An industrial production anomaly detection method based on information masking and related device

Abstract

The present invention belongs to the field of industrial production monitoring technology, and discloses an industrial production anomaly detection method based on information masking and related devices; wherein, the industrial production anomaly detection method based on information masking includes: obtaining original industrial signal data to be detected in industrial production; based on the obtained original industrial signal data, using a trained reconstruction model to reconstruct the data to obtain a data reconstruction result; obtaining a reconstruction error between the data reconstruction result and the original industrial signal data, and comparing the reconstruction error with a preset reconstruction error threshold, and obtaining an industrial production anomaly detection result based on the comparison result. The technical solution disclosed by the present invention enhances generalization by introducing information masking and reconstruction ideas, so that the technical solution of the present invention can achieve efficient anomaly detection in different workshops and different production lines, can adapt to different industrial environments, and can achieve real-time detection of abnormal situations in the industrial production process.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial production monitoring, and in particular relates to an industrial production anomaly detection method based on information masking and a related device. Background Art

[0002] In existing industrial production processes, anomaly detection is a key link in quality control and production efficiency optimization. It not only helps companies promptly detect abnormal conditions on the production line and prevent potential product defects, but also reduces downtime and resource waste, significantly improving overall operational efficiency.

[0003] As industrial production systems become increasingly complex and data volumes explode, existing traditional anomaly detection methods are gradually exposing their limitations. Specifically, most traditional anomaly detection methods rely on manually set fixed thresholds or simple statistical methods based on historical data, such as standard deviation and mean, to determine whether data deviates from the normal range. While these existing methods can be effective in relatively stable production environments with well-defined data characteristics, in today's highly dynamic and volatile industrial environments, data from industrial production processes often exhibit nonlinearity, high dimensionality, and strong noise. These characteristics make it difficult for fixed thresholds and simple statistical methods to effectively capture subtle changes and complex patterns in the data, resulting in frequent missed detections and false positives, significantly reducing detection effectiveness and generalization capabilities.

[0004] To sum up, with the advancement of technology, especially the integrated application of technologies such as artificial intelligence and big data, industrial anomaly detection is gradually moving towards a new stage of greater intelligence and precision. In order to adapt to the anomaly detection needs of different industrial environments, it is urgent to develop a new industrial production anomaly detection solution. Summary of the Invention

[0005] The present invention aims to provide an industrial production anomaly detection method and related apparatus based on information masking to address one or more of the aforementioned technical problems. The disclosed technical solution enhances generalizability by introducing information masking and reconstruction concepts, enabling efficient anomaly detection across diverse workshops and production lines. This adaptability allows for real-time detection of anomalies in industrial production processes.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides an industrial production anomaly detection method based on information masking, comprising the following steps:

[0008] Obtaining original industrial signal data to be tested in industrial production;

[0009] Based on the acquired original industrial signal data, the trained reconstruction model is used to reconstruct the data and obtain the data reconstruction result;

[0010] Obtaining a reconstruction error between the data reconstruction result and the original industrial signal data, and comparing the reconstruction error with a preset reconstruction error threshold, and obtaining an industrial production anomaly detection result based on the comparison result;

[0011] The reconstruction model is provided with a mask network layer, which is used to avoid the original industrial signal data X=[x1, x2, ..., x i ,…,x n ] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping.

[0012] A further improvement of the present invention is that

[0013] In the training step of the reconstruction model, the input data is the original reconstructed data, the output target data is the data reconstruction result of the original reconstructed data, and the model training is completed when the reconstruction error between the input data and the output target data reaches a preset convergence requirement;

[0014] Among them, the reconstruction model first divides the input original reconstructed data into multiple data segments, and then processes each data segment in sequence to obtain the final data reconstruction result; when processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the selected data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain the reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

[0015] A further improvement of the present invention is that

[0016] In the step of dividing the input original reconstructed data into multiple data segments, each data segment includes a selected data point and two data points adjacent before and after the selected data point; wherein, in the data segment corresponding to the first data point, the data points adjacent before the first data point are replaced by 0; and in the data segment corresponding to the last data point, the data points adjacent after the last data point are replaced by 0.

