Factory monitoring and safety system and method integrating multiple systems

Abstract

The invention discloses a multi-system fused factory monitoring and safety system and method, and relates to the technical field of industrial data processing, a subsystem information set, a historical monitoring data set and a historical abnormal label set are collected in advance, an incidence matrix calculation model is constructed based on the subsystem information set and the historical monitoring data set, and a multi-system fused factory monitoring and safety system is constructed; extracting a historical incidence matrix from a historical monitoring data set by using an incidence matrix calculation model, extracting an abnormal early warning feature vector from the historical incidence matrix, constructing and training an abnormal early warning judgment model based on the abnormal early warning feature vector and a historical abnormal label set, and performing abnormal early warning by using a time sequence of the historical incidence matrix. Constructing and training a matrix prediction model; under the condition of large-scale data, the system can quickly respond and provide accurate early warning and prediction, and the intelligent level and the safety management capability of the factory monitoring system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial data processing, and specifically to a factory monitoring and safety system and method integrating multiple systems. Background Art

[0002] In modern industrial production, factories are often composed of multiple complex subsystems, including production processes, material transportation, energy management, environmental control, safety and monitoring, etc. There are complex interdependent relationships among these subsystems, and any abnormality in one subsystem may have a significant impact on the overall production process. Therefore, realizing intelligent safety monitoring and early warning prediction for these subsystems has become an important task to ensure the safe and efficient operation of factories. However, existing technologies often face problems such as data islands, independent operation of systems, and poor information sharing, resulting in limited effectiveness of overall monitoring and risk management. Traditional monitoring systems usually only focus on a single subsystem or device, lacking the ability to identify complex inter-system relationships and being difficult to provide early warnings for potential failures.

[0003] The patent application with the publication number CN118134093A discloses a quality control system for a smart factory based on machine learning, including an adaptive module, a multimodal analysis module, a simulation module, and a feedback and adjustment module. The adaptive module is used to self-adjust and optimize the changes in production data, and is adjusted in real time through a machine learning model to adapt to new data patterns; the multimodal analysis module integrates and identifies features of different types of data, fuses multimodal data, and identifies data similarities and differences. However, this technical solution fails to solve the problem of data fusion of multiple subsystems.

[0004] Therefore, the present invention proposes a factory monitoring and safety system and method integrating multiple systems. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. For this purpose, the present invention proposes a factory monitoring and safety system and method integrating multiple systems, enabling the system to quickly respond and provide accurate early warning prediction under the condition of large-scale data, and improving the intelligent level and safety management ability of the factory monitoring system.

[0006] To achieve the above object, a factory monitoring and safety method integrating multiple systems is proposed, including the following steps:

[0007] Step 1: Pre-collect the subsystem information set, historical monitoring data set, and historical anomaly label set;

[0008] Step 2: Based on the subsystem information set and the historical monitoring data set, construct an association matrix calculation model;

[0009] Step 3: From the historical monitoring data set, extract the historical correlation matrix using the correlation matrix calculation model, and extract the abnormal warning feature vectors from the historical correlation matrix;

[0010] Step 4: Based on the abnormal warning feature vectors and the historical abnormal label set, construct and train an abnormal warning judgment model;

[0011] Step 5: Use the time series of the historical correlation matrix to construct and train a matrix prediction model;

[0012] Step 6: During the actual monitoring process, collect real-time monitoring data, obtain the real-time correlation matrix based on the real-time monitoring data and the correlation matrix calculation model, obtain the future correlation matrix based on the future correlation matrix and the matrix prediction model, and obtain the warning judgment for the future based on the future correlation matrix and the abnormal warning judgment model;

[0013] The collection method of the subsystem information set is as follows:

[0014] Obtain the basic information of each factory device in each subsystem in the factory from the technical specifications of the devices and sensors as the subsystem information set;

[0015] The collection method of the historical monitoring data set is as follows:

[0016] Extract the historical monitoring data generated by each subsystem and each device in each subsystem during the historical operation process from the factory's monitoring and data acquisition system or directly export it from the factory's back-end database;

[0017] The collection method of the historical abnormal label set is as follows:

[0018] Extract from the operator event log or abnormal report or collect from the factory's event management or reporting system, and obtain the abnormal label data each time an abnormal event occurs to form the historical abnormal label set;

[0019] The method of constructing the correlation matrix calculation model based on the subsystem information set and the historical monitoring data set is as follows:

