Method and system for diagnosing simultaneous anomaly of multiple indexes based on reconstruction probability

Abstract

The invention provides a multi-index simultaneous anomaly diagnosis method and system based on reconstruction probability, and relates to the technical field of data processing, and the method comprises the steps: inputting an obtained multi-index continuous sampling data sequence of a certain monitoring station into a coding network structure of a variational auto-encoder, obtaining a first reconstruction probability of mean value and standard deviation distribution of each input index data; inputting the normal index data into a coding network structure of a variational auto-encoder to obtain a second reconstruction probability of distribution of corresponding indexes of the normal data; the first reconstruction probability and the second reconstruction probability are compared, and if the first reconstruction probability is smaller than the second reconstruction probability and the product of the first reconstruction probability and the second reconstruction probability is smaller than a preset probability threshold value, it is considered that multiple indexes of a certain monitoring station are abnormal; otherwise, considering that the plurality of indexes of a certain monitoring station are not abnormal. The abnormal conditions of the multiple indexes are judged at the same time, and the accuracy and reliability of water quality anomaly detection are improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a diagnosis method and system for simultaneous abnormalities of multiple indicators based on reconstruction probability. Background Art

[0002] At present, despite significant progress in the control and treatment of industrial point sources and urban domestic sewage, my country's water pollution problem is still serious. Non-point source pollution such as farmland fertilizer runoff, pesticides, and intensive livestock and poultry breeding emissions have become one of the main sources of surface water pollution in the basin. When a surface water pollution accident occurs, usually multiple indicators (2 or more indicators) are abnormal at the same time. However, if a single indicator is monitored for abnormalities, it may be missed because the data abnormality of the indicator is not obvious. Therefore, it is necessary to judge the abnormal conditions of multiple indicators at the same time.

[0003] Therefore, the technical problem that urgently needs to be solved is: how to provide a diagnosis method and system for simultaneous abnormalities of multiple indicators based on reconstruction probability, to judge the abnormal conditions of multiple indicators at the same time, and to improve the accuracy and reliability of water quality abnormality detection. Summary of the invention

[0004] The purpose of this application is to provide a method and system for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability, to judge the abnormal conditions of multiple indicators simultaneously, and to improve the accuracy and reliability of water quality abnormality detection.

[0005] To achieve the above-mentioned purpose, as the first aspect of the present application, the present application provides a diagnostic method for simultaneous abnormalities of multiple indicators based on reconstruction probability, the method comprising: obtaining a multi-indicator continuous sampling data sequence of a certain monitoring site; inputting the obtained multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder, and obtaining a first reconstruction probability of the mean and standard deviation distribution of each input indicator data; inputting normal indicator data into the encoding network structure of the variational autoencoder, and obtaining a second reconstruction probability of the distribution of the corresponding indicator of the normal data; comparing the first reconstruction probability and the second reconstruction probability, if the first reconstruction probability is less than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is less than a preset probability threshold, it is considered that multiple indicators of a certain monitoring site are abnormal; otherwise, it is considered that multiple indicators of a certain monitoring site are not abnormal.

[0006] The above-mentioned method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability, wherein the method further includes: if multiple indicators of a monitoring site are abnormal, calculating the comprehensive abnormality scores of the multiple indicators.

[0007] The above-mentioned multi-indicator simultaneous abnormality diagnosis method based on reconstruction probability, wherein the multi-indicator continuous sampling data sequence is represented as: A = {A1, A2, A3, ..., AM}; A M Represents the continuous data sequence of the Mth indicator; A1={A 11 , A 12 , A 13 , …, A 1N}; A 1N Represents the sampling data of the first indicator at the Nth time.

[0008] As described above, the diagnosis method for simultaneous abnormality of multiple indicators based on reconstruction probability, wherein the encoding network structure of the variational autoencoder includes: an encoder and a decoder.

[0009] As described above, the diagnosis method for simultaneous abnormality of multiple indicators based on reconstruction probability, wherein the encoder adopts a convolution + LSTM structure, and the final output layer unit is the number of indicators*2.

[0010] As described above, in the multi-indicator simultaneous abnormality diagnosis method based on reconstruction probability, the decoder samples the mean and standard deviation of each indicator, and then fuses the output of each layer into the same dimension as the input data.

