Intelligent industrial site safety monitoring method and system

Abstract

The invention discloses an intelligent industrial field safety monitoring method and system. The method comprises the steps of data acquisition, data preprocessing, industrial field safety dual monitoring, model hyper-parameter optimization and industrial field safety comprehensive monitoring. The invention belongs to the technical field of industrial field safety monitoring, particularly relates to an intelligent industrial field safety monitoring method and system, and creatively provides an industrial field safety dual monitoring method so as to realize comprehensive prediction of potential safety risks. An internal structure activation function and a double-branch structure of a time sequence layer are innovatively designed, so that accurate identification of potential long-term risks is enhanced, the monitoring capability of concurrent security events is improved, and the accuracy and stability of a model prediction result are improved; by updating the inertia weight and obtaining the particle speed control parameter and the global optimal position variation strategy, the optimal hyper-parameter combination of the model is obtained, the output accuracy of the model is improved, and the safe and accurate monitoring of the industrial site is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial site safety monitoring, and specifically relates to an intelligent industrial site safety monitoring method and system. Background Art

[0002] With the rapid development of industrial automation and informatization, the complexity and diversity of industrial production sites are increasing. Traditional safety monitoring methods usually rely on manual inspections and simple sensor equipment. These methods have certain limitations in terms of monitoring coverage, response speed and data processing capabilities. In recent years, the rapid development of technologies such as the Internet of Things, big data and artificial intelligence has promoted the trend of intelligent safety monitoring. These technologies can realize real-time data collection and intelligent analysis, significantly improving the efficiency and accuracy of industrial site safety monitoring. However, there are technical problems in traditional industrial site safety monitoring methods that mainly monitor a single industrial safety risk and ignore concurrent safety risks; there are technical problems in the existing event hazard prediction model, such as insufficient concurrent event monitoring, difficulty in capturing long-term changes and potential risks of safety events, and instability in the face of complex and abnormal data, which leads to inaccurate model output results; there are technical problems in the traditional prediction model that the parameters are improperly set, and the parameter optimization algorithm has a weak ability to obtain the global optimal solution, which leads to inaccurate final prediction results. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an intelligent industrial site safety monitoring method and system. In view of the technical problem that the traditional industrial site safety monitoring method mainly monitors a single industrial safety risk and ignores the concurrent safety risks, this solution creatively proposes a method for dual monitoring of industrial site safety, which combines the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites. It can simultaneously identify a single safety hazard and the chain of concurrent safety risks caused by it, thereby achieving a comprehensive prediction of potential safety risks, solving the technical problem of the shortcomings of traditional methods in concurrent event monitoring, which not only improves the accuracy of risk prediction, but also reduces redundant calculations, and significantly enhances the real-time, accuracy and reliability of industrial site safety monitoring; in view of the technical problems of insufficient concurrent event monitoring, difficulty in capturing long-term changes and potential risks of safety events, and instability in the face of complex and abnormal data in the existing event hazard prediction model, which leads to inaccurate model output results, this solution innovatively designs the internal structure activation of the time series layer Function and dual-branch structure, the internal structure activation function of the time series layer can effectively identify and respond to sudden changes in complex safety data, solving the instability problem of traditional methods when processing abnormal data. The global safety dynamic extraction branch of the industrial site overcomes the shortcomings of the traditional method in long-term safety change monitoring by comprehensively capturing time changes and long-term dependencies, and enhances the accurate identification of potential long-term risks. The industrial site safety key event extraction branch accurately identifies concurrent risk events and avoids interference with irrelevant data, effectively improving the monitoring ability of concurrent safety events and significantly enhancing the accuracy and stability of the model prediction results; in view of the technical problems of improper parameter settings in traditional prediction models and weak global optimal solution acquisition capabilities of parameter optimization algorithms, which lead to inaccurate final prediction results, this scheme updates the inertia weight, obtains particle velocity control parameters and global optimal position variation strategies, enhances the global optimal solution search and improves the robustness of the algorithm, thereby obtaining the optimal hyperparameter combination of the model, improving the output accuracy of the model, and realizing accurate monitoring of industrial site safety.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides an intelligent industrial site safety monitoring method, which comprises the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Dual monitoring of industrial site safety;

[0008] Step S4: model hyperparameter optimization;

[0009] Step S5: Comprehensive monitoring of industrial site safety.

[0010] Furthermore, in step S1, the data collection specifically obtains the original data of industrial site safety monitoring from the industrial site through data collection; the original data of industrial site safety monitoring includes historical safety monitoring data and real-time safety monitoring data; the historical safety monitoring data and real-time safety monitoring data both include industrial site environment data, industrial site equipment data, industrial site external environment data and equipment operator behavior data; the historical safety monitoring data also includes historical safety event record data.

[0011] Further, in step S2, the data preprocessing is specifically used to perform data cleaning, data filtering, standardization, data completion and feature selection on the original data of industrial site safety monitoring to obtain preliminary data of industrial site safety monitoring; the data cleaning is used to eliminate invalid and inaccurate data, specifically by processing data missing values, data outliers and data duplicate values; the data filtering is used to remove high-frequency noise and interference signals in the data, specifically by smoothing the data using a filtering algorithm; the standardization is based on the maximum and minimum normalization method to standardize the data; the data completion is used to complete missing and uncollected data, specifically by using interpolation technology to fill in missing data points; the feature selection is to use correlation analysis methods to screen out features related to industrial site safety.

