groups of high-frequency harmonic data of the basic information corresponding to each fault type in the fault type set as the response characteristic of each fault type;

[0027] The corresponding relationship between each fault type and the response characteristic is determined based on the response characteristic of each fault type.

[0028] Optionally, based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set, determining the common characteristic data of the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set as the response characteristic of each fault type specifically includes:

[0029] The DBSCAN clustering algorithm is used to define clusters based on the data point aggregation density to analyze the high-frequency harmonic data of multiple groups of basic information corresponding to each fault type, and the common characteristic data of multiple groups of basic information corresponding to each fault type are determined as the response characteristics of each fault type.

[0030] The second aspect of the present application provides an electrolytic cell fault analysis device, the device comprising:

[0031] an acquisition unit, for acquiring a training data set; the training data set includes a fault type set of an electrolytic cell and a set of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution, and uneven thermal field distribution; the basic information includes current, voltage, and a current response obtained after applying a sinusoidal AC voltage signal;

[0032] A supplementing unit, used to supplement the training data set to generate a supplemented data set; the supplemented data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set;

[0033] A clustering unit, used to perform clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature; the response feature is the common feature data of multiple groups of basic information corresponding to the fault type;

[0034] The diagnosis unit is used to construct a fault type and response feature matching database based on the corresponding relationship between each fault type and the response feature, and perform electrolytic cell fault diagnosis based on the fault type and response feature matching database.

[0035] A third aspect of the present application provides a computer device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method provided in the first aspect.

[0036] A fourth aspect of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method provided in the first aspect is implemented.

[0037] A fifth aspect of the present application provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the method provided in the first aspect is implemented.

[0038] Compared with the prior art, this application has the following beneficial effects:

[0039] The present application first obtains a set of information corresponding to each fault type, and then supplements the data of a set of information corresponding to each fault type to construct a supplementary data set, so that the supplementary data set includes multiple sets of information corresponding to each fault type, and realizes obtaining a supplementary data set based on a small number of training data sets, and then clustering the multiple sets of basic information corresponding to each fault type in the fault type set in the supplementary data set, and determining the common characteristic data of the multiple sets of basic information corresponding to each fault type as the response feature, thereby determining the corresponding relationship between the fault type and the response feature, and finally performing electrolytic cell fault diagnosis based on the determined corresponding relationship. The data basis of the present application is a set of basic information corresponding to each fault type, and data compensation is performed on this, in order to avoid the problem of low data accuracy caused by too little data in the subsequent clustering process. Because the present application supplements the data, it can also ensure the accuracy of the established corresponding relationship between the fault type and the response feature when there is only a small amount of training data. Further, the accuracy of the database established based on this corresponding relationship and the accuracy of the diagnostic results of the fault diagnosis based on the database are ensured. It can be seen that the present application realizes accurate diagnosis of electrolytic cell faults based on a small amount of training data. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0041] Figure 1 A flow chart of a method for analyzing electrolytic cell failure provided in an embodiment of the present application;

[0042] Figure 2 A flowchart for supplementing the data provided in the embodiments of the present application;

[0043] Figure 3 A flow chart of clustering processing provided in an embodiment of the present application;

[0044] Figure 4 This is a structural diagram of the electrolytic cell fault analysis device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] As described above, although the current fault diagnosis based on machine learning overcomes the problem of multiple data collection requirements in directly monitoring the various operating status parameters of the electrolytic cell, it also requires a large amount of training data for training. After research, the inventors found that a small amount of training data can be supplemented, and then fault diagnosis can be performed based on the supplemented data set, thereby achieving accurate diagnosis of electrolytic cell faults based on a small amount of training data.

[0046] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] Figure 1 This is a flow chart of a method for analyzing electrolytic cell failure. Figure 1 Explanation of the electrolytic cell fault analysis method:

[0048] S101. Obtain a training data set.

[0049] The training data set includes a fault type set of the electrolytic cell and a set of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution and uneven thermal field distribution; the fault types are generally the above five fault types, and these five fault types are also the research focus of this application.

[0050] The basic information includes the current, voltage, and the current response obtained after applying a sinusoidal AC voltage signal. The basic information is generally obtained by experiment, so it is necessary to obtain the electrolytic cell for each fault type.

[0051] Electrolytic cells with fault types of electrode corrosion or diaphragm corrosion can be obtained from alkaline electrolytic cell parts that have been running stably at a certain time, and after confirming the degree of corrosion and performance degradation through electrochemical methods, they can be used to construct experimental fault states.