[0011] As a second aspect of the present application, the present application provides a diagnosis system for simultaneous abnormalities of multiple indicators based on reconstruction probability, the system comprising: a data acquisition module, used to acquire a multi-indicator continuous sampling data sequence of a certain monitoring site; a network module, used to input the acquired multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder, and obtain a first reconstruction probability of the mean and standard deviation distribution of each input indicator data; the network module is also used to input normal indicator data into the encoding network structure of the variational autoencoder, and obtain a second reconstruction probability of the distribution of the corresponding indicator of the normal data; a data comparator, used to compare the first reconstruction probability and the second reconstruction probability. If the first reconstruction probability is less than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is less than a preset probability threshold, it is considered that multiple indicators of a certain monitoring site are abnormal; otherwise, it is considered that multiple indicators of a certain monitoring site are not abnormal.

[0012] The diagnosis system for simultaneous abnormality of multiple indicators based on reconstruction probability as described above, wherein the system also includes: a data processor, which is used to calculate the comprehensive abnormality scores of multiple indicators if abnormalities occur in multiple indicators of a monitoring site.

[0013] As described above, in the multi-index simultaneous abnormality diagnosis system based on reconstruction probability, the encoding network structure of the variational autoencoder is set in the network module; the encoding network structure of the variational autoencoder includes: an encoder and a decoder.

[0014] As described above, the multi-index simultaneous abnormality diagnosis system based on reconstruction probability, wherein the encoder adopts a convolution + LSTM structure.

[0015] The beneficial effects achieved by this application are as follows:

[0016] (1) This application optimizes the model structure of the traditional VAE, can diagnose anomalies of multiple indicators at multiple times at the same time, can process multiple indicators at the same time, and improve the accuracy and reliability of water quality anomaly detection.

[0017] (2) The long short-term memory network used in this application has the following advantages:

[0018] Capturing long-term dependencies: LSTM can effectively learn long-term dependencies in sequence data through its unique design structure, especially the introduced "memory unit". This ability enables LSTM to accurately predict important events when processing time series data, even when the intervals and delays are very long, which is difficult to achieve in traditional RNN.

[0019] Solve the problem of gradient disappearance or explosion: Standard RNN is prone to gradient disappearance or explosion when processing long sequence data, making the model difficult to train. LSTM overcomes this limitation by introducing a "gate" mechanism (forget gate, input gate, and output gate) to finely control the reading, writing, and output of information, allowing the model to effectively learn long-term dependencies. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] 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 recorded in the present application. For those skilled in the art, other drawings can also be obtained based on these drawings.

[0021] Figure 1 This is a flow chart of a method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to an embodiment of the present application.

[0022] Figure 2 It is a structural schematic diagram of a diagnosis system for simultaneous abnormalities of multiple indicators based on reconstruction probability according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present application.

[0024] Embodiment 1

[0025] like Figure 1 As shown, the present application provides a method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability, the method comprising the following steps:

[0026] Step S1, obtaining a multi-indicator continuous sampling data sequence of a certain monitoring site.

[0027] Among them, the multi-indicator continuous sampling data sequence is: the data of M indicators at N consecutive moments of a monitoring station.

[0028] Among them, the multi-index continuous sampling data sequence is expressed as: A = {A1, A2, A3, ..., A M}; A M Represents the continuous data sequence of the Mth indicator.

[0029] Where A1={A 11 , A 12 , A 13 , …, A 1N}. 1N Represents the sampling data of the first indicator at the Nth time.

[0030] Step S2, inputting the acquired multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder to obtain the first reconstruction probability of the mean and standard deviation distribution of the inputted indicator data.

[0031] The mean of each indicator data is expressed as: μ = {μ1, μ2, μ3, …, μM}. μM represents the mean of the Mth indicator data. The standard deviation of each indicator data is: a = {a1, a2, a3, …, aM}. aM represents the standard deviation of the Mth indicator data.

[0032] The calculation formula of the first reconstruction probability is:

[0033]

[0034] Where CP represents the first reconstruction probability; U N Indicates the index data value at the Nth moment within the sampling time; U1 indicates the index data value at the 1st moment within the sampling time; represents the cumulative distribution function; μ represents the mean of the indicator data; a represents the standard deviation of the indicator data.

