Safety risk partition management method and system for safety supervision of chemical enterprise

Abstract

The invention discloses a safety risk partition management method and system for chemical enterprise safety supervision, and relates to the technical field of chemical enterprise management, and the method comprises the following steps: carrying out the feature extraction of safety monitoring data, and obtaining the safety performance feature data of each region; key risk factors influencing the safety performance of each area are obtained; establishing a risk prediction model based on the key risk factors, and predicting the safety performance trend of each region at the future moment by using the risk prediction model; and based on the safety performance trend, evaluating the safety performance state of each region, and taking safety management measures according to the evaluation result. According to the method, the safety performance characteristic data is analyzed by using a risk analysis algorithm, key risk factors influencing the safety performance of each area can be identified, an enterprise can be helped to quickly start emergency response, loss after an accident is reduced, the accident rate is effectively reduced, and the safe production environment of the enterprise is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical enterprise management. Specifically, it relates to a safety risk zoning management method and system for safety supervision of chemical enterprises. Background Art

[0002] During the daily production process of chemical enterprises, a large number of chemical substances are stored and used. These substances have high risks, such as flammable, explosive, and toxic characteristics. Therefore, the safety production work of chemical enterprises must be highly emphasized, and a sound safety management system and preventive measures should be established to ensure the safety of personnel, equipment, and the environment during the production process. In order to effectively control and reduce safety risks, chemical enterprises can adopt the method of safety risk zoning management to accurately identify and zone manage the safety risks of different regions, and conduct targeted supervision.

[0003] However, in the prior art, it is not convenient to analyze the safety performance characteristic data, the key risk factors affecting the safety performance of each region cannot be identified, and it cannot help enterprises more accurately grasp the specific safety challenges faced by different regions, thus reducing the prevention and control of high-risk regions. Moreover, it is not convenient to identify potential risk dynamics in advance to provide a prediction basis for safety decision-making. At the same time, when an enterprise discovers potential safety risks in some regions through safety performance trend assessment, it is not convenient to formulate emergency response measures in advance, making it inconvenient for enterprises to implement targeted safety management measures, thereby increasing the probability of accidents.

[0004] In response to the problems in the related art, no effective solution has been proposed yet. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention proposes a safety risk zoning management method and system for safety supervision of chemical enterprises, which solves the problems mentioned in the above background art, namely, it is not convenient to analyze the safety performance characteristic data, the key risk factors affecting the safety performance of each region cannot be identified, and it cannot help enterprises more accurately grasp the specific safety challenges faced by different regions, thus reducing the prevention and control of high-risk regions. Moreover, it is not convenient to identify potential risk dynamics in advance to provide a prediction basis for safety decision-making. At the same time, when an enterprise discovers potential safety risks in some regions through safety performance trend assessment, it is not convenient to formulate emergency response measures in advance, making it inconvenient for enterprises to implement targeted safety management measures, thereby increasing the probability of accidents.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: According to one aspect of the present invention, a safety risk zoning management method for safety supervision of chemical enterprises is provided. The method includes the following steps: S1. Obtain the safety monitoring data of each area in the chemical enterprise, and perform feature extraction on the safety monitoring data to obtain the safety performance feature data of each area; S2. Use the risk analysis algorithm to analyze the safety performance feature data to obtain the key risk factors affecting the safety performance of each area; S3. Based on the key risk factors, establish a risk prediction model and use the risk prediction model to predict the safety performance trend of each area at future times; S4. Based on the safety performance trend, evaluate the safety performance status of each area, and according to the evaluation results, take safety management measures.

[0007] Furthermore, obtaining the safety monitoring data of each area in the chemical enterprise and performing feature extraction on the safety monitoring data to obtain the safety performance feature data of each area includes the following steps: S11. Collect the safety monitoring data of each area in the chemical enterprise and calculate the key feature indicators related to area safety using the safety performance equation; S12. Use the data processing method to process the safety monitoring data of each area and extract the safety performance signals of each area; S13. Use the generalized demodulation algorithm to demodulate the extracted safety performance signals, and perform frequency analysis on the demodulated safety performance signals to extract the safety performance spectrum; S14. Combine the peak features in the safety performance spectrum and other safety performance indicators to extract the safety performance feature data related to the safety performance of each area.

[0008] Furthermore, using the risk analysis algorithm to analyze the safety performance feature data to obtain the key risk factors affecting the safety performance of each area includes the following steps: S21. Standardize the safety performance feature data of each area collected; S22. Based on the entropy weight method, allocate the weights of each index for the safety performance feature data of each area, and calculate the safety risk index of each area; S23. Based on the safety risk index of each area, use the clustering algorithm to perform a hierarchical analysis of the safety risks of each area to obtain areas with different risk levels; S24. Use the risk analysis algorithm to perform a correlation analysis on the areas with different risk levels to identify the key risk factors affecting the safety performance of each area.

