Distributed fault diagnosis method and device for large-scale industrial closed-loop control process

Abstract

The invention relates to a distributed fault diagnosis method and device for a large-scale industrial closed-loop control process, and the method comprises the steps: obtaining historical fault-free data in the large-scale industrial closed-loop process, and dividing the historical fault-free data based on a preset division rule, so as to obtain a plurality of variable blocks; performing variable type division on the plurality of variable blocks based on the action mechanism corresponding to each variable block to obtain a classification result of the variable blocks, and constructing a corresponding distributed variable topological structure based on the classification result; constructing a corresponding dynamic hidden variable model in combination with a distributed variable topological structure and a correlation structure existing in large closed-loop process data; and constructing a corresponding distributed fault diagnosis model based on the dynamic hidden variable model, and outputting a fault diagnosis result of the to-be-detected large system by using the distributed fault diagnosis model. Therefore, the technical problems that in the related technology, the constraint of closed-loop control is not considered, the local state of the system is difficult to effectively analyze, and the distributed dynamic characteristics in the data of the large industrial system are ignored, so that accurate fault diagnosis and fault positioning results are difficult to obtain are solved.

Description

Technical Field

[0001] The present application relates to the technical field of fault diagnosis, and in particular to a distributed fault diagnosis method and device for a large-scale industrial closed-loop control process. Background Art

[0002] As industrial systems develop towards intelligence and automation, their scale is becoming larger and their mechanisms are becoming more complex. Once a system fails to complete its specified functions, it may not only cause huge economic losses, but may even endanger the lives of operators. Therefore, the demand for timely detection of faults and determination of their causes, that is, the development of fault diagnosis technology, is increasing day by day. Due to the complex mechanism of modern industrial processes, it is very difficult to establish an accurate system analysis model of industrial processes, and it is difficult to apply model-driven methods. Data-driven fault diagnosis methods can effectively utilize industrial big data and mine the correlation relationships therein, thereby establishing a diagnostic model for real-time fault diagnosis. It does not require a systematic and accurate mechanism model and is highly practical, so it has received increasing attention. However, the scale of modern industrial processes is becoming larger and larger, and the controller design and structure are becoming more and more complex, which brings new challenges to data-driven fault diagnosis methods.

[0003] Closed-loop control, that is, introducing the system's output measurement information or state information as a feedback signal to effectively control the input signal, and the corresponding control strategy in which the input signal is not affected by the feedback signal is open-loop control. Compared with open-loop control, closed-loop control has the characteristics of high precision and good robustness, and is therefore widely used in large-scale industrial processes. However, unlike the open-loop control strategy, which has little impact on fault diagnosis, closed-loop control has brought significant impacts and many challenges to fault diagnosis, which can be mainly summarized as follows:

[0004] First, closed-loop control introduces additional dynamic relationships to process data, which greatly increases the difficulty of analyzing dynamic changes in the data;

[0005] Second, the robustness of closed-loop control makes it tend to suppress the impact of faults occurring within the loop, thereby weakening the fault amplitude and making it difficult to detect;

[0006] Third, the feedback mechanism of closed-loop control makes it possible for the fault signal on one variable to be transmitted to other variables, greatly increasing the difficulty of fault location and diagnosis.

[0007] However, related technologies rarely focus on the influence of closed-loop control when designing fault diagnosis methods, resulting in large errors in fault separation and diagnosis.

[0008] At the same time, large-scale industrial systems are composed of a large number of interacting devices or subsystems. In practical applications, it is necessary to effectively monitor the overall system and its subsystems or devices at the same time. The centralized data-driven method in the related art only establishes a centralized fault diagnosis model based on the overall system data, which makes it difficult to effectively analyze the local state of the system, which makes it difficult to diagnose and locate the subsequent faults, and cannot meet the actual application needs well. In addition, the data-driven fault diagnosis technology in the related art can only model the overall dynamic characteristics of the process data, ignoring the distributed dynamic characteristics that usually exist in the data of large industrial systems, so that the dynamic model established is not accurate enough, which leads to the limitation of its fault detection accuracy and insufficient fault diagnosis ability.

[0009] To sum up, in the related technologies, the influence of closed-loop control is not taken into account when designing the fault diagnosis method, and the centralized fault diagnosis model is difficult to effectively analyze the local state of the system. The data-driven fault diagnosis technology ignores the distributed dynamic characteristics of large industrial system data, which makes the related technologies have major problems in the process of fault diagnosis and fault location, making them difficult to use effectively and need to be improved. Summary of the invention

[0010] The present application provides a distributed fault diagnosis method and device for a large-scale industrial closed-loop control process to solve the technical problems in the related technology that the constraints of closed-loop control are not taken into account, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large-scale industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results.

[0011] The first aspect of the present application provides a distributed fault diagnosis method for a large-scale industrial closed-loop control process, which is applied to the model building stage, wherein the method includes the following steps: obtaining historical fault-free data in the large-scale industrial closed-loop process, and dividing the historical fault-free data based on a preset division rule to obtain a plurality of variable blocks; dividing the plurality of variable blocks into variable types based on the action mechanism corresponding to each variable block to obtain a classification result of the variable blocks, and constructing a corresponding distributed variable topology structure based on the classification result; constructing a corresponding dynamic latent variable model by combining the distributed variable topology structure and the correlation structure existing in the large-scale closed-loop process data; constructing a corresponding distributed fault diagnosis model based on the dynamic latent variable model, so as to use the distributed fault diagnosis model to output the fault diagnosis result of the large-scale system to be detected.

[0012] Optionally, in one embodiment of the present application, the expression of the dynamic latent variable model is:

[0013]

[0014]

[0015]

[0016]

[0017] in, represents the dimension-reduced dynamic latent variable derived from the variable block b and its correlation structure, e b represents the weight matrix, Indicates that the variables contained in the variable block b correspond to m b dimensional column vector, P b represents the load matrix, Indicates the static data changes contained in the data variable block b, G ab The corresponding elements of the topological matrix corresponding to the topological graph are represented. It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

[0018] Optionally, in one embodiment of the present application, constructing a corresponding distributed fault diagnosis model based on the dynamic latent variable model includes: constructing an optimization problem of the dynamic latent variable model using a preset optimization principle; solving the optimization problem to obtain a set of weighted matrices to be optimized and dynamic model parameters of the dynamic latent variable model, so as to construct the distributed fault diagnosis model using the set of weighted matrices to be optimized and dynamic model parameters.