[0035] Among them, the encoding network structure of the variational autoencoder includes: an encoder and a decoder.

[0036] Among them, the encoder adopts the convolution + LSTM (Long Short-Term Memory Network) structure, that is, the convolution structure is combined with the LSTM (Long Short-Term Memory Network) structure, and the final output layer unit is the number of indicators * 2, that is, the number of indicators multiplied by 2. This is different from the traditional VAE (Variational Autoencoder). The long short-term memory network used in this application has the following advantages:

[0037] Capturing long-term dependencies: LSTM can effectively learn long-term dependencies in sequence data through its unique design structure, especially the introduced "memory unit". This ability enables LSTM to accurately predict important events when processing time series data, even when the intervals and delays are very long, which is difficult to achieve in traditional RNN.

[0038] Solve the problem of gradient disappearance or explosion: Standard RNN is prone to gradient disappearance or explosion when processing long sequence data, making the model difficult to train. LSTM overcomes this limitation by introducing a "gate" mechanism (forget gate, input gate, and output gate) to finely control the reading, writing, and output of information, allowing the model to effectively learn long-term dependencies.

[0039] LSMT stands for Long Short-Term Memory Network, which is used to solve the problem of long-term dependency, that is, to remember information over a longer time interval when processing sequence data.

[0040] The basic structure of LSTM consists of four main components: input gate, forget gate, output gate, and memory cell (also called cell state).

[0041] The input gate is denoted as i t , is to determine which part of the current input should be updated to the cell state. It uses a sigmoid function to generate a value between 0 and 1, indicating how much new information should be "memorized". The mathematical expression is: t =σ(W i ·[h t-1 , x t ]+b i );

[0042] Where W i and W C are the weight matrices of the input gate and candidate cell states, respectively, and b i and b C is the corresponding bias term. t is the input for the current time step. is a candidate cell state.

[0043] The Forget Gate is denoted as f t, is to determine which information in the cell state should be forgotten. The sigmoid function is also used to determine the importance of past memories. A value close to 1 means that most of the memory is retained, and a value close to 0 means that most of the memory is forgotten. The mathematical expression is: t =σ(W f ·[h t-1 , x t ]+b f );

[0044] Where σ is the sigmoid activation function, W f is the weight matrix of the forget gate, b f is the bias term of the forget gate, h t-1 is the hidden state of the previous time step, x t is the input for the current time step.

[0045] The cell state is denoted as C t , is the core of LSTM, a vector that can store long-term information. It is updated by combining the forget gate and the cell state of the previous time step, as well as the input gate and a new candidate memory state through a dot product operation. The candidate memory state is obtained by the current input and an input weight matrix through the tanh activation function. The mathematical expression is:

[0046] Candidate cell states are denoted as At each time step, the current input x t and the previous hidden state h t-1 The information is processed to generate a candidate cell state, which contains information that may be added to the long-term state.

[0047] Output Gate is denoted as o t , controls which information in the cell state should be used to generate the output of the current time step. It combines the sigmoid function (determines which cell state content should be output) and the tanh function (scales the selected memory to ensure that the output is between -1 and 1). The mathematical expression is: t =σ(W o ·[h t-1 , x t ]+b o );

[0048] h t =o t *tanh(C t );

[0049] Where W o is the weight matrix of the output gate, b o is the bias term of the output gate.t-1 is the hidden state of the previous time step, x t is the input for the current time step.

[0050] Among them, the decoder samples the mean and standard deviation of each indicator, and then fuses the output of each layer into the same dimension as the input data.

[0051] Step S3, inputting the normal indicator data into the encoding network structure of the variational autoencoder to obtain a second reconstruction probability of the distribution of the corresponding indicator of the normal data.

[0052] The second reconstruction probability of the distribution of the corresponding indicator of normal data can be evaluated by using autoencoders for anomaly detection and condition monitoring. An autoencoder is an artificial neural network that learns efficient encodings of data values ​​in an unsupervised manner. The purpose of this network structure is to learn a representation (encoding) of a set of data, usually for dimensionality reduction. Both the decoder and the encoder are learned so that the autoencoder attempts to generate a representation as close to the original input as possible from the reduced encoding. In the context of anomaly detection and condition monitoring, the basic idea is to use an autoencoder to "compress" the sensor readings into a lower dimensional representation that captures the correlations and interactions between the various variables. The autoencoder network is trained using data that represents a "normal" operating state, with the goal of being able to reconstruct the input variables after dimensionality reduction. The main idea is that as the performance of the monitored equipment degrades, this will affect the interactions between the variables. When the state of the equipment changes, the error of the neural network in reconstructing the input variables will increase significantly. By monitoring the reconstruction error, an indication of the "health" of the monitored equipment can be obtained, as the reconstruction error will increase as the equipment degrades.