[0020] For each subsystem included in the subsystem information set:

[0021] According to the measurement data of each device in this subsystem in the historical monitoring data set, construct a device correlation matrix for each subsystem;

[0022] Extract the main eigenvectors of the device correlation matrices of each subsystem, and construct a subsystem correlation matrix between each subsystem based on the main eigenvectors of each subsystem;

[0023] The method of constructing a device correlation matrix for each subsystem is as follows:

[0024] Set the size T of the time window and the sliding step s;

[0025] In the historical monitoring data set, mark each unit time as t;

[0026] Mark the number of each subsystem as i;

[0027] Mark the number of each device in the i-th subsystem as ij;

[0028] For each unit time t, extract the parameter set composed of the measurement data of the ij-th device in each subsystem from unit time t - T to unit time t, and form the data window W(Sij,t); where Sij represents the parameter time series composed of the measurement data of the ij-th device in the i-th subsystem arranged in chronological order;

[0029] Use the correlation coefficient analysis method to calculate the correlation between devices in the i-th subsystem and generate a device correlation matrix;

[0030] The method for extracting the main eigenvector of the device correlation matrix of each subsystem is as follows:

[0031] Perform singular value decomposition on the device correlation matrix by using the singular value decomposition method to obtain the main eigenvector;

[0032] The method for obtaining the main eigenvector by using the singular value decomposition method is as follows:

[0033] Mark the device correlation matrix as A;

[0034] Use the singular value decomposition method to transform the matrix A into the form of A = U∑V T ;

[0035] Among them, both U and V are orthogonal matrices. The column vectors of U are called left singular vectors, and each column vector of V is a right singular vector; Σ is a singular value matrix, and the singular value matrix is a diagonal matrix containing singular values arranged in descending order;

[0036] Select the largest singular value from the singular value matrix, and use the left singular vector corresponding to the largest singular value as the main eigenvector;

[0037] The method for constructing the subsystem correlation matrix between each subsystem based on the main eigenvector of each subsystem is as follows:

[0038] Mark any two subsystems as i1 and i2 respectively;

[0039] For the main eigenvector of subsystem i1 and the main eigenvector of subsystem i2, use the Pearson correlation coefficient formula to calculate the subsystem correlation degree between subsystem i1 and subsystem i2;

[0040] Construct a subsystem correlation matrix according to the subsystem correlation degrees among the subsystems;

[0041] The method for extracting the historical correlation matrix from the historical monitoring data set using the correlation matrix calculation model is as follows:

[0042] For each time window in the historical monitoring process, calculate the subsystem correlation matrix among the subsystems within each time window through the calculation method of the subsystem correlation matrix, and use it as the historical correlation matrix;

[0043] The method for extracting the abnormal warning feature vector from the historical correlation matrix is as follows:

[0044] Use the singular value decomposition method for the historical correlation matrix of each time window to extract the main eigenvector of the historical correlation matrix as the abnormal warning feature vector;

[0045] The method for constructing and training the abnormal warning judgment model based on the abnormal warning feature vector and the historical abnormal label set is as follows:

[0046] Use any classification model as the abnormal warning judgment model;

[0047] The abnormal warning judgment model takes the abnormal warning feature vector as the input and the predicted value of the abnormal type number corresponding to the time window of the abnormal warning feature vector as the output;

[0048] The abnormal warning judgment model takes the cross-entropy error between the predicted value of the abnormal type number and the actual value of the abnormal type number of the corresponding time window as the first error function;

[0049] Use the abnormal warning feature vectors and the actual values of the abnormal type numbers within all time windows in the historical monitoring process to train the abnormal warning judgment model until the first error function converges;

[0050] The method for constructing and training the matrix prediction model using the time series of the historical correlation matrix is as follows:

[0051] Set the matrix prediction model as an LSTM network model;

[0052] The matrix prediction model takes the sequence of the subsystem correlation matrices within the past time windows as the input;

[0053] The output layer of the matrix prediction model outputs the predicted matrix of the subsystem correlation matrix within the time window after the next L time steps as the output; L is the preset prediction time step;

[0054] Take the true subsystem correlation matrix after L time steps corresponding to each set of inputs as the prediction target of the prediction matrix, and take the matrix cosine similarity between the true subsystem correlation matrix and the prediction matrix as the second error function;

[0055] Train the matrix prediction model until the second error function converges.