[0009] Furthermore, based on the entropy weight method, allocating the weights of each index for the safety performance feature data of each area and calculating the safety risk index of each area includes the following steps: S221. Analyze the safety performance characteristics of each region based on the expert scoring method and the eigenvector method to determine the preliminary weights of each criterion layer. S222. For each criterion layer, calculate the weight information entropy of each indicator within each criterion through the entropy weight method. S223. Use the weight information entropy to correct the weight vector between criterion layers and normalize the corrected weights. S224. Combine the safety performance characteristic data of each region, and based on the weights corrected by the weight information entropy, perform weighted summation and calculate the safety risk index of each region through the weighted score.

[0010] Furthermore, the formula for calculating the weight information entropy of each indicator within each criterion is: ; In the formula, X(t) represents the weight information entropy of the t-th criterion layer; ω(t, j) represents the weight of the j-th safety performance indicator in the t-th criterion layer; n(t) represents the total number of safety performance indicators in the t-th criterion layer.

[0011] Furthermore, the formula for calculating the safety risk index of each region through the weighted score is: ; In the formula, S i represents the safety risk index of the i-th region; ω j represents the weight of the j-th safety performance indicator; x ij represents the standardized value of the i-th region on the j-th safety performance indicator; n represents the total number of safety performance indicators.

[0012] Furthermore, using the risk analysis algorithm to conduct correlation analysis on regions with different risk levels and identify the key risk factors affecting the safety performance of each region includes the following steps: S231. Input the sampling inspection data set containing the safety performance and corresponding risk levels of each region, and set the minimum support and minimum confidence as the thresholds for correlation analysis. S232. Analyze the safety performance data set of each region, search for safety risk factors whose occurrence frequency is greater than or equal to the minimum support, and form a frequent risk factor set, which is sorted in descending order according to their support. S233. Based on the frequent risk factor set, gradually establish branches for the header of the frequent pattern tree, and sort each branch in descending order according to the support of the frequent items, and gradually insert and construct the frequent pattern tree. S234. Starting from the bottom of the frequent pattern tree, gradually traverse each node and recursively generate all combinations of frequent risk factors that meet the minimum support requirement. S235. Calculate the confidence levels among frequently occurring risk factors, retain the association rules with confidence levels greater than the minimum confidence level threshold, and determine the safety performance risk levels for each region. S236. Store the association rules between the selected frequently occurring risk factors and their corresponding safety performance risk levels, and identify the key risk factors affecting the safety performance of each region.

[0013] Furthermore, based on the key risk factors, establish a risk prediction model, and use the risk prediction model to predict the safety performance trends of each region at future times, including the following steps: S31. Input the set of key risk factors identified as affecting the safety performance of the region, decompose the input set of key risk factors, and extract several intrinsic mode functions. S32. Divide each intrinsic mode function into a training set and a test set according to the time series, and perform standardization processing. S33. Build a risk prediction model based on a neural network, optimize the key parameters of the risk prediction model, and use the optimized risk prediction model to predict the test set, and generate the safety performance trends of each region at future times.

[0014] Furthermore, based on a neural network, build a risk prediction model, optimize the key parameters of the risk prediction model, and use the optimized risk prediction model to predict the test set, and generate the safety performance trends of each region at future times, including the following steps: S331. Set the hyperparameter search range of the neural network model and the feature subset affecting the model performance. S332. Randomly initialize the parameter configuration and feature subset of the neural network, and construct a search space. S333. According to the initialized parameter configuration and feature subset of the neural network, construct and train the neural network model. S334. According to the fitness value of the risk prediction model, select several candidate neural network model configurations with the best performance. S335. For the remaining candidate neural network model configurations, perform random search, and calculate the fitness value of the new neural network model configuration. S336. Repeat steps S334 to S335, continue with multiple rounds of optimization until the maximum number of iterations is reached, select the neural network configuration with the best performance as the optimized risk prediction model; and use the optimized risk prediction model to predict the test set, and output the safety performance trends of each region at future times.

[0015] According to another aspect of the present invention, there is also provided a safety risk zoning management system for chemical enterprise safety supervision, and the system includes: A data acquisition module, which is used to acquire the safety monitoring data of each area in a chemical enterprise, extract features from the safety monitoring data, and obtain the safety performance feature data of each area; A risk analysis module, which is used to analyze the safety performance feature data by using a risk analysis algorithm to obtain the key risk factors affecting the safety performance of each area; A model prediction module, which is used to establish a risk prediction model based on the key risk factors and use the risk prediction model to predict the safety performance trends of each area at future times; A safety management module, which is used to evaluate the safety performance status of each area based on the safety performance trends, and take safety management measures according to the evaluation results.