[0019] Optionally, in one embodiment of the present application, constructing a corresponding distributed fault diagnosis model based on the dynamic latent variable model includes: using the dynamic latent variable model to construct two variables without time series correlation for each variable block to obtain the dynamic and static relationship of the corresponding variable block; constructing the detection index and index threshold of each variable block based on the dynamic and static relationship of each variable block and the principal component analysis method to obtain the index set and index threshold set corresponding to the multiple variable blocks; constructing a mapping relationship between the index set and the index threshold set and the fault diagnosis result, and constructing the distributed fault diagnosis model based on the mapping relationship.

[0020] Optionally, in one embodiment of the present application,

[0021] The expressions of the dynamic relationship index and the dynamic threshold in the detection index and the index threshold are:

[0022]

[0023]

[0024] Among them, J v represents the dynamic monitoring indicator function, represents the current dynamic residual, Represents a dynamic index, P v express v k The principal component space loadings, V represents the dynamic residual data matrix obtained from the historical fault-free data, express v k The residual subspace loadings of represents a statistical distribution derived parameter, J v,th Indicates the threshold value corresponding to the dynamic monitoring indicator. represents the alpha quantile of the chi-square distribution, l v Denotes another statistical distribution derived parameter, h v represents yet another statistical distribution derived parameter;

[0025] The expressions of the static relationship index and the static threshold in the detection index and the index threshold are:

[0026]

[0027]

[0028] Among them, J e represents the static monitoring indicator function, represents the dynamic residual obtained from static measurement, Represents a static indicator, P e express e k The principal component space loads, E represents the static residual data matrix obtained from the historical fault-free data, express e k The residual subspace loadings of represents a statistical distribution derived parameter, J e,th Indicates the threshold value corresponding to the static monitoring indicator, l e Denotes another statistical distribution derived parameter, h e represents yet another statistical distribution derived parameter.

[0029] The second aspect of the present application provides a distributed fault diagnosis method for a large-scale industrial closed-loop control process, which is applied to the model usage stage, wherein the method includes the following steps: obtaining current data in the closed-loop control process of the large-scale system to be detected; inputting the current data into a pre-built distributed fault diagnosis model to obtain the current fault diagnosis result of the large-scale system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in the large-scale closed-loop control process.

[0030] The third aspect of the present application provides a distributed fault diagnosis device for a large-scale industrial closed-loop control process, which is applied to the model construction stage, wherein the device includes: an acquisition module, which is used to acquire historical fault-free data in the large-scale industrial closed-loop process, and divide the historical fault-free data based on a preset division rule to obtain a plurality of variable blocks; a classification module, which is used to divide the plurality of variable blocks into variable types based on the action mechanism corresponding to each variable block, so as to obtain the classification results of the variable blocks, and construct a corresponding distributed variable topological structure based on the classification results; a first construction module, which is used to combine the distributed variable topological structure and the correlation structure existing in the large-scale closed-loop process data to construct a corresponding dynamic latent variable model; a second construction module, which is used to construct a corresponding distributed fault diagnosis model based on the dynamic latent variable model, so as to use the distributed fault diagnosis model to output the fault diagnosis results of the large-scale system to be detected.

[0031] Optionally, in one embodiment of the present application, the expression of the dynamic latent variable model is:

[0032]

[0033]

[0034]

[0035]

[0036] in, represents the dimension-reduced dynamic latent variable derived from the variable block b and its correlation structure, R b represents the weight matrix, Indicates that the variables contained in the variable block b correspond to m b dimensional column vector, P b represents the load matrix, Indicates the static data changes contained in the data variable block b, G ab The corresponding elements of the topological matrix corresponding to the topological graph are represented. It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

[0037] Optionally, in one embodiment of the present application, the second construction module includes: a first construction unit, used to construct the optimization problem of the dynamic latent variable model using a preset optimization principle; a calculation unit, used to solve the optimization problem and obtain the weighted matrix to be optimized and a set of dynamic model parameters of the dynamic latent variable model, so as to construct the distributed fault diagnosis model using the weighted matrix to be optimized and the set of dynamic model parameters.

[0038] Optionally, in one embodiment of the present application, the second construction module includes: a first construction unit, used to use the dynamic latent variable model to construct two variables without time series correlation for each variable block, so as to obtain the dynamic and static relationship of the corresponding variable block; a second construction unit, used to construct the detection index and index threshold of each variable block based on the dynamic and static relationship of each variable block and the principal component analysis method, so as to obtain the index set and index threshold set corresponding to the multiple variable blocks; a second construction unit, used to construct a mapping relationship between the index set and the index threshold set and the fault diagnosis result, and construct the distributed fault diagnosis model based on the mapping relationship.

[0039] Optionally, in one embodiment of the present application,

[0040] The expressions of the dynamic relationship index and the dynamic threshold in the detection index and the index threshold are:

[0041]

[0042]

[0043] Among them, J v represents the dynamic monitoring indicator function, represents the current dynamic residual, Represents a dynamic index, P v express v k The principal component space loadings, V represents the dynamic residual data matrix obtained from the historical fault-free data, express v k The residual subspace loadings of represents a statistical distribution derived parameter, J v,th Indicates the threshold value corresponding to the dynamic monitoring indicator. represents the alpha quantile of the chi-square distribution, l v Denotes another statistical distribution derived parameter, h v represents yet another statistical distribution derived parameter;

[0044] The expressions of the static relationship index and the static threshold in the detection index and the index threshold are:

[0045]

[0046]

[0047] Among them, J e represents the static monitoring indicator function, represents the dynamic residual obtained from static measurement, Represents a static indicator, P e express e k The principal component space loads, E represents the static residual data matrix obtained from the historical fault-free data, express e k The residual subspace loadings of represents a statistical distribution derived parameter, J e,th Indicates the threshold value corresponding to the static monitoring indicator, l e Denotes another statistical distribution derived parameter, h e represents yet another statistical distribution derived parameter.