[0056] A factory monitoring and safety system integrating multiple systems is proposed, including a historical data collection module, a correlation matrix calculation module, a warning model training module, a matrix prediction module, and a safety warning module; among them, each module is connected electrically;

[0057] The historical data collection module pre-collects a subsystem information set, a historical monitoring data set, and a historical anomaly label set, sends the subsystem information set and the historical monitoring data set to the correlation matrix calculation module, and sends the historical anomaly label set to the warning model training module;

[0058] The correlation matrix calculation module constructs a correlation matrix calculation model based on the subsystem information set and the historical monitoring data set, and sends the correlation matrix calculation model to the warning model training module;

[0059] The warning model training module extracts the historical correlation matrix from the historical monitoring data set using the correlation matrix calculation model, extracts the abnormal warning feature vector from the historical correlation matrix, constructs and trains an abnormal warning judgment model based on the abnormal warning feature vector and the historical anomaly label set, sends the historical correlation matrix to the matrix prediction module, and sends the abnormal warning judgment model to the safety warning module;

[0060] The matrix prediction module constructs and trains a matrix prediction model using the time series of the historical correlation matrix, and sends the matrix prediction model to the safety warning module;

[0061] The safety warning module collects real-time monitoring data during the actual monitoring process, obtains the real-time correlation matrix based on the real-time monitoring data and the correlation matrix calculation model, obtains the future correlation matrix based on the real-time correlation matrix and the matrix prediction model, and obtains the future warning judgment based on the future correlation matrix and the abnormal warning judgment model.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] The present invention pre-collects a subsystem information set, a historical monitoring data set, and a historical anomaly label set. Based on the subsystem information set and the historical monitoring data set, an association matrix calculation model is constructed. From the historical monitoring data set, a historical association matrix is extracted using the association matrix calculation model, and an anomaly warning feature vector is extracted from the historical association matrix. Based on the anomaly warning feature vector and the historical anomaly label set, an anomaly warning judgment model is constructed and trained. Using the time series of the historical association matrix, a matrix prediction model is constructed and trained. Finally, in the actual monitoring process, real-time monitoring data is collected, a real-time association matrix is obtained based on the real-time monitoring data and the association matrix calculation model, a future association matrix is obtained based on the real-time association matrix and the matrix prediction model, and a warning judgment for the future is obtained based on the future association matrix and the anomaly warning judgment model. This solution realizes the high efficiency of calculation in the correlation analysis process. By using dimensionality reduction techniques such as singular value decomposition to extract the main feature vectors of the devices and the anomaly correlation degree between subsystems, the data calculation amount is greatly reduced, the operation speed is improved, and the integrity of the main information is maintained. This enables the system to quickly respond and provide accurate warning predictions under large-scale data conditions, improving the intelligent level and safety management ability of the factory monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a flowchart of a factory monitoring and safety method integrating multiple systems in Embodiment 1 of the present invention;

[0065] Figure 2 is an example diagram of the topological relationship between each device and production line in a production system in Embodiment 1 of the present invention;

[0066] Figure 3 is a module connection diagram of a factory monitoring and safety system integrating multiple systems in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0068] Embodiment 1

[0069] As Figure 1 shown, a factory monitoring and safety method integrating multiple systems includes the following steps:

[0070] Step 1: Pre-collect a subsystem information set, a historical monitoring data set, and a historical anomaly label set;

[0071] Step 2: Based on the subsystem information set and the historical monitoring data set, construct an association matrix calculation model;

[0072] Step 3: From the historical monitoring data set, use the association matrix calculation model to extract the historical association matrix, and extract the abnormal warning feature vector from the historical association matrix;

[0073] Step 4: Based on the abnormal warning feature vector and the historical abnormal label set, construct and train an abnormal warning judgment model;

[0074] Step 5: Use the time series of the historical association matrix to construct and train a matrix prediction model;

[0075] Step 6: During the actual monitoring process, collect real-time monitoring data, obtain the real-time association matrix based on the real-time monitoring data and the association matrix calculation model, obtain the future association matrix based on the real-time association matrix and the matrix prediction model, and obtain the warning judgment for the future based on the future association matrix and the abnormal warning judgment model;

[0076] Among them, the collection method of the subsystem information set is:

[0077] Obtain the basic information of each factory equipment in each subsystem in the factory from the technical specifications of the equipment and sensors as the subsystem information set;

[0078] Specifically, the subsystems vary according to the different products produced by the factory. Exemplarily, the subsystems include, but are not limited to, the production process subsystem, and the factory equipment included includes, but is not limited to, processing equipment such as CNC machine tools, punching machines, milling machines, and lathes, assembly equipment such as automated assembly lines and robotic welding equipment, molding equipment such as injection molding machines and die-casting machines, the material and product conveying subsystem, and the factory equipment included includes, but is not limited to, conveyor belt systems, automated guided vehicles, and lifting equipment; the warehousing subsystem, and the factory equipment included includes, but is not limited to, stackers, automated storage racks, barcode scanners, and RFID systems, etc.; it may also include an energy management subsystem, an environmental control subsystem, a safety monitoring subsystem, a quality control subsystem, and a communication subsystem, etc.;

[0079] For example, a certain production system includes equipment or production lines such as an automated assembly line, data collector 1, central controller, CNC machining center, data collector 2, and injection molding machine, and the topological relationship of each piece of equipment and production line is as Figure 2 shown;

[0080] The subsystem information set includes, but is not limited to, the device information, sensor information (if sensors are included in the subsystem), the network topology between devices, and the operating parameters of each device in each subsystem; specifically, the device information includes, but is not limited to, the device type, model, and function description of each subsystem, and the sensor information includes, but is not limited to, the list of sensors used in each subsystem, the sensor type (such as temperature sensor, pressure sensor), measurement accuracy, and unit, etc.;

[0081] The collection method of the historical monitoring data set is as follows:

[0082] Extract from the factory's monitoring and data acquisition system or directly export from the factory's back-end database the historical monitoring data generated by each subsystem and each device in each subsystem during historical operation;

[0083] Specifically, the historical monitoring data includes, but is not limited to, the monitoring data of each subsystem should contain accurate timestamps, the measurement data of each subsystem during the monitoring period, and the parameter records of all manual operation events of the operator for each device;

[0084] The collection method of the historical anomaly label set is as follows:

[0085] Extract from the operator event log or anomaly report or collect from the factory's event management or reporting system, and obtain the anomaly label data each time an anomaly event occurs to form the historical anomaly label set;

[0086] Specifically, the anomaly label data includes the event type, event cause, time of event occurrence, the subsystem affected by the anomaly event, and the severity of the impact on the subsystem each time an anomaly event occurs; specifically, the severity can be set to harmless, low risk, medium risk, high risk, and shutdown for maintenance, etc., and can be further quantified as digital labels such as 1, 2, 3, 4, 5, etc. for subsequent use by the model;

[0087] Furthermore, the method of constructing the correlation matrix calculation model based on the subsystem information set and the historical monitoring data set is as follows:

[0088] For each subsystem included in the subsystem information set:

[0089] Construct a device correlation matrix for each subsystem according to the measurement data of each device in this subsystem in the historical monitoring data set;

[0090] Extract the main eigenvectors of the device correlation matrices of each subsystem, and construct a subsystem correlation matrix between each subsystem based on the main eigenvectors of each subsystem;

[0091] Specifically, the method for constructing the device association matrix for each subsystem is as follows:

[0092] Set the time window size T and the sliding step size s; preferably, the time window size T can be set to 5 minutes, that is, analyze the correlation between each subsystem within every 5 minutes; the sliding step size s can be set to 1 minute, that is, the time window changes every 1 minute;

[0093] Mark each unit time in the historical monitoring data set as t; the unit time can generally be set to 1 minute;

[0094] Mark the number of each subsystem as i;

[0095] Mark the number of each device in the i-th subsystem as ij;

[0096] For each unit time t, extract the parameter set composed of the measurement data of the ij-th device in each subsystem from unit time t - T to unit time t, and form a data window W(Sij,t); where Sij represents the parameter time series composed of the measurement data of the ij-th device in the i-th subsystem arranged in chronological order;

[0097] Use the correlation coefficient analysis method to calculate the correlation between each device in the i-th subsystem and generate a device association matrix;

[0098] Specifically, the correlation coefficient analysis method includes but is not limited to Pearson, Spearman, or mutual information;

[0099] Exemplarily, this embodiment provides a method for calculating the device association matrix using the Pearson correlation coefficient method here:

[0100] Mark the numbers of the other devices except the ij-th device in the i-th subsystem as ik;

[0101] Mark the association degree between the ij-th device and the jk-th device as Rijk;

[0102] Then the calculation formula for Riijk is:

[0103]

[0104] where sijt' is the t'-th measurement data in the time series of Sij, that is, in the data window corresponding to the time series of Sij, the measurement data of the ij-th device after the t'-th unit time;

[0105] is the average of all measurement data in the time series of Sij;

[0106] Similarly, sikt' is the t'-th measurement data within the time series of Sik, that is, within the data window corresponding to the time series of Sik, it is the measurement data of the ik-th device after the t'-th unit time;

[0107] is the average of all measurement data in the time series of Sik;

[0108] The correlation degrees between devices in the i-th subsystem form a device correlation matrix;

[0109] The value corresponding to the j-th row and k-th column in the device correlation matrix is Rijk;

[0110] Furthermore, the method for extracting the principal eigenvector of the device correlation matrix of each subsystem is:

[0111] Perform singular value decomposition on the device correlation matrix by using the method of singular value decomposition to obtain the principal eigenvector;

[0112] As an example, the way to obtain the principal eigenvector by using the method of singular value decomposition can be:

[0113] Mark the device correlation matrix as A;

[0114] Use the method of singular value decomposition to transform matrix A into A = U∑V T of the form;

[0115] where U and V are both orthogonal matrices, the column vectors of U are called left singular vectors, and each column vector of V is a right singular vector; Σ is the singular value matrix, and the singular value matrix is a diagonal matrix containing singular values arranged in descending order;

[0116] It should be noted that in matrix A, the left singular vectors correspond to the eigenmodes in the original data space. These vectors define the directions of different correlation patterns; while the right singular vectors correspond to the eigenmodes in the column space, usually related to the columns of the original data;

[0117] Select the largest singular value from the singular value matrix, and use the left singular vector corresponding to this largest singular value as the principal eigenvector;

[0118] It should be noted that the principal eigenvector represents a main correlation pattern among devices. This main correlation pattern represents the potential correlation relationship among devices in the subsystem. Each element in the principal eigenvector represents the relative contribution or weight of each device, and the positive or negative sign of each element indicates the consistency or symmetry of the change directions of each device in this pattern; the absolute value size of the element represents the intensity contribution of the device to the main correlation pattern.

[0119] Further, the method for constructing the subsystem correlation matrix between each subsystem based on the main eigenvectors of each subsystem is as follows:

[0120] Mark any two subsystems as i1 and i2 respectively;

[0121] For the main eigenvector of subsystem i1 and the main eigenvector of subsystem i2, use the Pearson correlation coefficient formula to calculate the subsystem correlation degree between subsystem i1 and subsystem i2;

[0122] It can be understood that the calculation method of this subsystem correlation degree is the same as the calculation method of the correlation degree between the above devices, which will not be elaborated here. The only difference is that the input vector of the correlation degree between the above devices is the time series of the measurement data of each device, while the input vector of the subsystem correlation degree is the main eigenvector of each subsystem;

[0123] Construct a subsystem correlation matrix according to the subsystem correlation degrees between each subsystem;

[0124] Further, the method for extracting the historical correlation matrix from the historical monitoring data set using the correlation matrix calculation model is as follows:

[0125] For each time window in the historical monitoring process, calculate the subsystem correlation matrix between each subsystem within each time window through the calculation method of the subsystem correlation matrix, and use it as the historical correlation matrix;

[0126] Then the method for extracting the abnormal warning feature vector from the historical correlation matrix is as follows:

[0127] Use the singular value decomposition method for the historical correlation matrix of each time window to extract the main eigenvector of the historical correlation matrix as the abnormal warning feature vector;

[0128] It can be understood that each element in the abnormal warning feature vector represents the contribution weight of a subsystem to the abnormality of the entire system, thus effectively characterizing the main correlation patterns between each subsystem. This correlation pattern may be an abnormal state;

[0129] Further, the method for constructing and training the abnormal warning judgment model based on the abnormal warning feature vector and the historical abnormal label set is as follows:

[0130] Use any classification model as the abnormal warning judgment model; the classification model includes but is not limited to DNN models, SVM models, etc.;

[0131] The abnormal warning judgment model takes the abnormal warning feature vector as the input and the predicted value of the abnormal type number corresponding to the time window of the abnormal warning feature vector as the output;