[0012] Further, in step S3, the industrial site safety dual monitoring includes industrial site safety single hazard monitoring, building a correlation event hazard prediction model, training the correlation event hazard prediction model, obtaining concurrent industrial site potential safety risks and obtaining industrial site safety dual monitoring results; specifically includes the following steps:

[0013] Step S31: Industrial site safety single hazard monitoring, used to obtain potential industrial site safety risks according to simple safety standard thresholds, specifically, according to the safety requirements and standards of the industrial site, the safety monitoring data of the real-time safety monitoring data is compared with the preset safety threshold to obtain a single industrial site potential safety risk, wherein the safety monitoring data is specifically the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data; comprising the following steps:

[0014] Step S311: setting safety standards and thresholds, specifically, setting the normal working range and safety threshold of each safety monitoring parameter according to the specific safety requirements of the industrial site, the characteristics of the production environment and the relevant national and industry safety standards; the safety threshold is set according to different safety requirements and working conditions;

[0015] Step S312: safety threshold comparison, specifically comparing each safety monitoring data with the set safety threshold, and determining whether each data point exceeds the preset safety threshold;

[0016] Step S313: obtaining a single industrial site potential safety risk, specifically by comparing the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data with the set safety threshold to obtain the single industrial site potential safety risk;

[0017] Step S32: constructing a correlation event hazard prediction model to predict concurrent safety accidents caused by a single potential safety risk and obtain concurrent industrial site potential safety risks, specifically including the following steps:

[0018] Step S321: constructing a time series layer, specifically including the following steps:

[0019] Step S3211: Design the internal structure activation function of the time series layer, specifically designing the data increase activation function and the data balance adjustment activation function. The formula used is as follows:

[0020] ;

[0021] ;

[0022] In the formula, represents the data augmentation activation function, represents the data balance adjustment activation function, represents the input variable of the data augmentation activation function, Represents the input variable of the data balance adjustment activation function; Indicates control The size of the part, Indicates the magnitude of the latter control. Indicates adjustment Partial sensitivity to input, represents the adjustment of the latter's sensitivity to input, represents the magnitude of the input variable in the numerator, It indicates the influence of the input variable in the control denominator;

[0023] Step S3212: Design the internal structure of the time series layer , the formula used is as follows:

[0024] ;

[0025] In the formula, represents the output of the forget gate, It represents the input data of the time series layer. Indicates the hidden state at the last moment. and They represent the weight matrices corresponding to the forget gate processing input data and processing hidden states, respectively. Represents the bias parameter of the forget gate, i t represents the output of the input gate, and They represent the weight matrices corresponding to the input gate processing input data and processing hidden states, respectively, and b i represents the bias parameter of the input gate, represents the candidate value of the new cell state, and Respectively represent the weight matrices corresponding to the cell state processing input data and processing hidden state, The bias parameter representing the cell state, It represents the cell state at the current moment. It indicates the cell state at the last moment. represents the output of the output gate, and They represent the weight matrices corresponding to the output gate processing input data and processing hidden states, respectively. represents the bias parameter of the output gate, Indicates the hidden state at the current moment;

[0026] Step S322: Establish an industrial site global safety dynamic extraction branch to analyze the development trajectory of a single potential safety risk and its impact on other potential safety events in the industrial site from the overall time series. Specifically, the time-distributed fully connected layer is used to enhance the data features of each time step, and then the Capture the long-term dependencies of the sequence and further process it through two fully connected layers to finally generate time step sequence features ;

[0027] Step S323: Establish an industrial site safety critical event extraction branch to analyze the dangerous trigger points of related events. Extract preliminary sequence features, then introduce the attention mechanism to calculate the attention weight of each time step, generate weighted features based on the weights, and finally generate compact global features through the fully connected layer ;

[0028] Step S324: Branch feature concatenation, the formula used is as follows:

[0029] ;

[0030] In the formula, Represents the comprehensive features after splicing;

[0031] Step S325: Establish the model output layer, the formula used is as follows:

[0032] ;

[0033] In the formula, It indicates the result of the risk prediction of the associated event. represents the weight matrix of the model output layer, Represents the bias parameter of the model output layer, Represents the Sigmoid activation function;

[0034] Step S33: training the associated event hazard prediction model, specifically using the historical safety monitoring data in the preliminary industrial site safety monitoring data to train the associated event hazard prediction model to obtain a trained associated event hazard prediction model;

[0035] Step S34: Acquire potential safety risks of concurrent industrial sites, specifically, use the real-time safety monitoring data and the potential safety risks of a single industrial site in the preliminary safety monitoring data of the industrial site as input data of the trained associated event hazard prediction model to obtain the potential safety risks of concurrent industrial sites;

[0036] Step S35: Obtain the industrial site safety dual monitoring results, specifically combining the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites to obtain the industrial site safety dual monitoring results.

[0037] Further, in step S4, the model hyperparameter optimization is specifically to obtain the optimal hyperparameter combination of the associated event risk prediction model by improving the particle swarm optimization algorithm, including the following steps:

[0038] Step S41: Initialize parameters, specifically by constructing initial algorithm parameters; the initial algorithm parameters include the number of particles N and the maximum number of iterations ;

[0039] Step S42: Initializing the particle swarm, specifically randomly generating particle positions, wherein the particle positions represent a combination of model hyperparameters;

[0040] Step S43: Calculate the fitness value, specifically calculate the fitness value f of the particles in the particle swarm i , the performance of the associated event hazard prediction model based on the individual location is used as the individual's fitness value;

[0041] Step S44: Update the inertia weight, specifically adjust the inertia weight according to the current number of iterations, and the formula used is as follows:

[0042] ;

[0043] In the formula, represents the inertia weight of the t-th iteration, represents the maximum value of the inertia weight, represents the minimum value of inertia weight, and t represents the current number of iterations;

[0044] Step S45: Obtaining particle speed control parameters, specifically, adaptively adjusting the particle speed control parameters to balance global search and local convergence capabilities. The formula used is as follows:

[0045] ;

[0046] In the formula, represents the particle velocity control parameter, represents the individual learning factor, which is used to control the speed at which particles move to the individual optimal position. represents the group learning factor, which is used to control the speed at which particles move to the global optimal position;

[0047] Step S46: Update the particle velocity using the following formula:

[0048] ;

[0049] In the formula, represents the velocity of the i-th particle in the t+1th iteration, represents the velocity of the ith particle in the tth iteration, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of individual particles, represents the global optimal position of the particle, and Represents a random number in the range [0,1];

[0050] Step S47: Update the particle position. The formula used is as follows:

[0051] ;

[0052] In the formula, represents the position of the i-th particle in the t+1-th iteration;

[0053] Step S48: Update the local optimal position of the individual particle, using the following formula:

[0054] ;

[0055] In the formula, represents the updated local optimal position of the individual particle, represents the Cauchy probability distribution, Represents the fitness value calculation function;

[0056] Step S49: Update the global optimal position of the particle, specifically compare the fitness values ​​of the local optimal positions of all particles, find the particle with the best objective function value, and update it to the global optimal position. If the global optimal position is not updated in 3 iterations, the global optimal position is mutated. The formula used is as follows:

[0057] ;

[0058] In the formula, Represents the global optimal position after mutation;

[0059] Step S410: search and determination, specifically, by constructing a search termination condition, searching and determining the global optimal position of the particle, and obtaining the data setting of the global optimal position of the particle;

[0060] The search termination conditions include threshold termination and iteration termination;

[0061] The threshold termination is specifically to set a fitness threshold, when the particle fitness value f i When it is above the fitness threshold, the search is completed;

[0062] The iteration termination specifically refers to terminating the iteration and obtaining the global optimal position of the particle when the maximum number of iterations is reached;

[0063] The global optimal position of the particle specifically refers to the optimal hyperparameter combination of the model.

[0064] Furthermore, in step S5, the comprehensive industrial site safety monitoring is used to comprehensively monitor and manage industrial site safety; specifically, based on the dual industrial site safety monitoring results, intelligent comprehensive industrial site safety monitoring is achieved to ensure the stability of industrial production and the safety of personnel.

[0065] The technical solution adopted by the present invention is as follows: an intelligent industrial site safety monitoring system provided by the present invention includes a data acquisition module, a data preprocessing module, an industrial site safety dual monitoring module, a model hyperparameter optimization module and an industrial site safety comprehensive monitoring module;

[0066] The data acquisition module acquires raw data of industrial site safety monitoring from the industrial site through acquisition, and sends the raw data of industrial site safety monitoring to the data preprocessing module;

[0067] The data preprocessing module receives the data sent by the data acquisition module, and performs data cleaning, data filtering, standardization processing, data completion and feature selection on the industrial site safety monitoring raw data to obtain preliminary industrial site safety monitoring data, and sends the preliminary industrial site safety monitoring data to the industrial site safety dual monitoring module;

[0068] The industrial site safety dual monitoring module receives the data sent by the data preprocessing module, and realizes the single danger monitoring of industrial site safety through the safety standard threshold judgment, and finally uses the real-time safety monitoring data and the single industrial site potential safety risk as the input data of the trained model to obtain multiple industrial site potential safety risks, and combines the single industrial site potential safety risk with the concurrent industrial site potential safety risk to obtain the industrial site safety dual monitoring result, and sends the industrial site safety dual monitoring result to the industrial site safety comprehensive monitoring module;

[0069] The model hyperparameter optimization module obtains the optimal hyperparameter combination of the associated event hazard prediction model by improving the particle swarm optimization algorithm, and sends the optimal hyperparameter combination of the model to the industrial site safety dual monitoring module;

[0070] The industrial site safety comprehensive monitoring module receives the results sent by the industrial site safety dual monitoring module, and implements intelligent industrial site safety comprehensive monitoring based on the industrial site safety dual monitoring results.

[0071] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0072] (1) In view of the technical problem that traditional industrial site safety monitoring methods mainly monitor a single industrial safety risk and ignore concurrent safety risks, this scheme creatively proposes a method of dual monitoring of industrial site safety, which combines the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites. It can simultaneously identify single safety hazards and the chain of concurrent safety risks caused by them, thereby achieving a comprehensive prediction of potential safety risks. It solves the technical problem of the shortcomings of traditional methods in concurrent event monitoring, not only improves the accuracy of risk prediction, but also reduces redundant calculations, significantly enhancing the real-time, precision and reliability of industrial site safety monitoring.

[0073] (2) In order to address the technical problems of insufficient concurrent event monitoring, difficulty in capturing long-term changes and potential risks of safety events, and instability in the face of complex and abnormal data in existing event hazard prediction models, which lead to inaccurate model output results, this solution innovatively designs the internal structure activation function and dual-branch structure of the time series layer. The internal structure activation function of the time series layer can effectively identify and respond to sudden changes in complex safety data, solving the instability problem of traditional methods when processing abnormal data. The industrial site global safety dynamic extraction branch overcomes the shortcomings of traditional methods in long-term safety change monitoring by comprehensively capturing time changes and long-term dependencies, and enhances the accurate identification of potential long-term risks. The industrial site safety key event extraction branch accurately identifies concurrent risk events and avoids interference with irrelevant data, effectively improving the monitoring capability of concurrent safety events and significantly enhancing the accuracy and stability of the model prediction results.

[0074] (3) In order to solve the technical problems of improper parameter settings in traditional prediction models and weak ability of parameter optimization algorithms to obtain global optimal solutions, which lead to inaccurate final prediction results, this scheme updates the inertia weight, obtains particle velocity control parameters and global optimal position variation strategy, enhances the global optimal solution search and improves the robustness of the algorithm, thereby obtaining the optimal hyperparameter combination of the model, improving the output accuracy of the model and achieving accurate monitoring of industrial site safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 A schematic diagram of a process flow of an intelligent industrial site safety monitoring method provided by the present invention;

[0076] Figure 2 A module schematic diagram of an intelligent industrial site safety monitoring system provided by the present invention;

[0077] Figure 3 This is a flow chart of step S3 industrial site safety dual monitoring;

[0078] Figure 4 Schematic diagram of the process of optimizing model hyperparameters in step S4;

[0079] Figure 5 A schematic diagram of the process of constructing a correlation event risk prediction model in step S32;

[0080] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0081] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0082] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0083] Example 1, see Figure 1 The present invention provides an intelligent industrial site safety monitoring method, which comprises the following steps:

[0084] Step S1: data collection;

[0085] Step S2: data preprocessing;

[0086] Step S3: Dual monitoring of industrial site safety;

[0087] Step S4: model hyperparameter optimization;

[0088] Step S5: Comprehensive monitoring of industrial site safety.