[0048] The fourth aspect of the present application provides a distributed fault diagnosis device for a large-scale industrial closed-loop control process, which is applied to the model usage stage, wherein the device includes: an acquisition module, used to obtain current data in the closed-loop control process of the large-scale system to be detected; a diagnosis module, used to input the current data into a pre-built distributed fault diagnosis model to obtain the current fault diagnosis result of the large-scale system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in the large-scale closed-loop control process.

[0049] The fifth aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the distributed fault diagnosis method for a large industrial closed-loop control process as described in the above embodiment.

[0050] The sixth aspect of the present application provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the distributed fault diagnosis method for a large-scale industrial closed-loop control process as described in the above embodiment.

[0051] The seventh aspect of the present application provides a computer program product, including a computer program, which, when executed, is used to implement the above-mentioned distributed fault diagnosis method for a large-scale industrial closed-loop control process.

[0052] The embodiment of the present application can obtain the historical fault-free data in the large-scale industrial closed-loop process, and divide the historical fault-free data to obtain multiple variable blocks, and construct a corresponding directed topological structure diagram according to the classification results of the variable blocks, and then construct a distributed variable topological structure, combine the distributed variable topological structure and the correlation structure existing in the large-scale closed-loop process data, and construct a corresponding dynamic hidden variable model, and finally use the dynamic hidden variable model to construct a corresponding distributed fault diagnosis model, so as to use the distributed fault diagnosis model to output the fault diagnosis result of the large-scale system to be detected, that is, by establishing a distributed dynamic model considering closed-loop control, effectively modeling the dynamic relationship of distributed data under closed-loop control, and by designing the indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large-scale closed-loop industrial process are effectively concurrently detected, so as to comprehensively analyze the information of each indicator and give the overall and local fault diagnosis results of the system. Thus, the technical problem that the constraints of closed-loop control are not considered in the related art, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large-scale industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results is solved.

[0053] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0055] Figure 1 A flow chart of a distributed fault diagnosis method for a large-scale industrial closed-loop control process provided according to an embodiment of the present application;

[0056] Figure 2 A schematic diagram of a distributed variable topology structure according to an embodiment of the present application;

[0057] Figure 3 A fault diagnosis logic diagram according to an embodiment of the present application;

[0058] Figure 4 It is a schematic diagram of the principle of a distributed fault diagnosis method for a large-scale industrial closed-loop control process according to an embodiment of the present application;

[0059] Figure 5 A schematic diagram of the structure of a distributed fault diagnosis device for a large-scale industrial closed-loop control process provided according to an embodiment of the present application;

[0060] Figure 6 A flowchart of another distributed fault diagnosis method for a large-scale industrial closed-loop control process provided according to an embodiment of the present application;

[0061] Figure 7 A schematic diagram of the structure of another distributed fault diagnosis device for a large-scale industrial closed-loop control process provided according to an embodiment of the present application;

[0062] Figure 8 It is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0063] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0064] The following describes a distributed fault diagnosis method and apparatus for a large-scale industrial closed-loop control process according to an embodiment of the present application with reference to the accompanying drawings. In view of the technical problem that the related technologies mentioned in the above background technology do not consider the constraints of closed-loop control, it is difficult to effectively analyze the local state of the system, and ignore the distributed dynamic characteristics in the data of large industrial systems, making it difficult to obtain accurate fault diagnosis and fault location results, the present application provides a distributed fault diagnosis method for a large industrial closed-loop control process. In this method, historical fault-free data in a large industrial closed-loop process can be obtained, and the historical fault-free data can be divided to obtain multiple variable blocks, and a corresponding directed topological structure graph is constructed according to the classification results of the variable blocks, and then a distributed variable topological structure is constructed. Combining the distributed variable topological structure and the correlation structure existing in the large closed-loop process data, a corresponding dynamic hidden variable model is constructed, and finally a corresponding distributed fault diagnosis model is constructed using the dynamic hidden variable model, so as to use the distributed fault diagnosis model to output the fault diagnosis results of the large system to be detected, that is, by establishing a distributed dynamic model considering closed-loop control, the dynamic relationship of distributed data under closed-loop control is effectively modeled, and by designing an indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large closed-loop industrial process are effectively concurrently detected, so as to comprehensively integrate the information of each indicator and give the overall and local fault diagnosis results of the system. This solves the technical problems in related technologies, that is, failure to consider the constraints of closed-loop control, difficulty in effectively analyzing the local state of the system, and neglect of the distributed dynamic characteristics in large industrial system data, making it difficult to obtain accurate fault diagnosis and fault location results.

[0065] Specifically, Figure 1 A flow chart of a distributed fault diagnosis method for a large-scale industrial closed-loop control process provided in an embodiment of the present application.

[0066] like Figure 1As shown, the distributed fault diagnosis method for the large-scale industrial closed-loop control process is applied to the model building stage, wherein the method includes the following steps:

[0067] In step S101, historical fault-free data in a large-scale industrial closed-loop process is obtained, and the historical fault-free data is divided based on a preset division rule to obtain a plurality of variable blocks.

[0068] During the actual implementation process, the embodiments of the present application can collect available data variables, that is, historical fault-free data in large-scale industrial closed-loop processes, and decompose them into several variable blocks based on the mechanism knowledge and structural information of large-scale industrial closed-loop processes. Each variable block corresponds to an important equipment or subsystem in the large-scale industrial closed-loop process, so that each variable block itself can reflect the operating status of key equipment and subsystems, and after the data of each variable block are integrated, it can organically reflect the overall status of the system.

[0069] In step S102, multiple variable blocks are classified into variable types based on the action mechanism corresponding to each variable block to obtain classification results of the variable blocks, and a corresponding distributed variable topology structure is constructed based on the classification results.