[0052] The electrolytic cell with abnormal gas accumulation at the electrode plate is designed with a visual window at the end plate, which can monitor the accumulation of product gas at the electrode plate in real time. By adjusting the gas production rate of the device, the experimental fault state can be controlled.

[0053] An electrolytic cell with uneven flow field distribution or uneven thermal field distribution can be constructed by applying artificial disturbance to a stably operating alkaline electrolytic cell, disturbing the flow field by mechanical methods and controlled liquid supply methods, or disturbing the thermal field by arranging resistance wires for fixed-point heating.

[0054] Now let's explain how to obtain the training data set:

[0055] Five faulty electrolytic cells are obtained; the fault types of the five faulty electrolytic cells correspond to the five fault types one by one.

[0056] The faulty operation state of the alkaline electrolytic cell with the determined fault type was manufactured to obtain five faulty electrolytic cells. In the subsequent experimental process, the corresponding faulty operation condition was manufactured for each faulty electrolytic cell.

[0057] Each faulty electrolytic cell is operated separately, and the current and voltage of each faulty electrolytic cell are collected.

[0058] In the process of acquiring the training data set, the present application applies a sinusoidal AC voltage signal to the electrolytic cell to collect the current response signal of the faulty electrolytic cell, and analyzes the high-frequency harmonic data of the basic information obtained, so the harmonic distortion analysis method is specifically applied.

[0059] The application of harmonic distortion analysis method in this application is now specifically described:

[0060] A sinusoidal AC voltage signal is applied to each faulty electrolytic cell, and the current response of each faulty electrolytic cell is collected.

[0061] When each faulty electrolytic cell is running stably, a sinusoidal AC voltage signal with a frequency of F and an amplitude of A is applied to it, and the current response of each faulty electrolytic cell is collected.

[0062] The values ​​of F and A can be adjusted according to actual conditions.

[0063] S102: Supplement the training data set to generate a supplemented data set.

[0064] The supplementary data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set.

[0065] Multiple groups of basic information corresponding to each fault type are obtained through data supplementation, thus avoiding the danger of multiple data collection for the faulty electrolytic cell.

[0066] Figure 2 For a flowchart to supplement the data, see Figure 2 The process of data supplementation, i.e. S102, is described below:

[0067] S201. Construct a data supplement model based on a generative adversarial neural network; the data supplement model includes a generative neural network and a discriminative neural network.

[0068] S202, training a data supplement model based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain a trained data supplement model, specifically including:

[0069] This application uses a generative adversarial neural network to construct a data supplement model, and uses a small amount of training data sets as the basis for the generative neural network and the discriminative neural network to engage in game play, so that the data generated by the generative neural network can be infinitely close to the real basic information provided by the training data sets, thereby obtaining a trained data supplement model.

[0070] A set of basic information corresponding to each fault type in the training data set is input into the generative neural network to obtain the output of the generative neural network, and the output of the generative neural network and a set of basic information corresponding to each fault type in the training data set are input into the initial discriminant neural network to obtain the output of the discriminant neural network.

[0071] Based on the output of the discriminant neural network and the generative adversarial loss function, the generative neural network is iteratively trained. Based on the output of the discriminant neural network and the discriminant adversarial loss function, the discriminant neural network is iteratively trained until the value of the generative adversarial loss function is minimized and the value of the discriminant adversarial loss function is maximized, thereby obtaining the trained data supplementation model.

[0072] The generated adversarial loss function LG is:

[0073] LG=-E z :p z (z)[logD(G(z))].

[0074] The discriminative adversarial loss function LD is:

[0075] LD=-E x :p data (x)[logD(x)]+E z :p z (z)[log(1-D(G(z))].

[0076] In both formulas, G() is the generator; D() is the discriminator; E z To generate the expected value; E x is the expected value of discrimination; p z (z) is a random noise distribution, which serves to improve the richness of the initial generated samples of the generator; p data (x) is the target data distribution.

[0077] S203, inputting a set of basic information corresponding to each fault type in the training data set into the trained data supplementation model to obtain multiple sets of basic information corresponding to each fault type.

[0078] The trained data supplement model can generate multiple sets of basic information based on a set of basic information in the training data, mainly because the trained data supplement model learns the data composition of a set of basic information.

[0079] S204: Generate a supplementary data set based on the training data set and multiple groups of basic information corresponding to each group of fault types.

[0080] Through the above steps S201 to S204, the data supplement of the training data set is achieved. Next, the corresponding relationship between the fault type and the response feature can be established based on the supplemented data set.