[0017] A further improvement of the present invention is that

[0018] When processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

[0019] The selected data segment is [x i-1 ,x i ,x i+1 ], the data segment input to the mask network layer is [x i-1 ,x i ,x i+1 ], the output masked data segment is [x i-1 ,0,x i+1 ], the reconstructed data obtained is

[0020] A further improvement of the present invention is that

[0021] In the step of obtaining the reconstruction error between the data reconstruction result and the original industrial signal data,

[0022] Original industrial signal data X=[x1,x2,…,x i ,…,x n ] and data reconstruction results The reconstruction error is expressed as ||·||2 represents the Euclidean norm.

[0023] A second aspect of the present invention provides an industrial production anomaly detection system based on information masking, comprising:

[0024] A data acquisition module is used to acquire the original industrial signal data to be detected in industrial production;

[0025] The reconstruction module is used to reconstruct the data based on the acquired original industrial signal data using the trained reconstruction model to obtain the data reconstruction result;

[0026] A comparison and analysis module is used to obtain a reconstruction error between the data reconstruction result and the original industrial signal data, and compare the reconstruction error with a preset reconstruction error threshold, and obtain an industrial production anomaly detection result based on the comparison result;

[0027] The reconstruction model is provided with a mask network layer, which is used to avoid the original industrial signal data X=[x1, x2, ..., x i ,…,xn ] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping.

[0028] A further improvement of the present invention is that

[0029] In the training step of the reconstruction model, the input data is the original reconstructed data, the output target data is the data reconstruction result of the original reconstructed data, and the model training is completed when the reconstruction error between the input data and the output target data reaches a preset convergence requirement;

[0030] Among them, the reconstruction model first divides the input original reconstructed data into multiple data segments, and then processes each data segment in sequence to obtain the final data reconstruction result; when processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the selected data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain the reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

[0031] A further improvement of the present invention is that

[0032] In the step of dividing the input original reconstructed data into multiple data segments, each data segment includes a selected data point and two data points adjacent before and after the selected data point; wherein, in the data segment corresponding to the first data point, the data points adjacent before the first data point are replaced by 0; and in the data segment corresponding to the last data point, the data points adjacent after the last data point are replaced by 0.

[0033] A further improvement of the present invention is that

[0034] When processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

[0035] The selected data segment is [x i-1 ,x i ,xi+1 ], the data segment input to the mask network layer is [x i-1 ,x i ,x i+1 ], the output masked data segment is [x i-1 ,0,x i+1 ], the reconstructed data obtained is

[0036] A further improvement of the present invention is that

[0037] In the step of obtaining the reconstruction error between the data reconstruction result and the original industrial signal data,

[0038] Original industrial signal data X=[x1,x2,…,x i ,…,x n ] and data reconstruction results The reconstruction error is expressed as ||·||2 represents the Euclidean norm.

[0039] In a third aspect of the present invention, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for detecting an industrial production anomaly based on information mask as described in any one of the first aspects of the present invention is implemented.

[0040] In a fourth aspect, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting an anomaly in industrial production based on information mask as described in any one of the first aspects of the present invention is implemented.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention discloses an industrial production anomaly detection method based on information masking, which can be used to perform anomaly detection on industrial signal data in industrial production to obtain anomaly detection results; wherein, the technical solution of the present invention is based on data reconstruction and information masking to complete anomaly detection, and the reconstruction model performs a masking operation on the input information, avoiding the direct mapping between the reconstructed data and the reconstructed data, forcing the model to learn more general data patterns to improve the detection level, and improving the generalization of the anomaly detection method. In summary, the present invention enhances generalization by introducing the technical means of information masking and data reconstruction, can achieve efficient anomaly detection in different workshops and different production lines, can adapt to different industrial environments, and can achieve real-time detection of abnormal situations in the industrial production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the present invention or 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 some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 The figure is a flow chart of an industrial production anomaly detection method based on information masking in an embodiment of the present invention.

[0045] Figure 2 4 is a flow chart of information mask operation in an embodiment of the present invention.

[0046] Figure 3 It is a schematic diagram of the architecture of the reconstruction model in an embodiment of the present invention.

[0047] Figure 4 FIG. 4 is a logic diagram of an anomaly scoring mechanism in an embodiment of the present invention.