[0070] As a possible implementation method, the embodiment of the present application can divide the variables in each variable block into three categories according to their action mechanism. The first category is the input variables that are directly regulated by the specific controller, which are called controlled variables; the second category is the variables that are the local output of a subsystem or device and affect other subsystems or devices, which are called local process variables; the third category is the variables that only reflect the output information or correspond to the overall system output, which are called output variables.

[0071] Furthermore, the embodiments of the present application can construct a directed topological structure graph of the corresponding variable blocks based on the division and classification results of the variable blocks to reflect the causal dynamic relationship between the variables within each variable block and the variables between the variable blocks, and obtain the corresponding distributed variable topological structure based on the directed topological structure graph.

[0072] like Figure 2 The figure shows an example of a directed topology diagram. Specifically, input variables may affect local process variables and output variables; local process variables may affect other local process variables, or may be used as reference signals to adjust input variables; output variables may be used as reference signals to adjust input variables. The establishment of the above model is conducive to the establishment of subsequent data-driven process models and fault diagnosis models.

[0073] In step S103, a corresponding dynamic latent variable model is constructed by combining the distributed variable topology structure and the correlation structure existing in the large closed-loop process data. The expression of the dynamic latent variable model is:

[0074]

[0075]

[0076]

[0077]

[0078] in, represents the dimension-reduced dynamic latent variable derived from the variable block b and its correlation structure, R b represents the weight matrix, Indicates that the variables contained in the variable block b correspond to m b dimensional column vector, P b represents the load matrix, Indicates the static data changes contained in the data variable block b, G ab The corresponding elements of the topological matrix corresponding to the topological graph are represented. It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

[0079] According to the distributed variable topology and the correlation structure existing in the large closed-loop process data, the embodiment of the present application can construct the following dynamic latent variable model. Assuming that the data can be divided into B data blocks, the distributed dynamic latent variable model corresponding to the bth data block can be expressed as shown in the following formula:

[0080]

[0081]

[0082]

[0083]

[0084] in, The variable block b and its correlation structure derive the dimension-reduced dynamic latent variable, which can be expressed as b dimensional column vector, the dynamic changes in the variable block can be considered to be fully represented by this reduced-dimensional column vector; Represents the weight matrix, which describes the mapping relationship between variables in data block b and latent variables; Indicates that the variables contained in the variable block b correspond to m b dimensional column vector; P b Represents the loading matrix, which describes the mapping relationship from latent variables to the original variables in the data variable block b; Indicates the static data changes contained in the data variable block b; G ab The corresponding element of the topological matrix corresponding to the topological graph is represented as 1 when there is a directed edge from variable block a to variable block b in the topological graph constructed in the above step, otherwise it is 0; It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

[0085] In step S104, a corresponding distributed fault diagnosis model is constructed based on the dynamic latent variable model, so as to output the fault diagnosis result of the large-scale system to be detected by using the distributed fault diagnosis model.

[0086] In the actual implementation process, the embodiments of the present application can use the dynamic latent variable model to build a corresponding distributed fault diagnosis model, thereby realizing fault diagnosis of large systems.

[0087] Optionally, in one embodiment of the present application, a corresponding distributed fault diagnosis model is constructed based on a dynamic latent variable model, including: constructing an optimization problem of the dynamic latent variable model using a preset optimization principle; solving the optimization problem to obtain a weighted matrix to be optimized and a set of dynamic model parameters of the dynamic latent variable model, so as to construct a distributed fault diagnosis model using the weighted matrix to be optimized and a set of dynamic model parameters.

[0088] After the above-mentioned distributed dynamic latent variable model is constructed, the embodiment of the present application can extract latent variables according to the following steps.

[0089] First, according to the principle of optimizing the prediction performance of the dynamic model, the embodiment of the present application can construct the following optimization problem:

[0090]

[0091]

[0092] in, and They correspond to the set of weighted matrices to be optimized and dynamic model parameters respectively, and the specific forms are as follows:

[0093]

[0094]

[0095]

[0096]

[0097] In addition, the optimization objective in the expression of the optimization problem The corresponding distributed dynamic latent variable model predicts this variable block, so the essence of this optimization problem is to make the distributed dynamic latent variable model predict the best performance. The prediction value can be calculated as follows:

[0098]

[0099]

[0100]

[0101] The optimization problem can be solved iteratively as follows.

[0102] For each variable block b, initialize the intermediate weight matrix variable W b is one-dimensional and R b And make it satisfy (W b ) T W b is a unit matrix.

[0103] For each variable block b, collect its historical data Where 1≤k≤N+s. The data matrix is ​​constructed as follows And perform the following singular value decomposition:

[0104]

[0105]

[0106] Then, the embodiment of the present application may iterate the following two steps until all W b convergence.

[0107] Update the latent variable matrix block by block as follows

[0108]

[0109] Then construct the latent variable expansion matrix corresponding to each data block as follows:

[0110]

[0111]

[0112] Then, the singular value decomposition of a matrix derived from the latent variable expansion matrix corresponding to each data block is performed as follows:

[0113]

[0114] Update W one by oneb for The front l b List.

[0115] The set of model parameters to be solved and It can be solved as follows:

[0116]

[0117]

[0118]

[0119] Optionally, in one embodiment of the present application, a corresponding distributed fault diagnosis model is constructed based on a dynamic latent variable model, including: using the dynamic latent variable model to construct two variables without time series correlation for each variable block to obtain the dynamic and static relationship of the corresponding variable block; constructing the detection index and index threshold of each variable block based on the dynamic and static relationship of each variable block and the principal component analysis method to obtain the index set and index threshold set corresponding to multiple variable blocks; constructing a mapping relationship between the index set and the index threshold set and the fault diagnosis result, and constructing a distributed fault diagnosis model based on the mapping relationship.