[0081] S103: performing clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature.

[0082] The response characteristics are common characteristic data of multiple groups of basic information corresponding to the fault type.

[0083] Find the common characteristic data of multiple groups of basic information, and use this common characteristic data as the identifier of this fault type. That is, when a fault occurs in the electrolytic cell, the fault information that appears includes the response characteristics of a certain fault, then the fault type of this faulty electrolytic cell can be limited to the fault type corresponding to the response characteristics.

[0084] Figure 3 For the flow chart of clustering processing, please refer to Figure 3 The clustering process (ie S103) is described as follows:

[0085] S301, performing fast Fourier transform on multiple groups of basic information corresponding to each fault type in the fault type set in the supplementary data set to obtain high-frequency harmonic amplitude and phase data, and performing normalized data preprocessing to obtain a processed supplementary data set.

[0086] The processed supplementary data set includes a fault type set and high-frequency harmonic data of multiple groups of basic information corresponding to each fault type in the fault type set.

[0087] S302: Based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set, determine the common characteristic data of the multiple groups of high-frequency harmonic data of the basic information corresponding to each fault type in the fault type set as the response characteristic of each fault type.

[0088] The DBSCAN clustering algorithm is used to define clusters based on the data point aggregation density to analyze the high-frequency harmonic data of multiple groups of basic information corresponding to each fault type, and the common characteristic data of multiple groups of basic information corresponding to each fault type are determined as the response characteristics of each fault type.

[0089] The specific application of DBSCAN clustering algorithm is:

[0090] After completing the normalization of high-frequency harmonic data, the core parameter neighborhood radius and the minimum number of samples to form a cluster are selected by drawing the k-distance distribution diagram of the data points.

[0091] After defining the core object, the algorithm runs in a way that a cluster is formed by traversing the sample set that can be reached with the core object density based on the core parameter neighborhood radius and the minimum number of samples that constitute the cluster. Based on this, all clusters in the data set are traversed to complete the classification of similar features and obtain the response features.

[0092] The parameter adjustment method based on the k-distance distribution graph is to calculate the k-distance of each point in the data set and arrange them in ascending order, check the inflection point of the distance change, and select the position where the distance change is extremely large as the core parameter neighborhood radius.

[0093] S303: Determine a correspondence between each fault type and a response feature based on the response feature of each fault type.

[0094] Through the above steps S301 to S303, the corresponding relationship between the fault type and the response characteristic is established. Next, a database can be constructed based on the corresponding relationship between the fault type and the response characteristic, and then the electrolytic cell fault diagnosis is performed based on the database.

[0095] S104: construct a fault type and response feature matching database based on the corresponding relationship between each fault type and the response feature, and perform electrolytic cell fault diagnosis based on the fault type and response feature matching database.

[0096] The present application can better contain the nonlinear information of the electrolytic cell's faulty operation, can better characterize the characteristics of the faulty operation, and better extract key features. At the same time, since the response speed of the electrical signal is much faster than physical quantities such as gas content and flow rate, real-time monitoring of the fault state can be achieved. At the same time, since the harmonics contain the real-time state of the electrolytic cell's operation, some possible trends can even achieve advance prediction of faults.

[0097] In addition, the present application supplements the training data and data, and can obtain a large amount of available data with fewer actual machine experiments, avoiding a large number of large-scale electrolytic cell actual machine failure experiments, reducing R&D costs, and at the same time ensuring safety to a greater extent.

[0098] Figure 4For the structural diagram of the electrolytic cell fault analysis device, see Figure 4 The function of each unit in the electrolytic cell fault analysis device is described as follows:

[0099] The acquisition unit is used to acquire the training data set.

[0100] The training data set includes a fault type set of the electrolytic cell and a group of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution and uneven thermal field distribution.

[0101] The basic information includes current, voltage, and current response obtained after applying a sinusoidal AC voltage signal. The basic information is generally obtained through experiments, so it is necessary to obtain an electrolytic cell for each fault type.

[0102] Electrolytic cells with fault types of electrode corrosion or diaphragm corrosion can be obtained from alkaline electrolytic cell parts that have been running stably at a certain time, and after confirming the degree of corrosion and performance degradation through electrochemical methods, they can be used to construct experimental fault states.

[0103] The electrolytic cell with abnormal gas accumulation at the electrode plate is designed with a visual window at the end plate, which can monitor the accumulation of product gas at the electrode plate in real time. By adjusting the gas production rate of the device, the experimental fault state can be controlled.