[0016] The beneficial effects of the present invention are as follows: 1. By acquiring the safety monitoring data of each area in a chemical enterprise and extracting features, the present invention can understand the safety performance of different areas in real time. By analyzing the safety performance feature data by using a risk analysis algorithm, the key risk factors affecting the safety performance of each area can be identified, which can help the enterprise more accurately grasp the specific safety challenges faced by different areas. By establishing a risk prediction model and predicting the safety performance trends of each area at future times, potential risk dynamics can be identified in advance, so as to provide a prediction basis for safety decision-making. When potential safety risks are found in some areas through the evaluation of safety performance trends, the enterprise can formulate emergency response measures in advance based on the evaluation results, which can help the enterprise quickly initiate an emergency response, reduce the losses after an accident, and provide a safer working environment for employees, thereby effectively reducing the accident rate and ensuring the safe production environment of the enterprise.

[0017] 2. Through correlation analysis by using a risk analysis algorithm, the present invention can identify the key risk factors affecting the safety performance of an area from a large amount of data. The identification of these key factors helps the chemical enterprise to early warn potential safety problems. By constructing a frequent pattern tree and association rules, the mutual relationships between different risk factors can be found, so as to provide forward-looking risk prediction for the enterprise. By calculating the safety risk index of each area and combining the key risk factors, the main factors affecting safety can be quickly determined in case of an anomaly, and a clear basis for safety decision-making can be provided, enabling the enterprise to promptly initiate an emergency response mechanism to avoid the spread of risks. By accurately dividing the safety management requirements of areas with different risk levels, the chemical enterprise can more reasonably allocate safety resources.

[0018] 3. The construction and optimization of the risk prediction model based on key risk factors in the present invention provide strong technical support for the safety management of chemical enterprises. Through methods such as data standardization, intrinsic mode extraction, and neural network optimization, it can accurately identify and predict future safety performance trends, and discover potential risks in advance. This not only improves the safety management efficiency of chemical enterprises, but also strengthens the emergency response ability, reduces the possibility of accidents, and ensures the safe operation of the enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 is a flowchart of a safety risk zoning management method for safety supervision of chemical enterprises according to an embodiment of the present invention; Figure 2 is a schematic block diagram of a safety risk zoning management system for safety supervision of chemical enterprises according to an embodiment of the present invention.

[0021] In the figure: 1. Data acquisition module; 2. Risk analysis module; 3. Safety management module; 4. Safety management module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments.

[0023] In the description of the present invention, unless otherwise specified, "a plurality of" means two or more. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0024] According to an embodiment of the present invention, a safety risk zoning management method and system for safety supervision of chemical enterprises are provided.

[0025] Now, the present invention will be further described in combination with the drawings and specific embodiments. As Figure 1 shown, the safety risk zoning management method for safety supervision of chemical enterprises according to an embodiment of the present invention includes the following steps: S1. Obtain the safety monitoring data of each area in the chemical enterprise, and perform feature extraction on the safety monitoring data to obtain the safety performance feature data of each area; Specifically, safety monitoring data is data collected through various monitoring devices, sensors, and manual observations that reflects the safety status of various areas in a chemical enterprise, including: 1) Gas monitoring data For example, the concentration data of flammable and explosive gases (such as methane, ammonia, chlorine, etc.).

[0026] Data source: Real-time monitoring of gas concentration through gas detectors and sensors.

[0027] 2) Temperature and humidity data Record the temperature and humidity conditions in each area. Especially in areas where chemicals and flammable items are stored, changes in temperature and humidity may directly affect safety.

[0028] Data source: Temperature and humidity sensors, environmental monitoring systems.

[0029] 3) Pressure monitoring data Involves pressure data of pressure vessels, pipelines, etc., used to determine whether the system is within the safe operating range.

[0030] Data source: Pressure sensors, pressure gauges.

[0031] 4) Flow rate monitoring data Monitor the flow rate and volume of liquids or gases in pipelines. Abnormal flow rate changes may be signals of leaks or equipment failures.

[0032] Data source: Flow meters, flow sensors.

[0033] 5) Equipment status data Includes the operating status (on / off, fault alarm, over-temperature, etc.) of key equipment (such as pumps, compressors, valves, etc.).

[0034] Data source: Equipment monitoring systems, alarm systems.

[0035] 6) Safety monitoring video data Obtain real-time video streams of each area through video monitoring systems, used to monitor safety hazards (such as fires, leaks, personnel behavior, etc.).

[0036] Data source: Closed-circuit television (CCTV) systems, infrared thermal imagers.

[0037] 7) Accident / abnormal event records Include records of historical accidents and abnormal events, such as equipment failures, worker injuries, fires, or chemical leaks, etc.

[0038] Data source: Event management systems, record systems.