[0053] The calculation method of the second reconstruction probability is the same as the calculation method of the first reconstruction probability.

[0054] Step S4, compare the first reconstruction probability and the second reconstruction probability. If the first reconstruction probability is less than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is less than a preset probability threshold, it is considered that multiple indicators of a certain monitoring site are abnormal; otherwise, it is considered that multiple indicators of a certain monitoring site are not abnormal.

[0055] Step S5: If multiple indicators of a monitoring site are abnormal, calculate the comprehensive abnormality scores of the multiple indicators.

[0056] Specifically, the calculation formula for the comprehensive anomaly score of multiple indicators is as follows:

[0057]

[0058] Among them, Zy represents the comprehensive anomaly score of multiple indicators; M represents the number of indicators; Bβ (Xi,N) represents the encoding result of the data at the Nth moment in the i-th indicator data sequence input into the encoder; B β (Z) represents the encoding result of the normal data of the i-th indicator input into the encoder; β represents the parameter of the encoder; Xi represents the input i-th indicator data sequence, N represents the N-th moment; J γ (Xi,N)] represents the result obtained when the data at the Nth moment in the i-th indicator data sequence is input into the decoder; J γ (Zi) represents the result of normal data of the i-th index being input to the decoder; γ represents the parameter of the decoder.

[0059] Embodiment 2

[0060] like Figure 2 As shown, the present application provides a diagnosis system 100 for simultaneous abnormality of multiple indicators based on reconstruction probability, the system comprising:

[0061] The data acquisition module 10 is used to acquire a multi-indicator continuous sampling data sequence of a monitoring site.

[0062] The network module 20 is used to input the acquired multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder to obtain the first reconstruction probability of the mean and standard deviation distribution of each input indicator data.

[0063] The network module 20 is also used to input the normal indicator data into the encoding network structure of the variational autoencoder to obtain a second reconstruction probability of the distribution of the corresponding indicator of the normal data.

[0064] The data comparator 30 is used to compare the first reconstruction probability and the second reconstruction probability. If the first reconstruction probability is smaller than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is smaller than a preset probability threshold, it is considered that multiple indicators of a certain monitoring site are abnormal; otherwise, it is considered that multiple indicators of a certain monitoring site are not abnormal.

[0065] The data processor 40 is used to calculate the comprehensive abnormality scores of the multiple indicators if multiple indicators of a monitoring site are abnormal.

[0066] The encoding network structure of the variational autoencoder is set in the network module 20. The encoding network structure of the variational autoencoder includes: an encoder and a decoder. The encoder adopts a convolution + LSTM structure.

[0067] The present application also provides a computer storage medium, the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the address mapping method of the large-capacity solid-state hard disk. The computer storage medium contains one or more program instructions, and the one or more program instructions are used by the processor to execute a diagnosis method for simultaneous abnormalities of multiple indicators based on reconstruction probability.

[0068] The embodiments disclosed in the present invention provide a computer-readable storage medium, in which computer program instructions are stored. When the computer program instructions are executed on a computer, the computer executes the above-mentioned method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability.

[0069] An embodiment of the present invention provides a processor for processing the above-mentioned diagnosis method for simultaneous abnormality of multiple indicators based on reconstruction probability.

[0070] In the embodiment of the present invention, the processor may be an integrated circuit chip having the ability to process signals. The processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0071] The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor for execution, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The processor reads the information in the storage medium and completes the steps of the above method in combination with its hardware.

[0072] The storage medium may be a memory, which may be, for example, a volatile memory or a nonvolatile memory, or may include both volatile and nonvolatile memory.

[0073] Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct RAM bus random access memory (DRRAM).