[0104] An electrolytic cell with uneven flow field distribution or uneven thermal field distribution can be constructed by applying artificial disturbance to a stably operating alkaline electrolytic cell, disturbing the flow field by mechanical methods and controlled liquid supply methods, or disturbing the thermal field by arranging resistance wires for fixed-point heating.

[0105] The acquisition unit includes an acquisition module, a first acquisition module and a second acquisition module. The function of each module in the acquisition unit is now described:

[0106] The acquisition module is used to acquire five faulty electrolytic cells; the fault types of the five faulty electrolytic cells correspond to the five fault types one by one.

[0107] The first acquisition module is used to operate each faulty electrolytic cell separately and acquire the current and voltage of each faulty electrolytic cell.

[0108] The second acquisition module is used to apply a sinusoidal AC voltage signal to each faulty electrolytic cell and acquire the current response of each faulty electrolytic cell.

[0109] A supplementing unit, used to supplement the training data set to generate a supplemented data set; the supplemented data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set;

[0110] The supplementary unit includes a construction module, a training module, an application module, and a generation module. The functions of each module in the supplementary unit are described below:

[0111] A construction module is used to construct a data supplement model based on a generative adversarial neural network; the data supplement model includes a generative neural network and a discriminative neural network.

[0112] The training module is used to train the data supplement model based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain a trained data supplement model.

[0113] The function of the training module can be refined as follows: input a set of basic information corresponding to each fault type in the training data set into the generative neural network to obtain the output of the generative neural network, input the output of the generative neural network and a set of basic information corresponding to each fault type in the training data set into the initial discriminant neural network to obtain the output of the discriminant neural network. Based on the output of the discriminant neural network and the generative adversarial loss function, the generative neural network is iteratively trained, and based on the output of the discriminant neural network and the discriminant adversarial loss function, the discriminant neural network is iteratively trained until the value of the generative adversarial loss function is minimized and the value of the discriminant adversarial loss function is maximized, thereby obtaining the trained data supplementation model.

[0114] The application module is used to input a set of basic information corresponding to each fault type in the training data set into the trained data supplement model to obtain multiple sets of basic information corresponding to each fault type.

[0115] A generating module is used to generate a supplementary data set based on the training data set and multiple groups of basic information corresponding to each group of fault types.

[0116] A clustering unit, used to perform clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature; the response feature is the common feature data of multiple groups of basic information corresponding to the fault type;

[0117] The clustering unit includes a processing module, a response feature determination module and a corresponding relationship determination module. The function of each module in the clustering unit is now described:

[0118] The processing module is used to perform fast Fourier transform on multiple groups of basic information corresponding to each fault type in the fault type set in the supplementary data set to obtain high-frequency harmonic amplitude and phase data, and perform normalized data preprocessing to obtain a processed supplementary data set.

[0119] The processed supplementary data set includes a fault type set and high-frequency harmonic data of multiple groups of basic information corresponding to each fault type in the fault type set.

[0120] A response feature determination module is used to determine the common feature data of the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set as the response feature of each fault type based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set.

[0121] The DBSCAN clustering algorithm is used to define clusters based on the data point aggregation density to analyze the high-frequency harmonic data of multiple groups of basic information corresponding to each fault type, and the common characteristic data of multiple groups of basic information corresponding to each fault type are determined as the response characteristics of each fault type.

[0122] The corresponding relationship determination module is used to determine the corresponding relationship between each fault type and the response feature based on the response feature of each fault type.

[0123] The diagnosis unit is used to construct a fault type and response feature matching database based on the corresponding relationship between each fault type and the response feature, and perform electrolytic cell fault diagnosis based on the fault type and response feature matching database.

[0124] A computer device comprises: a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement an electrolytic cell fault analysis method.

[0125] A computer-readable storage medium stores a computer program, which implements an electrolytic cell fault analysis method when executed by a processor.

[0126] A computer program product comprises a computer program, which implements an electrolytic cell fault analysis method when executed by a processor.

[0127] It should be noted that each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, storage medium and product embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments. The above-described device, equipment, storage medium and product embodiments are merely schematic, in which the units described as separate components may or may not be physically separated, and the components indicated as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0128] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for analyzing electrolytic cell failure, characterized in that: The method comprises: Acquire a training data set; the training data set includes a fault type set of the electrolytic cell and a set of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution, and uneven thermal field distribution; the basic information includes current, voltage, and current response obtained after applying a sinusoidal AC voltage signal; Supplementing the training data set to generate a supplemented data set; the supplemented data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set; Performing clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature; the response feature is common feature data of multiple groups of basic information corresponding to the fault type; Based on the corresponding relationship between each fault type and the response feature, a fault type and response feature matching database is constructed, and electrolytic cell fault diagnosis is performed based on the fault type and response feature matching database.