[0048] Figure 5 2 is a schematic diagram of an industrial production anomaly detection system based on information masking in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments and technical solutions are only part of the embodiments of the present invention, not all of the embodiments.

[0050] All other embodiments obtained by persons of ordinary skill in the art based on the technical solutions disclosed in the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.

[0051] See also Figure 1 The embodiment of the present invention discloses an industrial production anomaly detection method based on information masking, comprising the following steps:

[0052] Step 1: Obtain the original industrial signal data to be detected in industrial production;

[0053] Step 2: Based on the original industrial signal data obtained in step 1, the reconstruction model is used to reconstruct the data to obtain a data reconstruction result;

[0054] Step 3: Obtain a reconstruction error between the data reconstruction result and the original industrial signal data, compare the reconstruction error with a preset reconstruction error threshold, and obtain an industrial production anomaly detection result based on the comparison result;

[0055] In the reconstruction model, a mask network layer is provided to avoid the original industrial signal data X=[x1,x2,…,x i ,…,x n ] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping.

[0056] The embodiment of the present invention discloses a new method for detecting anomalies in industrial production, which combines information masking and data reconstruction technology to improve the wide adaptability of anomaly detection in industrial signal data. Specifically, the embodiment of the present invention proposes an anomaly detection scheme based on information masking, which is designed for industrial production environments and can effectively identify anomalies in the production process in real time. The core of the method is to use data reconstruction and information masking technology to process the input industrial signal data, thereby realizing anomaly detection. On the basis of data reconstruction, the present invention performs a masking operation on part of the information, that is, hides or modifies part of the data to avoid the model directly relying on the simple mapping relationship between the reconstructed data and the original data. This step prompts the model to learn deeper and more generalized data patterns; through information masking technology, the model is forced to learn the general laws between the data rather than the characteristics of specific samples, thereby improving its adaptability in different workshops, production lines and even industrial environments; the combination of information masking and data reconstruction enables the model to capture more subtle abnormal signals, thereby improving the accuracy and sensitivity of anomaly detection.

[0057] To further explain, the model provided by the embodiment of the present invention can effectively detect abnormal individuals in industrial signals. When the model is migrated from experimental data to similar products in different workshops during deployment, the detection performance is improved.

[0058] See also Figure 2 and Figure 3 In one embodiment of the present invention, further explanation of the reconstruction model is as follows:

[0059] The input data of the reconstruction model is the original reconstructed data, and the output target data is the data reconstruction result of the original reconstructed data. The input and output should be as consistent as possible. This model is trained on normal data and does not learn the pattern of abnormal samples. Therefore, the reconstruction error when reconstructing abnormal samples will become higher, resulting in poor reliability. In view of the problems existing in the above-mentioned existing reconstruction models, the reconstruction model of the embodiment of the present invention is different from the existing traditional network scheme that uses each dimension of the input data to construct a mapping to the output. The technical solution of the present invention sets a mask network layer to avoid the direct mapping of each data segment in the reconstructed data to the corresponding data segment of the data reconstruction result, forcing the model to learn more general data patterns to improve the detection level.

[0060] Specifically, first, the original reconstructed data is divided into multiple small data segments, such as [x1, x2, x3] is a data segment of length 3; then, a convolution kernel is used to perform a convolution operation on the data segment, and a mask network layer is inserted between the data segment and the convolution kernel. The input of the mask network layer is the data segment, and the output is the masked data segment. For example, the input is [x1, x2, x3], and the output is [x1, 0, x3]. Through this information masking processing method, the data point in the middle of the data segment is lost; the convolution kernel and the data segment have the same size, which ensures that the convolution kernel will not move on the data segment, and the output of the convolution kernel is It is a number after the convolution operation; the model uses multiple convolution kernels to perform convolution operations with the data segment at the same time. The results of multiple convolutions will be flattened to form a new sequence. For example, after the data segment [x1, x2, x3] passes through the mask network layer and the convolution operation, a new sequence [y1, y2, y3] is obtained. The new sequence does not contain the information of the x2 data point in the input data segment sequence, which results in information loss. By masking part of the data, the model is forced to reconstruct the result not to depend on the reconstructed data, but to learn a more global and general data pattern. The new sequence obtained after flattening the convolution result is reduced to 1 dimension using multiple linear layers to obtain the reconstructed data. In addition, by parallelizing multiple mask network layers and convolution kernels, we can reconstruct the [x1,x2,x3], [x2,x3,x4] order.