[0089] By performing the above operations, in order to address the technical problem that traditional industrial site safety monitoring methods mainly monitor a single industrial safety risk and ignore concurrent safety risks, this solution creatively proposes a method of dual monitoring of industrial site safety. It combines the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites, and can simultaneously identify single safety hazards and the chain concurrent safety risks caused by them, thereby achieving a comprehensive prediction of potential safety risks. It solves the technical problem of the shortcomings of traditional methods in concurrent event monitoring, which not only improves the accuracy of risk prediction, but also reduces redundant calculations, significantly enhancing the real-time, accuracy and reliability of industrial site safety monitoring.

[0090] Example 2, see Figure 1 and Figure 2This embodiment is based on the above embodiment. In step S1, the data collection is specifically to obtain the original data of industrial site safety monitoring from the industrial site through data collection; the original data of industrial site safety monitoring includes historical safety monitoring data and real-time safety monitoring data; the historical safety monitoring data and real-time safety monitoring data both include industrial site environment data, industrial site equipment data, industrial site external environment data and equipment operator behavior data; the historical safety monitoring data also includes historical safety event record data; the industrial site environment data includes field temperature data, field humidity data and field gas concentration data; the industrial site equipment data includes field equipment operation data, field pressure equipment pressure data and field equipment vibration data; the industrial site external environment data includes field external wind speed data and field precipitation data; the equipment operator behavior data includes operator working status, operator construction operation procedures and operator dangerous emergency response mode; the historical safety event record data includes safety event category, safety event cause, safety event basic information, safety event response measures and safety event related data; the safety event related data includes concurrent safety event record data and the causal relationship between safety events.

[0091] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S2, the data preprocessing is specifically used to perform data cleaning, data filtering, standardization processing, data completion and feature selection on the original data of industrial site safety monitoring to obtain preliminary data of industrial site safety monitoring; the data cleaning is used to eliminate invalid and inaccurate data, specifically by processing data missing values, data outliers and data duplicate values; the data filtering is used to remove high-frequency noise and interference signals in the data, specifically by using a filtering algorithm to smooth the data; the standardization processing is based on the maximum and minimum normalization method to standardize the data; the data completion is used to complete the missing and uncollected data, specifically by using interpolation technology to fill the missing data points; the feature selection is to use the correlation analysis method to screen out features related to industrial site safety.

[0092] Example 4, see Figure 1 , Figure 2 , Figure 3 and Figure 5 This embodiment is based on the above embodiment. In step S3, the dual monitoring of industrial site safety includes monitoring of industrial site safety single hazard, building a correlation event hazard prediction model, training the correlation event hazard prediction model, obtaining concurrent industrial site potential safety risks, and obtaining the dual monitoring results of industrial site safety; including the following steps:

[0093] Step S31: Industrial site safety single hazard monitoring, used to obtain potential industrial site safety risks according to simple safety standard thresholds, specifically, according to the safety requirements and standards of the industrial site, the safety monitoring data of the real-time safety monitoring data is compared with the preset safety threshold to obtain a single industrial site potential safety risk, wherein the safety monitoring data is specifically the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data; comprising the following steps:

[0094] Step S311: setting safety standards and thresholds, specifically, setting the normal working range and safety threshold of each safety monitoring parameter according to the specific safety requirements of the industrial site, the characteristics of the production environment and the relevant national and industry safety standards; the safety threshold is set according to different safety requirements and working conditions;

[0095] Step S312: safety threshold comparison, specifically comparing each safety monitoring data with the set safety threshold, and determining whether each data point exceeds the preset safety threshold;

[0096] Step S313: obtaining a single industrial site potential safety risk, specifically by comparing the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data with the set safety threshold to obtain the single industrial site potential safety risk;

[0097] Step S32: constructing a correlation event hazard prediction model to predict concurrent safety accidents caused by a single potential safety risk and obtain concurrent industrial site potential safety risks, specifically including the following steps:

[0098] Step S321: constructing a time series layer, specifically including the following steps:

[0099] Step S3211: Design the internal structure activation function of the time series layer, specifically designing the data increase activation function and the data balance adjustment activation function. The formula used is as follows:

[0100] ;

[0101] ;

[0102] In the formula, represents the data augmentation activation function, represents the data balance adjustment activation function, represents the input variable of the data augmentation activation function, Represents the input variable of the data balance adjustment activation function; Indicates control The size of the part, Indicates the magnitude of the latter control. Indicates adjustment Partial sensitivity to input, represents the adjustment of the latter's sensitivity to input, represents the magnitude of the input variable in the numerator, It indicates the influence of the input variable in the control denominator;

[0103] Step S3212: Design the internal structure of the time series layer , the formula used is as follows:

[0104] ;

[0105] In the formula, represents the output of the forget gate, It represents the input data of the time series layer. Indicates the hidden state at the last moment. and They represent the weight matrices corresponding to the forget gate processing input data and processing hidden states, respectively. Represents the bias parameter of the forget gate, i t represents the output of the input gate, and They represent the weight matrices corresponding to the input gate processing input data and processing hidden states, respectively, and b i represents the bias parameter of the input gate, represents the candidate value of the new cell state, and Respectively represent the weight matrices corresponding to the cell state processing input data and processing hidden state, The bias parameter representing the cell state, Indicates the current cell state. It indicates the cell state at the last moment. represents the output of the output gate, and They represent the weight matrices corresponding to the output gate processing input data and processing hidden states, respectively. represents the bias parameter of the output gate, Indicates the hidden state at the current moment;