[0074] The beneficial effects achieved by this application are as follows:

[0075] (1) This application optimizes the model structure of the traditional VAE, can diagnose anomalies of multiple indicators at multiple times at the same time, can process multiple indicators at the same time, and improve the accuracy and reliability of water quality anomaly detection.

[0076] (2) The long short-term memory network used in this application has the following advantages:

[0077] Capturing long-term dependencies: LSTM can effectively learn long-term dependencies in sequence data through its unique design structure, especially the introduced "memory unit". This ability enables LSTM to accurately predict important events when processing time series data, even when the intervals and delays are very long, which is difficult to achieve in traditional RNN.

[0078] Solve the problem of gradient disappearance or explosion: Standard RNN is prone to gradient disappearance or explosion when processing long sequence data, making the model difficult to train. LSTM overcomes this limitation by introducing a "gate" mechanism (forget gate, input gate, and output gate) to finely control the reading, writing, and output of information, allowing the model to effectively learn long-term dependencies.

[0079] In the description of this application, 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. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise clearly and specifically defined.

[0080] In the description of the present application, the word "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.

[0081] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability, characterized in that: The method includes: Obtain a multi-indicator continuous sampling data sequence at a monitoring site; Input the acquired multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder to obtain the first reconstruction probability of the mean and standard deviation distribution of each input indicator data; Input the normal index data into the encoding network structure of the variational autoencoder to obtain a second reconstruction probability of the distribution of the corresponding index of the normal data; Compare the first reconstruction probability and the second reconstruction probability. If the first reconstruction probability is less than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is less than the preset probability threshold, it is considered that multiple indicators of a monitoring site are abnormal; otherwise, it is considered that multiple indicators of a monitoring site are not abnormal.

2. The method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to claim 1, characterized in that: The method further includes: If multiple indicators of a monitoring site are abnormal, the comprehensive abnormality scores of multiple indicators are calculated.

3. The method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to claim 1, characterized in that: The multi-index continuous sampling data sequence is expressed as: A = {A1, A2, A3, ..., A M }; A M Represents the continuous data sequence of the Mth indicator; A1={A 11 , A 12 , A 13 , …, A 1N }; A 1N Represents the sampling data of the first indicator at the Nth time.

4. The method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to claim 1, characterized in that: The encoding network structure of the variational autoencoder includes: an encoder and a decoder.

5. The method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to claim 4 is characterized in that: The encoder adopts a convolution + LSTM structure, and the final output layer unit is the number of indicators * 2.

6. The method for diagnosing simultaneous abnormalities of multiple indicators based on reconstruction probability according to claim 4, characterized in that: The decoder samples the mean and standard deviation of each indicator, and then fuses the output of each layer into the same dimension as the input data.

7. A multi-index simultaneous abnormality diagnosis system based on reconstruction probability, characterized in that: The method according to any one of claims 1 to 6 is implemented, the system comprising: A data acquisition module is used to obtain a multi-indicator continuous sampling data sequence of a monitoring site; A network module is used to input the acquired multi-indicator continuous sampling data sequence of a certain monitoring site into the encoding network structure of the variational autoencoder to obtain the first reconstruction probability of the mean and standard deviation distribution of each input indicator data; The network module is further used to input the normal indicator data into the encoding network structure of the variational autoencoder to obtain a second reconstruction probability of the distribution of the corresponding indicator of the normal data; A data comparator is used to compare the first reconstruction probability and the second reconstruction probability. If the first reconstruction probability is less than the second reconstruction probability, and the product of the first reconstruction probability and the second reconstruction probability is less than a preset probability threshold, it is considered that multiple indicators of a certain monitoring site are abnormal; otherwise, it is considered that multiple indicators of a certain monitoring site are not abnormal.

8. The multi-index simultaneous abnormality diagnosis system based on reconstruction probability according to claim 7 is characterized in that: The system also includes: The data processor is used to calculate the comprehensive anomaly scores of multiple indicators if multiple indicators of a monitoring site are abnormal.

9. The multi-index simultaneous abnormality diagnosis system based on reconstruction probability according to claim 7 is characterized in that: The network module is provided with an encoding network structure of a variational autoencoder; The encoding network structure of the variational autoencoder includes: an encoder and a decoder.

10. The multi-index simultaneous abnormality diagnosis system based on reconstruction probability according to claim 9 is characterized in that: The encoder adopts a convolution + LSTM structure.

Images