[0061] To explain further, the original reconstructed data sequences x1 and x n Then we use the operation of adding 0 to the beginning part. Specifically, we design [0,x1,x2] and [x n-1 ,x n ,0] is used to get and Will arrive Arrange them in sequence to get the complete data reconstruction result

[0062] See also Figure 4 In one embodiment of the present invention, in the step of obtaining a reconstruction error between a data reconstruction result and the original industrial signal data, and comparing the reconstruction error with a preset reconstruction error threshold, and obtaining an industrial production anomaly detection result based on the comparison result,

[0063] Original industrial signal data X=[x1,x2,…,x i ,…,x n ] and data reconstruction results The reconstruction error is expressed as ||·||2 represents the Euclidean norm. If the calculated reconstruction error is greater than a preset reconstruction error threshold, the original industrial signal data is called abnormal data.

[0064] In a further preferred technical solution, the preset reconstruction error threshold is designed as follows: the reconstruction error of each sample is arranged from small to large, and α% of the samples in the last part are abnormal.

[0065] In a specific embodiment of the present invention, taking a specific industrial production scenario as an example, the signal data collected on the production line is first subjected to an information masking operation and then input into a reconstruction model for training. After the model training is completed, the newly collected data is monitored in real time through an anomaly scoring mechanism to promptly detect and report abnormal situations.

[0066] The following are device embodiments of the present invention, which can be used to perform the method embodiments of the present invention. For details not disclosed in the device embodiments, please refer to the method embodiments of the present invention.

[0067] See also Figure 5 In an embodiment of the present invention, a system for detecting anomalies in industrial production based on information masking is provided, comprising:

[0068] A data acquisition module is used to acquire the original industrial signal data to be detected in industrial production;

[0069] The reconstruction module is used to reconstruct the data based on the acquired original industrial signal data using the trained reconstruction model to obtain the data reconstruction result;

[0070] A comparison and analysis module is used to obtain a reconstruction error between the data reconstruction result and the original industrial signal data, and compare the reconstruction error with a preset reconstruction error threshold, and obtain an industrial production anomaly detection result based on the comparison result;

[0071] The reconstruction model is provided with a mask network layer, which is used to avoid the original industrial signal data X=[x1, x2, ..., x i ,…,x n] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping.

[0072] In one embodiment of the present invention, a computer device is provided, the computer device including a processor and a memory, the memory being used to store a computer program, the computer program including program instructions, and the processor being used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used to perform the operation of the industrial production anomaly detection method based on information mask.

[0073] In one embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory). The computer-readable storage medium is a memory device in a computer device for storing programs and data. It is understood that the computer-readable storage medium herein may include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides storage space, which stores the terminal's operating system. In addition, the storage space also stores one or more instructions suitable for being loaded and executed by a processor. These instructions may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium herein may be a high-speed RAM (Random Access Memory) memory or a non-volatile memory, such as at least one disk storage device. The processor may load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the industrial production anomaly detection method based on information mask in the above-mentioned embodiment.

[0074] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, optical storage, etc.) that contain computer-usable program code.

[0075] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0076] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for detecting anomalies in industrial production based on information masking, characterized in that: The following steps are involved: Obtaining original industrial signal data to be tested in industrial production; Based on the acquired original industrial signal data, the trained reconstruction model is used to reconstruct the data and obtain the data reconstruction result; Obtaining a reconstruction error between the data reconstruction result and the original industrial signal data, and comparing the reconstruction error with a preset reconstruction error threshold, and obtaining an industrial production anomaly detection result based on the comparison result; The reconstruction model is provided with a mask network layer, which is used to avoid the original industrial signal data X=[x1, x2, ..., x i ,…,x n ] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping of In the training step of the reconstruction model, the input data is the original reconstructed data, the output target data is the data reconstruction result of the original reconstructed data, and the model training is completed when the reconstruction error between the input data and the output target data reaches a preset convergence requirement; Among them, the reconstruction model first divides the input original reconstructed data into multiple data segments, and then processes each data segment in sequence to obtain the final data reconstruction result; when processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the selected data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain the reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

2. The method for detecting anomalies in industrial production based on information masking according to claim 1, characterized in that: In the step of dividing the input original reconstructed data into multiple data segments, each data segment includes a selected data point and two data points adjacent before and after the selected data point; wherein, in the data segment corresponding to the first data point, the data points adjacent before the first data point are replaced by 0; and in the data segment corresponding to the last data point, the data points adjacent after the last data point are replaced by 0.