[0039] Specifically, safety performance characteristic data includes: Hazardous gas concentration level: Calculate the change trend of gas concentration, such as the maximum concentration, average concentration, concentration fluctuation range of a certain harmful gas, etc.

[0040] Temperature and humidity fluctuation range: Calculate the fluctuation range of temperature and humidity in the area and determine whether it is within the safe range. Severe fluctuations in temperature and humidity may cause equipment damage or abnormal reactions.

[0041] Equipment failure rate: Measure the frequency of equipment failures. Equipment with a higher failure frequency may be a potential source of safety hazards.

[0042] Abnormal pressure value: Monitor and calculate the data of pressure sensors in the area to determine whether there is excessive pressure. Ultra-high or ultra-low pressure may both be dangerous signals.

[0043] Leakage occurrence frequency: Calculate the occurrence frequency of leakage events through monitoring abnormal event records in the data, abnormal changes in flow rate, etc.

[0044] Analysis of personnel safety behavior: Analyze the personnel behavior data obtained through surveillance videos to determine whether there are behaviors that may cause safety problems, such as violation of operating regulations, personnel gathering, etc.

[0045] S2. Use risk analysis algorithms to analyze the safety performance characteristic data to obtain the key risk factors affecting the safety performance of each area; Specifically, the key risk factors include: 1) Hazardous gas leakage: Leakage of flammable, explosive, and toxic gases (such as methane, ammonia, chlorine, hydrogen, etc.).

[0046] 2) Abnormal temperature: Abnormal fluctuations in high or low temperatures, especially temperature changes during storage or reaction processes.

[0047] 3) Pressure fluctuation: Pressure fluctuations in equipment or pipelines, especially in important equipment such as pressure vessels and reactors.

[0048] 4) Equipment failure or malfunction: Failure or malfunction of key equipment (such as pumps, compressors, valves, etc.).

[0049] 5) Chemical leakage or spillage: Leakage, spillage of chemicals, or solvent dissolution reactions.

[0050] 6) Failure of safety equipment: Failure of safety equipment (such as fire alarm systems, automatic fire extinguishing systems, pressure relief valves, etc.).

[0051] 7) Operational errors or improper personnel behavior: Employee operational errors, violation of operating procedures, improper behavior (such as ignoring safety operating procedures, not wearing appropriate safety equipment, etc.).

[0052] 8) Environmental factors (such as wind speed, humidity, etc.): Extreme weather conditions or environmental factors, such as high humidity, strong wind, rainfall, etc.

[0053] 9) Abnormal production load: Abnormal changes in load during the production process (such as too high or too low load).

[0054] 10) Unstable chemical reactions: Unstable factors such as temperature, pressure, and chemical substance ratio during the reaction process.

[0055] 11) Aging or corroded equipment: Equipment corrodes or ages due to long-term use or external environmental influences.

[0056] S3. Based on the key risk factors, establish a risk prediction model, and use the risk prediction model to predict the safety performance trends of each region at future times; S4. Based on the safety performance trends, evaluate the safety performance status of each region, and according to the evaluation results, take safety management measures.

[0057] It should be noted that, first of all, ensure that the safety performance trend data of each region has been obtained and processed. These data usually come from monitoring systems, sensors, and historical data analysis.

[0058] Data source: Safety monitoring data such as gas concentration, temperature and humidity, equipment failure, pressure, flow rate, historical accident records, etc.

[0059] Data processing: Clean and standardize these data, remove noise data, and ensure data quality.

[0060] According to the collected safety monitoring data, analyze the safety performance trends of each region. Through time series analysis of the data, the change trends of the safety performance of each region can be depicted. According to the obtained safety performance trends, comprehensively evaluate the safety performance of each region. The following methods can be used for evaluation: Evaluation criteria: Normal state: The safety performance indicators are stable and within the safe range.

[0061] Warning state: The safety performance indicators are close to the threshold, and risks may occur.

[0062] Dangerous state: The safety performance indicators have exceeded the safety threshold, and there are relatively large risks.

[0063] Evaluation method: Classify the safety performance status of each region based on the set threshold, and divide it into multiple levels, such as "normal", "warning", and "dangerous".

[0064] Comprehensive assessment: By integrating multiple safety performance indicators such as gas concentration, equipment status, temperature and humidity, etc., the overall safety performance score of the area is obtained. According to the evaluation results, corresponding safety management measures are taken for each area. Different safety performance states correspond to different management measures: Normal state: If the safety performance of the area is in a normal state, the existing safety measures and monitoring methods shall be maintained to ensure continuous data monitoring and regular inspections.

[0065] Measures: Maintain regular safety inspections and equipment maintenance.

[0066] Continue to implement the existing safety management measures and preventive measures.

[0067] Alert state: If the safety performance of a certain area is close to the danger threshold, indicating that there are certain risks in the area, preventive and enhanced monitoring measures should be taken immediately.