[0106] Step S322: Establish an industrial site global safety dynamic extraction branch to analyze the development trajectory of a single potential safety risk and its impact on other potential safety events in the industrial site from the overall time series. Specifically, the time-distributed fully connected layer is used to enhance the data features of each time step, and then the Capture the long-term dependencies of the sequence and further process it through two fully connected layers to finally generate time step sequence features ; Specifically include the following steps:

[0107] Step S3221: Time-distributed fully connected layer, used to enhance the data features of each time step, the formula used is as follows:

[0108] ;

[0109] In the formula, represents the output feature of the tth time step of the time-distributed fully connected layer, represents the Sigmoid activation function, represents the weight matrix of the time-distributed fully connected layer, represents the input data of the time-distributed fully connected layer, Represents the bias parameter of the time-distributed fully connected layer;

[0110] Step S3222: Time series layer, used to capture the long-term dependency of the sequence, the formula used is as follows:

[0111] ;

[0112] In the formula, Represents the output features of the t-th time step of the time series layer;

[0113] Step S3223: Two fully connected layers to generate time step sequence features , the formula used is as follows:

[0114] ;

[0115] ;

[0116] In the formula, represents the output features of the tth time step of the first fully connected layer, represents the ReLU activation function, represents the weight matrix of the first fully connected layer, represents the bias parameter of the first fully connected layer, represents the weight matrix of the second fully connected layer, represents the bias parameter of the second fully connected layer, Representing the global safety dynamics of industrial sites, extracting branches to generate time step sequence features;

[0117] Step S323: Establish an industrial site safety critical event extraction branch to analyze the dangerous trigger points of related events. Extract preliminary sequence features, then introduce the attention mechanism to calculate the attention weight of each time step, generate weighted features based on the weights, and finally generate compact global features through the fully connected layer ; Specifically include the following steps:

[0118] Step S3231: Extract the key event branch time series layer, which is used to extract preliminary sequence features. The formula used is as follows:

[0119] ;

[0120] In the formula, represents the output feature of the t-th time step of the time series layer in the industrial site safety critical event extraction branch, Represents the input data of the time series layer in the industrial site safety critical event extraction branch;

[0121] Step S3232: Attention mechanism layer, used to calculate the attention weight of each time step, the formula used is as follows:

[0122] ;

[0123] In the formula, represents the weighted eigenvector, represents the total number of time steps, Represents the weight matrix of the attention layer;

[0124] Step S3233: Key event extraction branch fully connected layer, the formula used is as follows:

[0125] ;

[0126] In the formula, The branch for extracting critical safety events in industrial sites generates compact global features. represents the weight matrix of the fully connected layer of the key event extraction branch, represents the bias parameter of the fully connected layer of the key event extraction branch, Indicates flattening the weighted feature vector into a one-dimensional vector;

[0127] Step S324: Branch feature concatenation, the formula used is as follows:

[0128] ;

[0129] In the formula, Represents the comprehensive features after splicing;

[0130] Step S325: Establish the model output layer, the formula used is as follows:

[0131] ;

[0132] In the formula, It indicates the result of the risk prediction of the associated event. represents the weight matrix of the model output layer, Represents the bias parameter of the model output layer, Represents the Sigmoid activation function;

[0133] Step S33: training the associated event hazard prediction model, specifically using the historical safety monitoring data in the preliminary industrial site safety monitoring data to train the associated event hazard prediction model to obtain a trained associated event hazard prediction model;

[0134] Step S34: Acquire potential safety risks of concurrent industrial sites, specifically, use the real-time safety monitoring data in the preliminary safety monitoring data of the industrial site and the potential safety risks of a single industrial site as input data of the trained associated event hazard prediction model to obtain potential safety risks of concurrent industrial sites, wherein the potential safety risks of concurrent industrial sites include safety event probabilities and safety event categories;

[0135] Step S35: Obtain the industrial site safety dual monitoring results, specifically combining the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites to obtain the industrial site safety dual monitoring results.

[0136] By performing the above operations, in order to solve the technical problems of insufficient concurrent event monitoring, difficulty in capturing long-term changes and potential risks of safety events, and instability in the face of complex and abnormal data in the existing event hazard prediction models, which leads to inaccurate model output results, this solution innovatively designs the internal structure activation function and dual-branch structure of the time series layer. The internal structure activation function of the time series layer can effectively identify and respond to sudden changes in complex safety data, and solves the instability problem of traditional methods when processing abnormal data. The industrial site global safety dynamic extraction branch overcomes the shortcomings of traditional methods in long-term safety change monitoring by comprehensively capturing time changes and long-term dependencies, and enhances the accurate identification of potential long-term risks. The industrial site safety key event extraction branch avoids interference with irrelevant data by accurately identifying concurrent risk events, effectively improves the monitoring capability of concurrent safety events, and significantly enhances the accuracy and stability of the model prediction results.