[0120] Optionally, in one embodiment of the present application,

[0121] The expressions of dynamic relationship indicators and dynamic thresholds in detection indicators and indicator thresholds are:

[0122]

[0123]

[0124] Among them, J v represents the dynamic monitoring indicator function, represents the current dynamic residual, Represents a dynamic index, P v Indicates v k The principal component space loads, V represents the dynamic residual data matrix obtained from the historical fault-free data, Indicates v k The residual subspace loadings of represents a statistical distribution derived parameter, J v,th Indicates the threshold value corresponding to the dynamic monitoring indicator. represents the alpha quantile of the chi-square distribution, l v Denotes another statistical distribution derived parameter, h v represents yet another statistical distribution derived parameter;

[0125] The expressions of the static relationship index and static threshold in the detection index and index threshold are:

[0126]

[0127]

[0128] Among them, J e represents the static monitoring indicator function, represents the dynamic residual obtained from static measurement, Represents a static indicator, P e Indicates e k The principal component space loads, E represents the static residual data matrix obtained from the historical fault-free data, Indicates e k The residual subspace loadings of represents a statistical distribution derived parameter, J e,th Indicates the threshold value corresponding to the static monitoring indicator, l e Denotes another statistical distribution derived parameter, h e represents yet another statistical distribution derived parameter.

[0129] Furthermore, the embodiment of the present application can construct the following two variables without time series correlation for each variable block b according to the distributed latent variable model to monitor the dynamic and static relationship therein:

[0130]

[0131]

[0132] in, Monitor the dynamic relationship in variable block b, Monitor the static changes in variable block b. Next, the principal component analysis method can be used to construct detection indicators. For example, for The following principal component decomposition can be performed:

[0133]

[0134] in, The data matrix composed of fault-free residuals obtained by introducing the distributed dynamic latent variable model expression and two variables without time series correlation to monitor the dynamic and static relationship expression of the historical fault-free data is obtained. is the corresponding principal component loading matrix and is the corresponding residual component loading matrix.

[0135] From this, the following index can be constructed To monitor And given the confidence α, the threshold can be designed as follows:

[0136]

[0137]

[0138] in, represents the alpha quantile of the chi-square distribution with n degrees of freedom, for The number of columns, parameters and It can be calculated as follows:

[0139]

[0140]

[0141]

[0142] Similarly, for monitoring static changes Monitoring indicators and their thresholds can also be constructed in a similar way, namely and The specific expression can be given as follows:

[0143]

[0144]

[0145] in,

[0146]

[0147]

[0148]

[0149] In addition, for the overall dynamic and static relationship of a large closed-loop control process, it is necessary to integrate the information of the distributed latent variable model to give a monitoring method. The following two variables are constructed to effectively monitor the overall dynamic and static relationship of a large process:

[0150]

[0151] in,

[0152]

[0153]

[0154]

[0155] The v in the previous text k Dynamic relationships in data can be monitored, ek It can be used to monitor static relationships. Accordingly, the following two sets of indicators can be designed: and And the corresponding threshold J v,th , J e,th .

[0156] For v k , the indicators and thresholds can be designed as follows:

[0157]

[0158]

[0159] in,

[0160]

[0161]

[0162]

[0163] For e k , the indicators and their thresholds can be designed as follows:

[0164]

[0165]

[0166] in,

[0167]

[0168]

[0169]

[0170] So there are two sets J and J th , which represent the indicator set and indicator threshold set respectively.

[0171]

[0172]

[0173] Finally, the embodiment of the present application can design a mapping relationship from the results given by the indicator set to the fault diagnosis results, that is, the fault diagnosis logic, which can be briefly described as the following two steps:

[0174] 1. If any indicator in the indicator set alarms, it is considered that an abnormality has occurred. Specifically, if J v or J e If the threshold is exceeded, it is considered that the global dynamic data change of the large system is abnormal; if Jv or J e If the threshold is exceeded, it is considered that the subsystem or device corresponding to data block b is abnormal.

[0175] 2. If an abnormality occurs, analyze the alarm conditions of each indicator. If the indicator corresponding to the large system alarms, it is considered that this abnormality is a fault that affects the normal operation of the closed-loop process. Otherwise, analyze the variables in the subsystems or devices corresponding to all abnormal indicators. If the third type of variables, that is, output variables, are not included, it can be considered that the abnormality is only a local disturbance; otherwise, it is considered that the abnormality is a fault that affects the system output.

[0176] Among them, the fault diagnosis logic diagram can be as follows Figure 3 shown.

[0177] Combination Figure 3 and Figure 4 As shown, the working principle of the distributed fault diagnosis method for a large-scale industrial closed-loop control process of the embodiment of the present application is described in detail by taking an embodiment as an example.

[0178] like Figure 4 As shown, an embodiment of the present application may include two stages: an offline stage for establishing a fault diagnosis model and an online stage for implementing fault diagnosis. The former corresponds to several steps for obtaining a distributed fault diagnosis model from historical data, and the latter corresponds to several steps for obtaining the fault status of the current process using online process data and the fault diagnosis model.

[0179] Among them, the offline stage of fault diagnosis model establishment mainly includes the following steps, namely, historical fault-free data collection, data variable block division, establishment of distributed dynamic latent variable model, and establishment of distributed fault diagnosis model, which can be explained as follows.

[0180] Historical trouble-free data collection, collect historical process data of large closed-loop control processes for a continuous period of time and combine them into a data matrix Each row corresponds to the transpose of a column of vectors consisting of all the measurement information of the elephant closed-loop process at a moment, and is arranged in rows in chronological order.

[0181] Data variable block division, that is, according to the variable division result of step S101 of the embodiment of the present application, the matrix Divide into B data arrays in, For x k The column vector corresponding to the b-th subsystem or device in is transposed.

[0182] Establish a distributed dynamic latent variable model, that is, according to step 2 to step 5 of the invention, solve the model parameter set in the distributed dynamic latent variable model, that is, and And build the corresponding dynamic latent variable model.

[0183] Establish a distributed fault diagnosis model, that is, according to step S103 of the invention, construct a fault diagnosis index function set J and its corresponding threshold set J th .

[0184] The online implementation stage of fault diagnosis mainly includes the following steps, namely, online distributed latent variable solution, distributed fault diagnosis indicator solution and threshold analysis, and giving fault diagnosis results based on fault diagnosis logic, which can be given as follows.

[0185] Online distributed latent variable solution, that is, based on online data And according to the variable division principle in step S101, we can obtain Then, according to the distributed dynamic latent variable model, we solve That is, the real-time value of the distributed latent variable.