[0068] Measures: Increase the safety monitoring frequency of this area and use more sensors for monitoring.

[0069] Conduct additional inspections on relevant equipment, especially equipment prone to failure.

[0070] Increase the safety training of staff to ensure that all employees understand the operating procedures for dealing with potential risks.

[0071] Set up an emergency alarm system in the area to ensure timely response in case of abnormalities.

[0072] Danger state: If the safety performance of the area has exceeded the safety threshold, emergency response measures must be taken to minimize the risk of accidents.

[0073] Measures: Activate the emergency plan, immediately stop the relevant production activities, and control the potential hazards in the area.

[0074] Carry out area isolation and evacuate unnecessary personnel.

[0075] Activate the emergency response mechanism and organize a professional team to handle or prevent accidents.

[0076] Check all equipment and operation procedures to ensure there are no further safety hazards.

[0077] Analyze and summarize all relevant data and resume normal production as soon as possible. After the implementation of safety management measures, the safety performance of the area must be continuously monitored, and the safety management measures should be continuously adjusted according to the new monitoring data and trend feedback: Monitoring: Continuously collect the safety monitoring data of each area to ensure that all safety performance indicators are within the safe range.

[0078] Feedback mechanism: Regularly evaluate the effectiveness of safety management measures to ensure that the measures taken can effectively address potential safety risks. If the measures are not effective, adjust and optimize them in a timely manner.

[0079] Dynamic adjustment: Dynamically adjust the safety performance of the area based on continuous monitoring and feedback information. For example, if the alert status persists and the risk intensifies, it may be necessary to upgrade to a dangerous state and take more stringent measures. Finally, to ensure compliance and continuous improvement of safety management, all evaluations and safety management measures should be carefully documented and reported regularly: Report content: The safety performance trends and evaluation results of each area.

[0080] Specific safety measures taken for alert and dangerous states.

[0081] Emergency response records, incident handling situations, and preventive measures.

[0082] Purpose of the report: To help management understand the safety status of the enterprise, timely adjust the overall safety management strategy, and provide data support for future risk assessments.

[0083] In this alternative embodiment, obtaining the safety monitoring data of each area in a chemical enterprise and performing feature extraction on the safety monitoring data to obtain the safety performance feature data of each area includes the following steps: S11. Collect the safety monitoring data of each area in the chemical enterprise and calculate the key feature indicators related to area safety using the safety performance equation; S12. Process the safety monitoring data of each area using a data processing method to extract the safety performance signals of each area; S13. Demodulate the extracted safety performance signals using a generalized demodulation algorithm, perform frequency analysis on the demodulated safety performance signals, and extract the safety performance spectrum; S14. Combine the peak features in the safety performance spectrum and other safety performance indicators to extract the safety performance feature data related to the safety performance of each area.

[0084] It needs to be explained that, first of all, the enterprise collects safety monitoring data of each area through the installed safety monitoring equipment (such as gas sensors, temperature and humidity sensors, pressure sensors, equipment status monitoring, etc.). These monitoring data may include: gas concentration, equipment temperature, pressure value, flow data, equipment health status, etc.; Based on the monitoring data, the preset safety performance equation is used to calculate the safety indicators of each area. The safety performance equation is usually based on experience or existing research results, and reflects the safety of the area by weighted calculation of the combination of different monitoring data. For example, changes in gas concentration and fluctuations in equipment pressure may become the main factors affecting safety performance. These calculated key characteristic indicators can help identify potential safety risk points; pre-process the collected safety monitoring data of each area. This may include removing noise data, processing missing values, and data standardization. By processing the data, ensure that its quality meets the requirements of subsequent analysis; pre-process the collected safety monitoring data of each area. This may include removing noise data, processing missing values, and data standardization. By processing the data, ensure that its quality meets the requirements of subsequent analysis; demodulate the extracted safety performance signal using a generalized demodulation algorithm. Demodulation is the process of extracting feature information from complex signals. Through this algorithm, periodic or non-periodic safety change patterns can be identified from time series data; next, the demodulated safety performance signal is subjected to frequency analysis. Frequency analysis can help discover hidden frequency components in the signal, such as periodic changes in the system, or the frequency distribution of sudden changes. Through these frequency components, risk factors related to safety can be further identified; through frequency analysis, the safety performance spectrum is extracted. The spectrum can reflect the frequency distribution of the signal and reveal the changing characteristics of the signal. For example, a gas leak may produce significant changes in a specific frequency band, which can be clearly shown through spectrum analysis; from the safety performance spectrum, special attention is paid to the peak characteristics. A peak usually refers to a significant peak in the spectrum, which may indicate an abnormality or potential risk in the system. For example, a significant peak in a specific frequency band in the spectrum may mean that there is a fault in a certain equipment or process, or that the concentration of a certain chemical gas has suddenly increased; the peak characteristics in the safety performance spectrum are combined with other safety performance indicators (such as temperature, pressure, gas concentration, etc.) for comprehensive analysis. These comprehensive data can reveal potential safety hazards in each area. For example, a sudden spike in gas concentration in a certain area at a specific moment, combined with changes in temperature and pressure, may indicate that there is a risk of excessive chemical reactions in that area.