3. The method for detecting anomalies in industrial production based on information masking according to claim 2, characterized in that: When processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size. The selected data segment is [x i-1 ,x i ,x i+1 ], the data segment of the mask network layer input is [x i-1 ,x i ,x i+1 ], the output masked data segment is [x i-1 ,0,x i+1 ], the reconstructed data obtained is 4. The method for detecting anomalies in industrial production based on information masking according to claim 1, characterized in that: In the step of obtaining the reconstruction error between the data reconstruction result and the original industrial signal data, Original industrial signal data X=[x1,x2,…,x i ,…,x n ] and data reconstruction results The reconstruction error is expressed as ||·||2 represents the Euclidean norm.

5. An industrial production anomaly detection system based on information masking, characterized in that: include: A data acquisition module is used to acquire the original industrial signal data to be detected in industrial production; The reconstruction module is used to reconstruct the data based on the acquired original industrial signal data using the trained reconstruction model to obtain the data reconstruction result; A comparison and analysis module is used to obtain a reconstruction error between the data reconstruction result and the original industrial signal data, and compare the reconstruction error with a preset reconstruction error threshold, and obtain an industrial production anomaly detection result based on the comparison result; The reconstruction model is provided with a mask network layer, which is used to avoid the original industrial signal data X=[x1, x2, ..., x i ,…,x n ] to data reconstruction results In the mapping process, the i-th original data point x i To the i-th reconstructed data point Direct mapping of In the training step of the reconstruction model, the input data is the original reconstructed data, the output target data is the data reconstruction result of the original reconstructed data, and the model training is completed when the reconstruction error between the input data and the output target data reaches a preset convergence requirement; Among them, the reconstruction model first divides the input original reconstructed data into multiple data segments, and then processes each data segment in sequence to obtain the final data reconstruction result; when processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the selected data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain the reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size.

6. The industrial production anomaly detection system based on information masking according to claim 5 is characterized in that: In the step of dividing the input original reconstructed data into multiple data segments, each data segment includes a selected data point and two data points adjacent before and after the selected data point; wherein, in the data segment corresponding to the first data point, the data points adjacent before the first data point are replaced by 0; and in the data segment corresponding to the last data point, the data points adjacent after the last data point are replaced by 0.

7. The industrial production anomaly detection system based on information masking according to claim 6 is characterized in that: When processing each data segment in sequence, for the selected data segment, multiple convolution kernels are used to perform convolution operations with the data segment at the same time, the results of the multiple convolution operations are flattened to form a new sequence, and the new sequence is reduced to one dimension using a multi-linear layer to obtain reconstructed data corresponding to the selected data segment; when multiple convolution kernels are used to perform convolution operations with the data segment at the same time, a mask network layer is inserted between the data segment and the convolution kernel, the input of the mask network layer is the data segment, and the output is the masked data segment, and the convolution kernel and the data segment have the same size. The selected data segment is [x i-1 ,x i ,x i+1 ], the data segment of the mask network layer input is [x i-1 ,x i ,x i+1 ], the output masked data segment is [x i-1 ,0,x i+1 ], the reconstructed data obtained is 8. The industrial production anomaly detection system based on information masking according to claim 5 is characterized in that: In the step of obtaining the reconstruction error between the data reconstruction result and the original industrial signal data, Original industrial signal data X=[x1,x2,…,x i ,…,x n ] and data reconstruction results The reconstruction error is expressed as ||·||2 represents the Euclidean norm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the industrial production anomaly detection method based on information mask according to any one of claims 1 to 4 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting anomalies in industrial production based on information masks according to any one of claims 1 to 4 is implemented.

Images