[0186] Distributed fault diagnosis index solution and threshold analysis, that is, according to the previous step Then will Substitute two variables without time-series correlation to monitor the dynamic and static relationship between them. k Solution and Substitute x into the two variable expressions for effectively monitoring the overall dynamic and static relationship of the large process k Solution and Then, the above variables are substituted into the expressions for monitoring static changes and monitoring dynamic changes to solve the real-time values ​​of each indicator in the indicator set J, and then compared with J th Compare with the comparison threshold in to obtain the real-time alarm indicator set

[0187] The fault diagnosis results are given based on the fault diagnosis logic, that is, according to the real-time alarm indicator set and Figure 3 Diagnostic logic to obtain real-time fault diagnosis results.

[0188] According to the distributed fault diagnosis method of a large industrial closed-loop control process proposed in the embodiment of the present application, the historical fault-free data in the large industrial closed-loop process can be obtained, and the historical fault-free data can be divided to obtain multiple variable blocks, and a corresponding directed topological structure diagram can be constructed according to the classification results of the variable blocks, and then a distributed variable topological structure can be constructed, and a corresponding dynamic hidden variable model can be constructed by combining the distributed variable topological structure and the correlation structure existing in the large closed-loop process data, and finally a corresponding distributed fault diagnosis model can be constructed by using the dynamic hidden variable model, so as to output the fault diagnosis result of the large system to be detected by using the distributed fault diagnosis model, that is, by establishing a distributed dynamic model considering closed-loop control, the dynamic relationship of distributed data under closed-loop control can be effectively modeled, and by designing an indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large closed-loop industrial process can be effectively concurrently detected, so as to comprehensively analyze the information of each indicator and give the overall and local fault diagnosis results of the system. Thus, the technical problem that the constraints of closed-loop control are not considered in the related art, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results is solved.

[0189] Next, a distributed fault diagnosis device for a large-scale industrial closed-loop control process proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0190] Figure 5 It is a block diagram of a distributed fault diagnosis device for a large-scale industrial closed-loop control process according to an embodiment of the present application.

[0191] like Figure 5 As shown, the distributed fault diagnosis device 10 for the large-scale industrial closed-loop control process is applied to the model building stage, wherein the device 10 includes: an acquisition module 101, a classification module 102, a first building module 103 and a second building module 104.

[0192] Specifically, the acquisition module 101 is used to acquire historical fault-free data in a large-scale industrial closed-loop process, and divide the historical fault-free data based on a preset division rule to obtain a plurality of variable blocks.

[0193] The classification module 102 is used to classify the multiple variable blocks into variable types based on the action mechanism corresponding to each variable block to obtain the classification results of the variable blocks, and to construct a corresponding distributed variable topology structure based on the classification results.

[0194] The first construction module 103 is used to construct a corresponding dynamic latent variable model by combining the distributed variable topology structure and the correlation structure existing in the large closed-loop process data.

[0195] The second construction module 104 is used to construct a corresponding distributed fault diagnosis model based on the dynamic latent variable model, so as to output the fault diagnosis result of the large-scale system to be detected by using the distributed fault diagnosis model.

[0196] Optionally, in one embodiment of the present application, the expression of the dynamic latent variable model is:

[0197]

[0198]

[0199]

[0200]

[0201] in, represents the dimension-reduced dynamic latent variable derived from the variable block b and its correlation structure, R b represents the weight matrix, Indicates that the variables contained in the variable block b correspond to m b dimensional column vector, P b represents the load matrix, Indicates the static data changes contained in the data variable block b, G ab The corresponding elements of the topological matrix corresponding to the topological graph are represented. It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

[0202] Optionally, in one embodiment of the present application, the second building module 104 includes: a first building unit and a computing unit.

[0203] Among them, the first construction unit is used to construct the optimization problem of the dynamic latent variable model using a preset optimization principle.

[0204] The computing unit is used to solve the optimization problem and obtain the weighted matrix to be optimized and the set of dynamic model parameters of the dynamic latent variable model, so as to construct a distributed fault diagnosis model using the weighted matrix to be optimized and the set of dynamic model parameters.

[0205] Optionally, in one embodiment of the present application, the second construction module 104 includes: a first construction unit, a second construction unit and a second construction unit.

[0206] The first construction unit is used to construct two variables without time series correlation for each variable block by using a dynamic latent variable model, so as to obtain a dynamic and static relationship between the corresponding variable blocks.

[0207] The second construction unit is used to construct the detection index and index threshold of each variable block based on the dynamic and static relationship of each variable block and the principal component analysis method, so as to obtain the index set and index threshold set corresponding to multiple variable blocks.

[0208] The second construction unit is used to construct a mapping relationship between the indicator set and the indicator threshold set and the fault diagnosis result, and to construct a distributed fault diagnosis model based on the mapping relationship.

[0209] Optionally, in one embodiment of the present application,

[0210] The expressions of dynamic relationship indicators and dynamic thresholds in detection indicators and indicator thresholds are:

[0211]

[0212]

[0213] Among them, J v represents the dynamic monitoring indicator function, represents the current dynamic residual, Represents a dynamic index, P v express v k The principal component space loads, V represents the dynamic residual data matrix obtained from the historical fault-free data, express v k The residual subspace loadings of represents a statistical distribution derived parameter, J v,th Indicates the threshold value corresponding to the dynamic monitoring indicator. represents the alpha quantile of the chi-square distribution, l v Denotes another statistical distribution derived parameter, h v represents yet another statistical distribution derived parameter.

[0214] The expressions of the static relationship index and static threshold in the detection index and index threshold are:

[0215]

[0216]

[0217] Among them, J e represents the static monitoring indicator function, represents the dynamic residual obtained from static measurement, Represents a static indicator, P e Indicates e k The principal component space loads, E represents the static residual data matrix obtained from the historical fault-free data, Indicates e k The residual subspace loadings of represents a statistical distribution derived parameter, J e,th Indicates the threshold value corresponding to the static monitoring indicator, l e Denotes another statistical distribution derived parameter, h e represents yet another statistical distribution derived parameter.