[0085] Extraction of safety performance features: Finally, key feature data related to safety performance for each area are extracted from these data. These feature data usually include: abnormal gas concentration, temperature change, equipment status, spikes in the spectrum, etc. These data are the basis for subsequent risk assessment and prediction, and can help enterprises give early warnings and take appropriate safety management measures.

[0086] In this alternative embodiment, analyzing the safety performance feature data using a risk analysis algorithm to obtain the key risk factors affecting the safety performance of each area includes the following steps: S21. Standardize the safety performance feature data collected for each area; S22. Based on the entropy weight method, assign weights to each index of the safety performance feature data for each area, and calculate the safety risk index for each area; S23. Based on the safety risk index of each area, use a clustering algorithm to analyze the safety risks of each area to obtain areas with different risk levels; S24. Use a risk analysis algorithm to perform correlation analysis on areas with different risk levels to identify the key risk factors affecting the safety performance of each area.

[0087] Specifically, the risk analysis algorithm is an improved FP-growth algorithm. First, use the entropy weight method to assign weights to the risk indicators of the safety performance feature data for each area respectively to calculate the risk index of the safety performance feature data in different areas. Second, use the risk index as a feature and perform risk clustering based on the MiniBatchKmeans algorithm to obtain the risk levels of each area. Finally, use the improved FP-growth algorithm with constraints to mine the association rules of the key risk factors of the safety performance of each area, mine the association relationship between the risk levels of each area and the safety performance feature data of each area, and perform correlation analysis on the mined results, which helps chemical enterprises give early warnings of potential safety problems. Compared with the traditional FP-growth algorithm, the present invention deletes invalid redundant rules on the basis of the traditional FP-growth algorithm to improve the rule mining efficiency. MiniBatchKmeans clustering does not have to use all sample data, but selects a small part of the data from the dataset to represent all data and optimizes the objective function, that is, adopts a method called batch processing. The basic idea of this algorithm for clustering is basically the same as that of K-Means clustering, except that on the basis of the model constructed by the K-Means algorithm, each time a small batch of data subsets randomly selected from the whole is used for training, so as to cluster the sample data. Since the scale of the sample data is reduced, the iteration speed is improved, achieving time optimization.

[0088] In this alternative embodiment, based on the entropy weight method, the weight distribution of each index for the safety performance characteristic data of each region is carried out, and calculating the safety risk index of each region includes the following steps: S221. Analyze the safety performance characteristics of each region based on the expert scoring method and the eigenvector method to determine the preliminary weight of each criterion layer; S222. For each criterion layer, calculate the weight information entropy of each index within each criterion through the entropy weight method; S223. Use the weight information entropy to correct the weight vector between criterion layers, and normalize the corrected weights; S224. Combine the safety performance characteristic data of each region, and based on the weights corrected by the weight information entropy, perform weighted summation, and calculate the safety risk index of each region through the weighted score.

[0089] In this alternative embodiment, the formula for calculating the weight information entropy of each index within each criterion is: ; In the formula, X(t) represents the weight information entropy of the t-th criterion layer; ω(t, j) represents the weight of the j-th safety performance index of the t-th criterion layer; n(t) represents the total number of safety performance indexes in the t-th criterion layer.

[0090] In this alternative embodiment, the formula for calculating the safety risk index of each region through the weighted score is: ; In the formula, S i represents the safety risk index of the i-th region; ω j represents the weight of the j-th safety performance index; x ij represents the standardized value of the i-th region on the j-th safety performance index; n represents the total number of safety performance indexes.

[0091] In this alternative embodiment, using the risk analysis algorithm to conduct correlation analysis on regions with different risk levels to identify the key risk factors affecting the safety performance of each region includes the following steps: S231. Input the sampling inspection data set containing the safety performance and corresponding risk levels of each region, and set the minimum support and minimum confidence as the thresholds for correlation analysis; S232. Analyze the safety performance data set of each region, search for safety risk factors whose occurrence frequency is greater than or equal to the minimum support, and form a set of frequent risk factors with the safety risk factors, and sort them in descending order according to their support; S233. Based on the frequent risk factor set, gradually establish branches for the header of the frequent pattern tree, arrange each branch in descending order of the support degree of the frequent items, and gradually insert and construct the frequent pattern tree; S234. Starting from the bottom of the frequent pattern tree, gradually traverse each node and recursively generate all combinations of frequent risk factors that meet the minimum support degree requirement; S235. Calculate the confidence degree between frequent risk factors, retain the association rules with a confidence degree greater than the minimum confidence degree threshold, and determine the safety performance risk level of each region; S236. Store the association rules between the selected frequent risk factors and their corresponding safety performance risk levels, and identify the key risk factors affecting the safety performance of each region.