[0132] Specifically, the predicted value of the abnormal type number can simultaneously contain the information of the number of the event type and the number of the severity level. For example, if the number of the event type of a short circuit is 0, the event type of the short circuit can be divided into 5 abnormal type numbers, namely 1, 2, 3, 4, and 5, which respectively represent a harmless short circuit, a low-risk short circuit, a medium-risk short circuit, a high-risk short circuit, and a short circuit for shutdown and maintenance; the abnormal type numbers of the remaining event types can be inferred by analogy;

[0133] The abnormal warning judgment model uses the cross-entropy error between the predicted value of the abnormal type number and the actual value of the abnormal type number in the corresponding time window as the first error function;

[0134] Use the abnormal warning feature vectors and the actual values of the abnormal type numbers in all time windows during the historical monitoring process to train the abnormal warning judgment model until the first error function converges;

[0135] It can be understood that through the training of the abnormal warning judgment model, it is possible to judge the type and danger level of abnormal events occurring inside the factory by analyzing the real-time correlation between each subsystem and between each device in the subsystem;

[0136] Further, the method of constructing and training the matrix prediction model using the time series of the historical correlation matrix is as follows:

[0137] Set the matrix prediction model as an LSTM network model; it can be understood that the LSTM model is suitable for processing and predicting data based on time series. It can capture data dependencies with long time spans through memory cells and adapt to dynamic changes and correlations in time series;

[0138] The matrix prediction model takes the sequence of subsystem correlation matrices in the past time window as the input;

[0139] The output layer of the matrix prediction model outputs a prediction matrix for the subsystem correlation matrix in the time window after the next L time steps as the output; L is a preset prediction time step;

[0140] Take the real subsystem correlation matrix after L time steps corresponding to each group of inputs as the prediction target of the prediction matrix, and take the matrix cosine similarity between the real subsystem correlation matrix and the prediction matrix as the second error function;

[0141] Train the matrix prediction model until the second error function converges;

[0142] It can be understood that by training the matrix prediction model, the subsystem correlation matrix after the next L time steps can be predicted through the past subsystem correlation matrix, so as to determine whether there is an abnormal state after the next L time steps through the predicted subsystem correlation matrix;

[0143] Therefore, the real-time monitoring data includes the real-time measurement data collected by all devices in all subsystems;

[0144] The method for obtaining the real-time correlation matrix based on the real-time monitoring data and the correlation matrix calculation model is to construct a real-time subsystem correlation matrix according to the real-time measurement data collected by each device in each subsystem as the real-time correlation matrix;

[0145] The method for obtaining the future correlation matrix based on the real-time correlation matrix and the matrix prediction model is: combining the real-time correlation matrix and the subsystem correlation matrices within several past time windows as the input of the matrix prediction model, and obtaining the predicted matrix after L time steps output by the matrix prediction model as the future correlation matrix;

[0146] The method for obtaining the future warning judgment based on the future correlation matrix and the abnormal warning judgment model is: inputting the future correlation matrix into the abnormal warning judgment model to obtain the predicted value of the abnormal type number output by the abnormal warning judgment model. According to this predicted value, the event type and severity that may occur after the next L time steps can be judged, and thus corresponding warnings can be given in advance according to the event type and severity according to the pre-set warning mode.

[0147] Embodiment 2

[0148] As Figure 3 shown, a factory monitoring and safety system integrating multiple systems includes a historical data collection module, a correlation matrix calculation module, a warning model training module, a matrix prediction module, and a safety warning module; among them, each module is electrically connected;

[0149] The historical data collection module pre-collects a subsystem information set, a historical monitoring data set, and a historical abnormal label set, and sends the subsystem information set and the historical monitoring data set to the correlation matrix calculation module, and sends the historical abnormal label set to the warning model training module;

[0150] The correlation matrix calculation module constructs a correlation matrix calculation model based on the subsystem information set and the historical monitoring data set, and sends the correlation matrix calculation model to the warning model training module;

[0151] The early warning model training module extracts the historical correlation matrix from the historical monitoring data set using the correlation matrix calculation model, extracts the abnormal early warning feature vectors from the historical correlation matrix, constructs and trains an abnormal early warning judgment model based on the abnormal early warning feature vectors and the historical abnormal label set, and sends the historical correlation matrix to the matrix prediction module and the abnormal early warning judgment model to the security early warning module;

[0152] The matrix prediction module constructs and trains a matrix prediction model using the time series of the historical correlation matrix and sends the matrix prediction model to the security early warning module;

[0153] The security early warning module collects real-time monitoring data during the actual monitoring process, obtains the real-time correlation matrix based on the real-time monitoring data and the correlation matrix calculation model, obtains the future correlation matrix based on the real-time correlation matrix and the matrix prediction model, and obtains the early warning judgment for the future based on the future correlation matrix and the abnormal early warning judgment model.