[0137] Example 5, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S4, the model hyperparameter optimization is specifically to obtain the optimal hyperparameter combination of the associated event risk prediction model by improving the particle swarm optimization algorithm, including the following steps:

[0138] Step S41: Initialize parameters, specifically by constructing initial algorithm parameters; the initial algorithm parameters include the number of particles N and the maximum number of iterations ;

[0139] Step S42: Initializing the particle swarm, specifically randomly generating particle positions, wherein the particle positions represent a combination of model hyperparameters;

[0140] Step S43: Calculate the fitness value, specifically calculate the fitness value f of the particles in the particle swarm i , the performance of the associated event hazard prediction model based on the individual location is used as the individual's fitness value;

[0141] Step S44: Update the inertia weight, specifically adjust the inertia weight according to the current number of iterations, and the formula used is as follows:

[0142] ;

[0143] In the formula, represents the inertia weight of the t-th iteration, represents the maximum value of the inertia weight, represents the minimum value of inertia weight, and t represents the current number of iterations;

[0144] Step S45: Obtaining particle speed control parameters, specifically, adaptively adjusting the particle speed control parameters to balance global search and local convergence capabilities. The formula used is as follows:

[0145] ;

[0146] In the formula, represents the particle velocity control parameter, represents the individual learning factor, which is used to control the speed at which particles move to the individual optimal position. represents the group learning factor, which is used to control the speed at which particles move to the global optimal position;

[0147] Step S46: Update the particle velocity using the following formula:

[0148] ;

[0149] In the formula, represents the velocity of the i-th particle in the t+1th iteration, represents the velocity of the ith particle in the tth iteration, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of individual particles, represents the global optimal position of the particle, and Represents a random number in the range [0,1];

[0150] Step S47: Update the particle position. The formula used is as follows:

[0151] ;

[0152] In the formula, represents the position of the i-th particle in the t+1-th iteration;

[0153] Step S48: Update the local optimal position of the individual particle, using the following formula:

[0154] ;

[0155] In the formula, represents the updated local optimal position of the individual particle, represents the Cauchy probability distribution, Represents the fitness value calculation function;

[0156] Step S49: Update the global optimal position of the particle, specifically compare the fitness values ​​of the local optimal positions of all particles, find the particle with the best objective function value, and update it to the global optimal position. If the global optimal position is not updated in 3 iterations, the global optimal position is mutated. The formula used is as follows:

[0157] ;

[0158] In the formula, Represents the global optimal position after mutation;

[0159] Step S410: search and determination, specifically, by constructing a search termination condition, searching and determining the global optimal position of the particle, and obtaining the data setting of the global optimal position of the particle;

[0160] The search termination conditions include threshold termination and iteration termination;

[0161] The threshold termination is specifically to set a fitness threshold, when the particle fitness value f i When it is above the fitness threshold, the search is completed;

[0162] The iteration termination specifically refers to terminating the iteration and obtaining the global optimal position of the particle when the maximum number of iterations is reached;

[0163] The global optimal position of the particle specifically refers to the optimal hyperparameter combination of the model.

[0164] By performing the above operations, in order to solve the technical problems of improper parameter settings in traditional prediction models and weak ability of parameter optimization algorithms to obtain global optimal solutions, which lead to inaccurate final prediction results, this solution updates the inertia weight, obtains particle velocity control parameters and global optimal position variation strategy, enhances the global optimal solution search and improves the robustness of the algorithm, thereby obtaining the optimal hyperparameter combination of the model, improving the output accuracy of the model and achieving accurate monitoring of industrial site safety.

[0165] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the comprehensive industrial site safety monitoring is used to comprehensively monitor and manage the industrial site safety. Specifically, based on the dual monitoring results of the industrial site safety, intelligent comprehensive industrial site safety monitoring is realized to ensure the stability of industrial production and the safety of personnel.

[0166] Embodiment 7, see Figure 1 and Figure 2 , This embodiment is based on the above embodiment, and the technical solution adopted by the present invention is as follows: The technical solution adopted by the present invention is as follows: The present invention provides an intelligent industrial site safety monitoring system, including a data acquisition module, a data preprocessing module, an industrial site safety dual monitoring module, a model hyperparameter optimization module and an industrial site safety comprehensive monitoring module;

[0167] The data acquisition module acquires raw data of industrial site safety monitoring from the industrial site through acquisition, and sends the raw data of industrial site safety monitoring to the data preprocessing module;

[0168] The data preprocessing module receives the data sent by the data acquisition module, and performs data cleaning, data filtering, standardization processing, data completion and feature selection on the industrial site safety monitoring raw data to obtain preliminary industrial site safety monitoring data, and sends the preliminary industrial site safety monitoring data to the industrial site safety dual monitoring module;

[0169] The industrial site safety dual monitoring module receives the data sent by the data preprocessing module, and realizes the single danger monitoring of industrial site safety through the safety standard threshold judgment, and finally uses the real-time safety monitoring data and the single industrial site potential safety risk as the input data of the trained model to obtain multiple industrial site potential safety risks, and combines the single industrial site potential safety risk with the concurrent industrial site potential safety risk to obtain the industrial site safety dual monitoring result, and sends the industrial site safety dual monitoring result to the industrial site safety comprehensive monitoring module;

[0170] The model hyperparameter optimization module obtains the optimal hyperparameter combination of the associated event hazard prediction model by improving the particle swarm optimization algorithm, and sends the optimal hyperparameter combination of the model to the industrial site safety dual monitoring module;

[0171] The industrial site safety comprehensive monitoring module receives the results sent by the industrial site safety dual monitoring module, and implements intelligent industrial site safety comprehensive monitoring based on the industrial site safety dual monitoring results.