[0092] To facilitate the understanding of the above technical solution of the present invention, the following will detail the use of the risk analysis algorithm to perform association analysis on regions with different risk levels in the actual process of the present invention, and identify the key risk factors affecting the safety performance of each region.

[0093] Step 1. Input the data set: First, input the safety performance data and corresponding risk levels of each region. Each region has multiple safety performance indicators and the safety risk level of this region.

[0094] Set the thresholds: Set the minimum support degree and the minimum confidence degree. These two thresholds are used for association rule mining to determine which rules are valid.

[0095] Suppose the data set is as shown in Table 1: Table 1: Data Set Minimum support degree: Set to 0.4, which means that only risk factors with an occurrence frequency greater than or equal to 40% will be considered.

[0096] Minimum confidence degree: Set to 0.7, which means that only association rules with a confidence degree greater than or equal to 70% are valid.

[0097] Step 2. Analyze the data set: Scan the data sets of all regions and find the occurrence frequency of each safety performance factor.

[0098] Frequent risk factor set: According to the minimum support degree, screen out risk factors with an occurrence frequency greater than or equal to the threshold.

[0099] Calculate the frequency: F1 appears 3 times (in regions R1, R3, R4).

[0100] F2 appears 3 times (in regions R1, R2, R5).

[0101] F3 appears 4 times (in regions R1, R2, R4, R5).

[0102] F4 appears 4 times (in regions R2, R3, R4, R5).

[0103] F5 appears 2 times (in regions R3, R5).

[0104] Support calculation: Support of F1: 3 / 5 = 0.6.

[0105] Support of F2: 3 / 5 = 0.6.

[0106] Support of F3: 4 / 5 = 0.8.

[0107] Support of F4: 4 / 5 = 0.8.

[0108] Support of F5: 2 / 5 = 0.4.

[0109] Frequent risk factor set (support greater than or equal to 0.4): F1, F2, F3, F4, F5.

[0110] Sorted in descending order of support: 1) F3 (0.8).

[0111] 2) F4 (0.8).

[0112] 3) F1 (0.6).

[0113] 4) F2 (0.6).

[0114] 5) F5 (0.4).

[0115] Step 3: Construct the frequent pattern tree: Based on the frequent risk factor set, construct the frequent pattern tree, gradually insert each frequent risk factor, and sort according to its support.

[0116] Construction process: For the frequent risk factor set, first insert the elements F3 and F4 with the highest support to build the top of the tree. Then, insert F1, F2, F5 in turn, ensuring that each node is sorted in descending order of support.

[0117] Step 4: Recursive traversal: Starting from the bottom of the tree, recursively traverse each node and generate all frequent risk factor combinations that meet the minimum support.

[0118] Combination generation: Starting from the bottom of the tree, traverse the node combinations such as F1, F2, F5. All combinations must meet the minimum support (0.4).

[0119] Suppose the following are the frequent combinations that meet the support requirements: {F1, F2} (support 0.4).

[0120] {F3, F4} (support 0.8).

[0121] {F1, F3} (support 0.6).

[0122] Step Five, confidence calculation: Calculate the confidence between frequent risk factors and filter out the association rules with confidence greater than the minimum threshold.

[0123] Confidence formula: ; For example, calculate the confidence of {F1, F2}: Support({F1, F2}) = 0.4.

[0124] Support(F1) = 0.6.

[0125] Confidence(F1 → F2) = 0.4 / 0.6 = 0.67.

[0126] Since 0.67 is greater than the minimum confidence of 0.7, the rule F1 → F2 does not meet the requirements.

[0127] Step Six, store association rules: Store the association rules that meet the confidence requirements and the safety risk levels of the regions.

[0128] Identify key risk factors: According to these rules, identify the key risk factors that affect the safety performance of the region.

[0129] Suppose through association rule mining, the following association rules are obtained: {F3, F4} → High risk.

[0130] {F1, F3} → Medium risk.

[0131] Final key risk factors: F3, F4 are the key factors in high-risk areas.

[0132] F1, F3 are the key factors in medium-risk areas.

[0133] In this alternative embodiment, based on the key risk factors, a risk prediction model is established, and using the risk prediction model to predict the safety performance trends of each region at future times includes the following steps: S31. Input the set of key risk factors that affect the safety performance of the region, and decompose the input set of key risk factors to extract several intrinsic mode functions; S32. Divide each intrinsic mode function into a training set and a test set according to the time series and perform standardization processing; S33. Construct a risk prediction model based on a neural network (i.e., a long short-term memory neural network model), optimize the key parameters of the risk prediction model, and use the optimized risk prediction model to predict the test set and generate the safety performance trends of each region at future times.