[0218] It should be noted that the aforementioned explanation of the embodiment of the distributed fault diagnosis method for a large-scale industrial closed-loop control process is also applicable to the distributed fault diagnosis device for a large-scale industrial closed-loop control process of this embodiment, and will not be repeated here.

[0219] According to the distributed fault diagnosis device of the large-scale industrial closed-loop control process proposed in the embodiment of the present application, the historical fault-free data in the large-scale industrial closed-loop process can be obtained, and the historical fault-free data can be divided to obtain multiple variable blocks, and the corresponding directed topological structure diagram can be constructed according to the classification results of the variable blocks, and then the distributed variable topological structure can be constructed, and the corresponding dynamic hidden variable model can be constructed by combining the distributed variable topological structure and the correlation structure existing in the large-scale closed-loop process data, and finally the corresponding distributed fault diagnosis model can be constructed by using the dynamic hidden variable model, so as to output the fault diagnosis result of the large-scale system to be detected by using the distributed fault diagnosis model, that is, by establishing a distributed dynamic model considering closed-loop control, the dynamic relationship of distributed data under closed-loop control can be effectively modeled, and by designing the indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large-scale closed-loop industrial process can be effectively concurrently detected, so as to comprehensively analyze the information of each indicator and give the overall and local fault diagnosis results of the system. Thus, the technical problem that the constraints of closed-loop control are not considered in the related art, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large-scale industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results is solved.

[0220] The above is an explanation of the distributed fault diagnosis method for a large-scale industrial closed-loop control process in the model building phase of the embodiment of the present application. The following is an explanation of the model use phase.

[0221] Specifically, Figure 6 A flow chart of a distributed fault diagnosis method for a large-scale industrial closed-loop control process provided in an embodiment of the present application.

[0222] like Figure 6 As shown, the distributed fault diagnosis method for the large-scale industrial closed-loop control process is applied to the model building stage, wherein the method includes the following steps:

[0223] In step S601, current data in the closed-loop control process of the large-scale system to be detected is obtained.

[0224] In step S602, current data is input into a pre-built distributed fault diagnosis model to obtain current fault diagnosis results of the large system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in a large closed-loop control process.

[0225] According to the distributed fault diagnosis method of a large industrial closed-loop control process proposed in the embodiment of the present application, the historical fault-free data in the large industrial closed-loop process can be obtained, and the historical fault-free data can be divided to obtain multiple variable blocks, and a corresponding directed topological structure diagram can be constructed according to the classification results of the variable blocks, and then a distributed variable topological structure can be constructed, and a corresponding dynamic hidden variable model can be constructed by combining the distributed variable topological structure and the correlation structure existing in the large closed-loop process data, and finally a corresponding distributed fault diagnosis model can be constructed by using the dynamic hidden variable model, so as to output the fault diagnosis result of the large system to be detected by using the distributed fault diagnosis model, that is, by establishing a distributed dynamic model considering closed-loop control, the dynamic relationship of distributed data under closed-loop control can be effectively modeled, and by designing an indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large closed-loop industrial process can be effectively concurrently detected, so as to comprehensively analyze the information of each indicator and give the overall and local fault diagnosis results of the system. Thus, the technical problem that the constraints of closed-loop control are not considered in the related art, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results is solved.

[0226] Next, a distributed fault diagnosis device for a large-scale industrial closed-loop control process proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0227] Figure 7 It is a block diagram of a distributed fault diagnosis device for a large-scale industrial closed-loop control process according to an embodiment of the present application.

[0228] like Figure 7 As shown, the distributed fault diagnosis device 20 of the large-scale industrial closed-loop control process is applied to the model building stage, wherein the device 20 includes: an acquisition module 201 and a diagnosis module 202 .

[0229] Specifically, the acquisition module 201 is used to acquire current data in the closed-loop control process of the large-scale system to be detected.

[0230] The diagnosis module 202 is used to input current data into a pre-built distributed fault diagnosis model to obtain the current fault diagnosis result of the large system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in a large closed-loop control process.

[0231] It should be noted that the aforementioned explanation of the embodiment of the distributed fault diagnosis method for a large-scale industrial closed-loop control process is also applicable to the distributed fault diagnosis device for a large-scale industrial closed-loop control process of this embodiment, and will not be repeated here.

[0232] According to the distributed fault diagnosis device of the large-scale industrial closed-loop control process proposed in the embodiment of the present application, the historical fault-free data in the large-scale industrial closed-loop process can be obtained, and the historical fault-free data can be divided to obtain multiple variable blocks, and the corresponding directed topological structure diagram can be constructed according to the classification results of the variable blocks, and then the distributed variable topological structure can be constructed, and the corresponding dynamic hidden variable model can be constructed by combining the distributed variable topological structure and the correlation structure existing in the large-scale closed-loop process data, and finally the corresponding distributed fault diagnosis model can be constructed by using the dynamic hidden variable model, so as to output the fault diagnosis result of the large-scale system to be detected by using the distributed fault diagnosis model, that is, by establishing a distributed dynamic model considering closed-loop control, the dynamic relationship of distributed data under closed-loop control can be effectively modeled, and by designing the indicator set derived from the distributed dynamic model, the overall and local dynamic and static relationships of the large-scale closed-loop industrial process can be effectively concurrently detected, so as to comprehensively analyze the information of each indicator and give the overall and local fault diagnosis results of the system. Thus, the technical problem that the constraints of closed-loop control are not considered in the related art, it is difficult to effectively analyze the local state of the system, and the distributed dynamic characteristics in the data of large-scale industrial systems are ignored, making it difficult to obtain accurate fault diagnosis and fault location results is solved.

[0233] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0234] A memory 801 , a processor 802 , and a computer program stored in the memory 801 and executable on the processor 802 .

[0235] When the processor 802 executes the program, the distributed fault diagnosis method for a large industrial closed-loop control process provided in the above embodiment is implemented.

[0236] Furthermore, the electronic device further comprises:

[0237] The communication interface 803 is used for communication between the memory 801 and the processor 802 .

[0238] The memory 801 is used to store computer programs that can be executed on the processor 802 .