[0172] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0173] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0174] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. An intelligent industrial site safety monitoring method, characterized in that: The method comprises the following steps: Step S1: data collection, specifically, obtaining the original data of industrial site safety monitoring from the industrial site through data collection; Step S2: data preprocessing, specifically for data cleaning, data filtering, standardization, data completion and feature selection of the original data of industrial site safety monitoring to obtain preliminary data of industrial site safety monitoring; Step S3: Dual monitoring of industrial site safety, used to obtain the results of dual monitoring of industrial site safety, specifically identifying the potential safety risks of a single industrial site through safety standards and threshold design, designing the internal structure activation function of the time series layer, improving the internal structure of the time series layer, establishing a dual-branch structure including the time series layer, constructing an associated event hazard prediction model through the dual-branch structure and the model output layer, and performing model training, using real-time data and a single industrial site potential safety risk as input data of the trained model, obtaining concurrent industrial site potential safety risks, combining the single industrial site potential safety risks with the concurrent industrial site potential safety risks, and obtaining the results of dual monitoring of industrial site safety; comprising the following steps: Step S31: monitoring single safety hazard of industrial site; Step S32: constructing a risk prediction model for associated events; Step S33: training the risk prediction model for associated events; Step S34: obtaining potential safety risks of concurrent industrial sites; Step S35: obtaining dual safety monitoring results of industrial sites; Step S4: model hyperparameter optimization, specifically, obtaining the optimal hyperparameter combination of the correlation event hazard prediction model by updating the inertia weight, obtaining the particle speed control parameters and improving the particle swarm optimization algorithm through the global optimal position mutation strategy; Step S5: Comprehensive industrial site safety monitoring: Based on the industrial site safety dual monitoring results, intelligent comprehensive industrial site safety monitoring is achieved.

2. The intelligent industrial site safety monitoring method according to claim 1 is characterized in that: In step S3, the industrial site safety dual monitoring specifically includes the following steps: Step S31: Industrial site safety single hazard monitoring, used to obtain potential industrial site safety risks according to simple safety standard thresholds, specifically, according to the safety requirements and standards of the industrial site, the safety monitoring data of the real-time safety monitoring data is compared with the preset safety threshold to obtain a single industrial site potential safety risk, wherein the safety monitoring data is specifically the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data; comprising the following steps: Step S311: setting safety standards and thresholds, specifically, setting the normal working range and safety threshold of each safety monitoring parameter according to the specific safety requirements of the industrial site, the characteristics of the production environment and the relevant national and industry safety standards; the safety threshold is set according to different safety requirements and working conditions; Step S312: safety threshold comparison, specifically comparing each safety monitoring data with the set safety threshold, and determining whether each data point exceeds the preset safety threshold; Step S313: obtaining a single industrial site potential safety risk, specifically by comparing the industrial site environment data and the industrial site equipment data in the preliminary industrial site safety monitoring data with the set safety threshold to obtain the single industrial site potential safety risk; Step S32: constructing a correlation event hazard prediction model to predict concurrent safety accidents caused by a single potential safety risk and obtain concurrent industrial site potential safety risks, specifically including the following steps: Step S321: constructing a time series layer, specifically including the following steps: Step S3211: Design the internal structure activation function of the time series layer, specifically designing the data increase activation function and the data balance adjustment activation function. The formula used is as follows: ； ； In the formula, represents the data augmentation activation function, represents the data balance adjustment activation function, represents the input variable of the data augmentation activation function, Represents the input variable of the data balance adjustment activation function; Indicates control The size of the part, Indicates the magnitude of the latter control. Indicates adjustment Partial sensitivity to input, represents the adjustment of the latter's sensitivity to input, represents the magnitude of the input variable in the numerator, It indicates the influence of the input variable in the control denominator; Step S3212: Design the internal structure of the time series layer , the formula used is as follows: ； In the formula, represents the output of the forget gate, It represents the input data of the time series layer. Indicates the hidden state at the last moment. and They represent the weight matrices corresponding to the forget gate processing input data and processing hidden states, respectively. Represents the bias parameter of the forget gate, i t represents the output of the input gate, and They represent the weight matrices corresponding to the input gate processing input data and processing hidden states, respectively, and b i represents the bias parameter of the input gate, represents the candidate value of the new cell state, and Respectively represent the weight matrices corresponding to the cell state processing input data and processing hidden state, The bias parameter representing the cell state, Indicates the current cell state. It indicates the cell state at the last moment. represents the output of the output gate, and They represent the weight matrices corresponding to the output gate processing input data and processing hidden states, respectively. represents the bias parameter of the output gate, Indicates the hidden state at the current moment; Step S322: Establish an industrial site global safety dynamic extraction branch to analyze the development trajectory of a single potential safety risk and its impact on other potential safety events in the industrial site from the overall time series. Specifically, the time-distributed fully connected layer is used to enhance the data features of each time step, and then the Capture the long-term dependencies of the sequence and further process it through two fully connected layers to finally generate time step sequence features ; Step S323: Establish an industrial site safety critical event extraction branch to analyze the dangerous trigger points of related events. Extract preliminary sequence features, then introduce the attention mechanism to calculate the attention weight of each time step, generate weighted features based on the weights, and finally generate compact global features through the fully connected layer ; Step S324: Branch feature concatenation, the formula used is as follows: ； In the formula, Represents the comprehensive features after splicing; Step S325: Establish the model output layer, the formula used is as follows: ； In the formula, It indicates the result of the risk prediction of the associated event. represents the weight matrix of the model output layer, Represents the bias parameter of the model output layer, Represents the Sigmoid activation function; Step S33: training the associated event hazard prediction model, specifically using the historical safety monitoring data in the preliminary industrial site safety monitoring data to train the associated event hazard prediction model to obtain a trained associated event hazard prediction model; Step S34: Acquire potential safety risks of concurrent industrial sites, specifically, use the real-time safety monitoring data and the potential safety risks of a single industrial site in the preliminary safety monitoring data of the industrial site as input data of the trained associated event hazard prediction model to obtain the potential safety risks of concurrent industrial sites; Step S35: Obtain the industrial site safety dual monitoring results, specifically combining the potential safety risks of a single industrial site with the potential safety risks of concurrent industrial sites to obtain the industrial site safety dual monitoring results.