[0154] The methods, devices, and equipment of the present application can be implemented in many ways. For example, the methods, devices, and equipment of the present application can be implemented through software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present application are not limited to the above specific description order unless otherwise specifically stated. In addition, in some embodiments, the present application can also be implemented as a program recorded in a recording medium, and these programs include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers a recording medium storing a program for executing the method according to the present application.

[0155] In addition, parts of the above technical solutions provided in the embodiments of the present application that are consistent with the corresponding technical solutions in the prior art in terms of implementation principles are not described in detail to avoid excessive elaboration.

[0156] As described above in the specific embodiments, the purpose, technical solutions, and beneficial effects of the present invention are further described in detail. It should be understood that the above is only the specific embodiments of the present invention and is not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

[0157] The above preset parameters or preset thresholds are set by those skilled in the art according to the actual situation or obtained through a large number of data simulations.

[0158] The above embodiments are only used to illustrate the technical methods of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical methods of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A factory monitoring and safety method integrating multiple systems, characterized in that: The following steps are involved: Step 1: Collect subsystem information sets, historical monitoring data sets, and historical abnormal label sets in advance; Step 2: Based on the subsystem information set and the historical monitoring data set, a correlation matrix calculation model is constructed; Step 3: From the historical monitoring data set, use the correlation matrix calculation model to extract the historical correlation matrix, and extract the abnormal warning feature vector from the historical correlation matrix; Step 4: Build and train an anomaly warning judgment model based on the anomaly warning feature vector and the historical anomaly label set; Step 5: Use the time series of the historical correlation matrix to build and train the matrix prediction model; Step 6: In the actual monitoring process, real-time monitoring data is collected, and a real-time correlation matrix is ​​obtained based on the real-time monitoring data and the correlation matrix calculation model. Based on the real-time correlation matrix and the matrix prediction model, a future correlation matrix is ​​obtained. Based on the future correlation matrix and the abnormal warning judgment model, a warning judgment for the future is obtained.

2. A factory monitoring and safety method integrating multiple systems according to claim 1, characterized in that: The subsystem information set is collected in the following manner: Obtain the basic information of each factory equipment in each subsystem as a subsystem information set; The historical monitoring data set is collected in the following manner: Extract from the factory's monitoring and data acquisition system or from the factory's backend database, export the historical monitoring data generated by each subsystem and each device in each subsystem during the historical operation process; The collection method of the historical abnormal label set is: Extracted from operator event logs or exception reports or collected from the factory's event management or reporting system, exception tag data is obtained each time an abnormal event occurs to form a historical exception tag collection.

3. A factory monitoring and safety method integrating multiple systems according to claim 2, characterized in that: The method of constructing the association matrix calculation model based on the subsystem information set and the historical monitoring data set is: For each subsystem contained in the subsystem information collection: According to the measurement data of each device of the subsystem in the historical monitoring data set, a device association matrix is ​​constructed for each subsystem; The main eigenvectors of the device association matrix of each subsystem are extracted, and the subsystem association matrix between each subsystem is constructed based on the main eigenvectors of each subsystem.

4. A factory monitoring and safety method integrating multiple systems according to claim 3, characterized in that: The method of constructing the device association matrix for each subsystem is as follows: Set the time window size T and sliding step size s; In the historical monitoring data set, each unit time is marked as t; Label each subsystem as i; Mark the number of each device in the i-th subsystem as ij; For each unit time t, extract the parameter set composed of the measurement data of the ijth device in each subsystem from the unit time tT to the unit time t, and form a data window W(Sij,t); where Sij represents the parameter time series composed of the measurement data of the ijth device in the i-th subsystem arranged in time order; The correlation coefficient analysis method is used to calculate the correlation between each device in the i-th subsystem and generate a device correlation matrix.