2. The method according to claim 1, characterized in that The obtaining of the training data set specifically includes: Five faulty electrolytic cells are obtained; the fault types of the five faulty electrolytic cells correspond to the five fault types one by one; Operate each faulty electrolytic cell separately, and collect the current and voltage of each faulty electrolytic cell; A sinusoidal AC voltage signal is applied to each faulty electrolytic cell, and the current response of each faulty electrolytic cell is collected.

3. The method according to claim 1, characterized in that The supplementing the training data set to generate a supplemented data set specifically includes: Constructing a data supplement model based on a generative adversarial neural network; the data supplement model includes a generative neural network and a discriminative neural network; The data supplement model is trained based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain a trained data supplement model; Inputting a set of basic information corresponding to each fault type in the training data set into the trained data supplementation model to obtain multiple sets of basic information corresponding to each fault type; A supplementary data set is generated based on the training data set and multiple groups of basic information corresponding to each group of fault types.

4. The method according to claim 3, characterized in that: The data supplement model is trained based on a set of basic information corresponding to each fault type in the fault type set in the training data set to obtain the trained data supplement model, specifically including: Inputting a set of basic information corresponding to each fault type in the training data set into the generative neural network to obtain an output of the generative neural network, and inputting the output of the generative neural network and a set of basic information corresponding to each fault type in the training data set into an initial discriminant neural network to obtain an output of the discriminant neural network; Based on the output of the discriminant neural network and the generative adversarial loss function, the generative neural network is iteratively trained. Based on the output of the discriminant neural network and the discriminant adversarial loss function, the discriminant neural network is iteratively trained until the value of the generative adversarial loss function is minimized and the value of the discriminant adversarial loss function is maximized, thereby obtaining the trained data supplementation model.

5. The method according to claim 1, characterized in that The clustering process of the supplementary data set to establish a corresponding relationship between each fault type and a response feature specifically includes: Performing fast Fourier transform on multiple groups of basic information corresponding to each fault type in the fault type set in the supplementary data set to obtain high-frequency harmonic amplitude and phase data, and performing normalized data preprocessing to obtain a processed supplementary data set; the processed supplementary data set includes the fault type set and high-frequency harmonic data of multiple groups of basic information corresponding to each fault type in the fault type set; Based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set, determining the common characteristic data of the multiple groups of high-frequency harmonic data of the basic information corresponding to each fault type in the fault type set as the response characteristic of each fault type; The corresponding relationship between each fault type and the response characteristic is determined based on the response characteristic of each fault type.

6. The method according to claim 5, characterized in that The determining, based on the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set, common feature data of the high-frequency harmonic data of the multiple groups of basic information corresponding to each fault type in the fault type set as the response feature of each fault type specifically includes: The DBSCAN clustering algorithm is used to define clusters based on the data point aggregation density to analyze the high-frequency harmonic data of multiple groups of basic information corresponding to each fault type, and the common characteristic data of multiple groups of basic information corresponding to each fault type are determined as the response characteristics of each fault type.

7. An electrolytic cell fault analysis device, characterized in that: The device comprises: an acquisition unit, for acquiring a training data set; the training data set includes a fault type set of an electrolytic cell and a set of basic information corresponding to each fault type in the fault type set; the fault type set includes five fault types; the five fault types are electrode corrosion, diaphragm corrosion, abnormal accumulation of gas production at the plate, uneven flow field distribution, and uneven thermal field distribution; the basic information includes current, voltage, and a current response obtained after applying a sinusoidal AC voltage signal; A supplementing unit, used to supplement the training data set to generate a supplemented data set; the supplemented data set includes a fault type set and multiple groups of basic information corresponding to each fault type in the fault type set; A clustering unit, used to perform clustering processing on the supplementary data set to establish a corresponding relationship between each fault type and a response feature; the response feature is the common feature data of multiple groups of basic information corresponding to the fault type; The diagnosis unit is used to construct a fault type and response feature matching database based on the corresponding relationship between each fault type and the response feature, and perform electrolytic cell fault diagnosis based on the fault type and response feature matching database.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images