3. The intelligent industrial site safety monitoring method according to claim 1 is characterized in that: In step S4, the model hyperparameter optimization is specifically to obtain the optimal hyperparameter combination of the associated event risk prediction model by improving the particle swarm optimization algorithm, including the following steps: Step S41: Initialize parameters, specifically by constructing initial algorithm parameters; the initial algorithm parameters include the number of particles N and the maximum number of iterations ; Step S42: Initializing the particle swarm, specifically randomly generating particle positions, wherein the particle positions represent a combination of model hyperparameters; Step S43: Calculate the fitness value, specifically calculate the fitness value f of the particles in the particle swarm i , the performance of the associated event hazard prediction model based on the individual location is used as the individual's fitness value; Step S44: Update the inertia weight, specifically adjust the inertia weight according to the current number of iterations, and the formula used is as follows: ； In the formula, represents the inertia weight of the t-th iteration, represents the maximum value of the inertia weight, represents the minimum value of inertia weight, and t represents the current number of iterations; Step S45: Obtaining particle speed control parameters, specifically, adaptively adjusting the particle speed control parameters to balance global search and local convergence capabilities. The formula used is as follows: ； In the formula, represents the particle velocity control parameter, represents the individual learning factor, which is used to control the speed at which particles move to the individual optimal position. represents the group learning factor, which is used to control the speed at which particles move to the global optimal position; Step S46: Update the particle velocity using the following formula: ； In the formula, represents the velocity of the i-th particle in the t+1th iteration, represents the velocity of the ith particle in the tth iteration, represents the position of the i-th particle in the t-th iteration, represents the local optimal position of individual particles, represents the global optimal position of the particle, and Represents a random number in the range [0,1]; Step S47: Update the particle position. The formula used is as follows: ； In the formula, represents the position of the i-th particle in the t+1-th iteration; Step S48: Update the local optimal position of the individual particle, using the following formula: ； In the formula, represents the updated local optimal position of the individual particle, represents the Cauchy probability distribution, Represents the fitness value calculation function; Step S49: Update the global optimal position of the particle, specifically compare the fitness values ​​of the local optimal positions of all particles, find the particle with the best objective function value, and update it to the global optimal position. If the global optimal position is not updated in 3 iterations, the global optimal position is mutated. The formula used is as follows: ； In the formula, Represents the global optimal position after mutation; Step S410: search and determination, specifically, by constructing a search termination condition, searching and determining the global optimal position of the particle, and obtaining the data setting of the global optimal position of the particle; The search termination conditions include threshold termination and iteration termination; The threshold termination is specifically to set a fitness threshold, when the particle fitness value f i When it is above the fitness threshold, the search is completed; The iteration termination specifically refers to terminating the iteration and obtaining the global optimal position of the particle when the maximum number of iterations is reached; The global optimal position of the particle specifically refers to the optimal hyperparameter combination of the model.

4. The intelligent industrial site safety monitoring method according to claim 1 is characterized in that: In step S5, the comprehensive industrial site safety monitoring is used to comprehensively monitor and manage the industrial site safety; specifically, based on the industrial site safety dual monitoring results, intelligent comprehensive industrial site safety monitoring is achieved to ensure the stability of industrial production and the safety of personnel.

5. The intelligent industrial site safety monitoring method according to claim 1 is characterized in that: In step S1, the data collection specifically involves obtaining the original industrial site safety monitoring data from the industrial site through data collection; the original industrial site safety monitoring data includes historical safety monitoring data and real-time safety monitoring data; both the historical safety monitoring data and the real-time safety monitoring data include industrial site environment data, industrial site equipment data, industrial site external environment data and equipment operator behavior data; the historical safety monitoring data also includes historical safety event record data.

6. The intelligent industrial site safety monitoring method according to claim 1 is characterized in that: In step S2, the data preprocessing is specifically used to perform data cleaning, data filtering, standardization, data completion and feature selection on the original data of industrial site safety monitoring to obtain preliminary data of industrial site safety monitoring; the data cleaning is used to eliminate invalid and inaccurate data, specifically by processing missing data values, data outliers and data duplication values; the data filtering is used to remove high-frequency noise and interference signals in the data, specifically by smoothing the data using a filtering algorithm; the standardization is based on the maximum and minimum normalization method to standardize the data; the data completion is used to complete missing and uncollected data, specifically by using interpolation technology to fill in missing data points; the feature selection is to use correlation analysis methods to screen out features related to industrial site safety.

7. An intelligent industrial site safety monitoring system, used to implement an intelligent industrial site safety monitoring method as claimed in any one of claims 1 to 6, characterized in that: It includes a data acquisition module, a data preprocessing module, an industrial site safety dual monitoring module, a model hyperparameter optimization module and an industrial site safety comprehensive monitoring module.

8. The intelligent industrial site safety monitoring system according to claim 7 is characterized in that: The data acquisition module acquires raw data of industrial site safety monitoring from the industrial site through acquisition, and sends the raw data of industrial site safety monitoring to the data preprocessing module; The data preprocessing module receives the data sent by the data acquisition module, and performs data cleaning, data filtering, standardization processing, data completion and feature selection on the industrial site safety monitoring raw data to obtain preliminary industrial site safety monitoring data, and sends the preliminary industrial site safety monitoring data to the industrial site safety dual monitoring module; The industrial site safety dual monitoring module receives the data sent by the data preprocessing module, and realizes the single danger monitoring of industrial site safety through the safety standard threshold judgment, and finally uses the real-time safety monitoring data and the single industrial site potential safety risk as the input data of the trained model to obtain multiple industrial site potential safety risks, and combines the single industrial site potential safety risk with the concurrent industrial site potential safety risk to obtain the industrial site safety dual monitoring result, and sends the industrial site safety dual monitoring result to the industrial site safety comprehensive monitoring module; The model hyperparameter optimization module obtains the optimal hyperparameter combination of the associated event hazard prediction model by improving the particle swarm optimization algorithm, and sends the optimal hyperparameter combination of the model to the industrial site safety dual monitoring module; The industrial site safety comprehensive monitoring module receives the results sent by the industrial site safety dual monitoring module, and implements intelligent industrial site safety comprehensive monitoring based on the industrial site safety dual monitoring results.

Images