5. A factory monitoring and safety method integrating multiple systems according to claim 4, characterized in that: The method of extracting the main eigenvector of the device association matrix of each subsystem is: The device association matrix is ​​subjected to singular value decomposition by using the singular value decomposition method to obtain the main eigenvector.

6. A factory monitoring and safety method integrating multiple systems according to claim 5, characterized in that: The method of using singular value decomposition to obtain the main eigenvector is: Label the device association matrix as A; Use the singular value decomposition method to transform the matrix A into A = U∑V T form; Wherein, U and V are both orthogonal matrices, the column vector of U is called a left singular vector, and each column vector of V is a right singular vector; Σ is a singular value matrix, which is a diagonal matrix containing singular values ​​arranged in descending order; The largest singular value is selected from the singular value matrix, and the left singular vector corresponding to the largest singular value is taken as the main eigenvector.

7. A factory monitoring and safety method integrating multiple systems according to claim 6, characterized in that: The method of constructing the subsystem association matrix between each subsystem based on the main eigenvector of each subsystem is: Label any two subsystems as i1 and i2; For the main eigenvector of subsystem i1 and the main eigenvector of subsystem i2, the Pearson correlation coefficient formula is used to calculate the subsystem correlation between subsystem i1 and subsystem i2; The subsystem correlation matrix is ​​constructed according to the subsystem correlation between each subsystem.

8. A factory monitoring and safety method integrating multiple systems according to claim 7, characterized in that: The method of constructing and training the abnormal warning judgment model based on the abnormal warning feature vector and the historical abnormal label set is: Use any classification model as an abnormal warning judgment model; The abnormal warning judgment model takes the abnormal warning feature vector as input and takes the predicted value of the abnormal type number of the time window corresponding to the abnormal warning feature vector as output; The abnormal warning judgment model uses the cross entropy error between the predicted value of the abnormal type number and the actual value of the abnormal type number in the corresponding time window as the first error function; The abnormal warning judgment model is trained using the actual values ​​of the abnormal warning feature vectors and abnormal type numbers in all time windows in the historical monitoring process until the first error function reaches convergence.

9. A factory monitoring and safety method integrating multiple systems according to claim 8, characterized in that: The method of constructing and training the matrix prediction model using the time series of the historical correlation matrix is ​​as follows: Set the matrix prediction model to LSTM network model; The matrix prediction model takes as input a sequence of subsystem association matrices within a past time window; The output layer of the matrix prediction model outputs the prediction matrix of the subsystem association matrix in the time window after L time steps in the future as output; L is the preset prediction time step; The true subsystem association matrix after L time steps corresponding to each group of inputs is used as the prediction target of the prediction matrix, and the matrix cosine similarity between the true subsystem association matrix and the prediction matrix is ​​used as the second error function; The matrix prediction model is trained until the second error function reaches convergence.

10. A multi-system integrated factory monitoring and safety system, which is used to implement the multi-system integrated factory monitoring and safety method described in any one of claims 1 to 9, characterized in that: It includes a historical data collection module, a correlation matrix calculation module, a warning model training module, a matrix prediction module and a safety warning module; wherein each module is electrically connected; The historical data collection module collects subsystem information sets, historical monitoring data sets and historical abnormal label sets in advance, and sends the subsystem information sets and historical monitoring data sets to the association matrix calculation module, and sends the historical abnormal label sets to the early warning model training module; The correlation matrix calculation module constructs a correlation matrix calculation model based on the subsystem information set and the historical monitoring data set, and sends the correlation matrix calculation model to the early warning model training module; The early warning model training module extracts the historical association matrix from the historical monitoring data set using the association matrix calculation model, and extracts the abnormal warning feature vector from the historical association matrix. Based on the abnormal warning feature vector and the historical abnormal label set, the abnormal warning judgment model is constructed and trained, and the historical association matrix is ​​sent to the matrix prediction module, and the abnormal warning judgment model is sent to the safety warning module; The matrix prediction module uses the time series of the historical correlation matrix to build and train the matrix prediction model, and sends the matrix prediction model to the security warning module; The security warning module collects real-time monitoring data during the actual monitoring process, obtains a real-time correlation matrix based on the real-time monitoring data and the correlation matrix calculation model, obtains a future correlation matrix based on the real-time correlation matrix and the matrix prediction model, and obtains a future warning judgment based on the future correlation matrix and the abnormal warning judgment model.

Images