[0239] The memory 801 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0240] If the memory 801, the processor 802 and the communication interface 803 are implemented independently, the communication interface 803, the memory 801 and the processor 802 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0241] Optionally, in a specific implementation, if the memory 801, the processor 802 and the communication interface 803 are integrated on a chip, the memory 801, the processor 802 and the communication interface 803 can communicate with each other through an internal interface.

[0242] The processor 802 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0243] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the distributed fault diagnosis method for a large-scale industrial closed-loop control process as described above is implemented.

[0244] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the distributed fault diagnosis method for a large industrial closed-loop control process provided by an embodiment of the present invention.

[0245] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0246] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0247] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0248] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0249] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0250] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0251] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0252] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A distributed fault diagnosis method for a large industrial closed-loop control process, characterized in that: Applied to the model building stage, wherein the method comprises the following steps: Acquire historical fault-free data in a large industrial closed-loop process, and divide the historical fault-free data based on a preset division rule to obtain a plurality of variable blocks; Classifying the multiple variable blocks into variable types based on the action mechanism corresponding to each variable block to obtain classification results of the variable blocks, and constructing a corresponding distributed variable topology structure based on the classification results; Combining the distributed variable topology structure with the correlation structure existing in the large closed-loop process data, a corresponding dynamic latent variable model is constructed; A corresponding distributed fault diagnosis model is constructed based on the dynamic latent variable model, so as to utilize the distributed fault diagnosis model to output the fault diagnosis result of the large-scale system to be detected.

2. The method according to claim 1, characterized in that The expression of the dynamic latent variable model is: in, represents the dimension-reduced dynamic latent variable derived from the variable block b and its correlation structure, R b represents the weight matrix, Indicates that the variables contained in the variable block b correspond to m b dimensional column vector, P b represents the load matrix, Indicates the static data changes contained in the data variable block b, G ab The corresponding elements of the topological matrix corresponding to the topological graph are represented. It represents the prediction of the latent variables corresponding to variable block b based on the variables in variable block a. Indicates the dynamic data changes contained in variable block b, Represents dynamic model parameters.

3. The method according to claim 1, characterized in that The corresponding distributed fault diagnosis model is constructed based on the dynamic latent variable model, including: Constructing the optimization problem of the dynamic latent variable model using a preset optimization principle; The optimization problem is solved to obtain the weighted matrix to be optimized and a set of dynamic model parameters of the dynamic latent variable model, so as to construct the distributed fault diagnosis model using the weighted matrix to be optimized and the set of dynamic model parameters.

4. The method according to claim 1, characterized in that: The corresponding distributed fault diagnosis model is constructed based on the dynamic latent variable model, including: Using the dynamic latent variable model to construct two variables without time series correlation for each variable block, so as to obtain the dynamic and static relationship of the corresponding variable block; Based on the dynamic and static relationship of each variable block and the principal component analysis method, the detection index and index threshold of each variable block are constructed to obtain the index set and index threshold set corresponding to the multiple variable blocks; A mapping relationship between the indicator set and the indicator threshold set and the fault diagnosis result is constructed, and the distributed fault diagnosis model is constructed based on the mapping relationship.

5. The method according to claim 4, characterized in that in, The expressions of the dynamic relationship index and the dynamic threshold in the detection index and the index threshold are: Among them, J v represents the dynamic monitoring indicator function, represents the current dynamic residual, Represents a dynamic index, P v express v k The principal component space loadings, V represents the dynamic residual data matrix obtained from the historical fault-free data, express v k The residual subspace loadings of represents a statistical distribution derived parameter, J v,th Indicates the threshold value corresponding to the dynamic monitoring indicator. represents the alpha quantile of the chi-square distribution, l v Denotes another statistical distribution derived parameter, h v represents yet another statistical distribution derived parameter; The expressions of the static relationship index and the static threshold in the detection index and the index threshold are: Among them, J e represents the static monitoring indicator function, represents the dynamic residual obtained from static measurement, Represents a static indicator, P e express e k The principal component space loads, E represents the static residual data matrix obtained from the historical fault-free data, express e k The residual subspace loadings of represents a statistical distribution derived parameter, J e,th Indicates the threshold value corresponding to the static monitoring indicator, l e Denotes another statistical distribution derived parameter, h e represents yet another statistical distribution derived parameter.

6. A distributed fault diagnosis method for a large industrial closed-loop control process, characterized in that: Applied to the model use phase, wherein the method comprises the following steps: Obtain current data during the closed-loop control process of the large system to be tested; The current data is input into a pre-built distributed fault diagnosis model to obtain a current fault diagnosis result of the large-scale system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in a large-scale closed-loop control process.

7. A distributed fault diagnosis device for a large industrial closed-loop control process, characterized in that: Applied to the model building stage, wherein the device comprises: An acquisition module, used for acquiring historical fault-free data in a large industrial closed-loop process, and dividing the historical fault-free data based on a preset division rule to obtain a plurality of variable blocks; A classification module, used for classifying the plurality of variable blocks into variable types based on the action mechanism corresponding to each variable block, so as to obtain classification results of the variable blocks, and constructing a corresponding distributed variable topology structure based on the classification results; A first construction module is used to construct a corresponding dynamic latent variable model by combining the distributed variable topology structure and the correlation structure existing in the large closed-loop process data; The second building module is used to build a corresponding distributed fault diagnosis model based on the dynamic latent variable model, so as to output the fault diagnosis result of the large-scale system to be detected by using the distributed fault diagnosis model.

8. A distributed fault diagnosis device for a large industrial closed-loop control process, characterized in that: Applied to the model use stage, wherein the device comprises: An acquisition module is used to acquire current data in the closed-loop control process of the large system to be tested; The diagnostic module is used to input the current data into a pre-built distributed fault diagnosis model to obtain the current fault diagnosis result of the large system to be detected, wherein the distributed fault diagnosis model is constructed from historical fault-free data in a large closed-loop control process.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a distributed fault diagnosis method for a large industrial closed-loop control process as described in any one of claims 1-5 or 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement a distributed fault diagnosis method for a large-scale industrial closed-loop control process as described in any one of claims 1-5 or 6.

Images