[0134] It should be explained that, first, input the identified key risk factors that affect the safety performance of the region. For example, assume that the identified key risk factors are: F3 (equipment failure), F1 (gas concentration), and F4 (ambient temperature); decompose these key risk factors and extract the intrinsic mode functions (IMFs) of each factor. The intrinsic mode function is a component extracted from the original time series through methods such as empirical mode decomposition (EMD), which can reveal different frequency components and trends in the data. Each key risk factor will generate multiple intrinsic mode functions for further analysis of their impact on safety performance; for each extracted intrinsic mode function, based on the time series data, divide it into a training set and a test set. Generally, the training set accounts for 70%-80%, and the remaining part is used for the test set. For example, select the first 3 years as the training set and the data for the remaining 1 year as the test set; perform standardization processing on the data in the training set and the test set. Standardization can transform the data into a standard normal distribution with a mean of 0 and a variance of 1 to eliminate the dimensional differences between different indicators and make the model more stable and accurate. Commonly used standardization methods include Z-score standardization or Min-Max standardization; based on the standardized training set data, construct a risk prediction model based on the neural network algorithm. The neural network simulates the neural structure of the human brain through multiple layers of neuron nodes, can automatically learn and extract the non-linear relationships in the data, and is suitable for complex data prediction tasks; optimize the key parameters of the neural network model, such as the learning rate, the number of network layers, the number of neurons, etc. These parameters have an important impact on the training efficiency and prediction effect of the model. Methods such as cross-validation or grid search can be used for parameter optimization; use the optimized neural network model to predict the test set and generate the safety performance values (such as safety risk indices, change trends of various safety indicators, etc.) of each region in the future for a period of time. According to the prediction results, the change trends of the safety performance of different regions can be identified to provide a reference for decision-making.

[0135] In this alternative embodiment, constructing a risk prediction model based on a neural network, optimizing the key parameters of the risk prediction model, and using the optimized risk prediction model to predict the test set and generate the safety performance trends of each region at future times include the following steps: S331. Set the search range of the hyperparameters of the neural network model and the feature subset that affects the model performance; S332. Randomly initialize the parameter configuration and feature subset of the neural network, and construct the search space; S333. Construct and train a neural network model according to the initialized parameter configuration and feature subset of the neural network; S334. Select several candidate neural network model configurations with the best performance according to the fitness value of the risk prediction model; S335. For the remaining candidate neural network model configurations, perform random search and calculate the fitness value of the new neural network model configuration; S336. Repeat steps S334 to S335, continue with multiple rounds of optimization until the maximum number of iterations is reached, select the neural network configuration with the best performance as the optimized risk prediction model; and use the optimized risk prediction model to predict the test set, and output the safety performance trend of each region at future times.

[0136] It should be noted that in order to optimize the neural network model, hyperparameters of the neural network need to be set, such as the number of network layers, the number of neurons in each layer, the learning rate, the activation function, the batch size, etc. The setting range of each hyperparameter will affect the training effect and performance of the neural network.

[0137] For example: Learning rate (LearningRate) range: 0.001 to 0.1.

[0138] Number of network layers: It can be selected from 1 layer to 5 layers.

[0139] Number of neurons in each layer: For example, set the range of the number of neurons in each layer from 32 to 256.

[0140] Activation function: It can be selected as ReLU, Tanh or Sigmoid, etc.

[0141] Batch size (BatchSize): For example, select 16, 32, 64, 16, 32, 64, etc.

[0142] For the key risk factors affecting safety performance, which features are selected as inputs will affect the prediction ability of the model. Through feature selection algorithms (such as mutual information method, correlation analysis, etc.), select the subset that is most critical to safety risks. For example, select gas concentration, equipment failure and environmental temperature as input features; initialize the hyperparameters of the neural network (such as learning rate, number of layers, number of neurons in each layer, etc.) by randomization method. These initial configurations provide a starting point for model optimization and avoid the local optimum problem that may be caused by fixed hyperparameter settings.

[0143] For example: Initialize the learning rate to 0.01.

[0144] Randomly select 3 network layers, with 128 neurons in each layer.

[0145] Select ReLU as the activation function.

[0146] Select 32 as the batch size.

[0147] According to the initialized hyperparameter configuration, construct a searchable parameter space. This is a high-dimensional space, where each dimension corresponds to a hyperparameter and its possible values. Optimization algorithms (such as random search, grid search, Bayesian optimization, bee algorithm, etc.) will find the optimal hyperparameter combination from this space; Build a neural network model according to the initialized hyperparameter configuration. For example, set the corresponding network structure according to the number of network layers, the number of neurons in each layer, etc., and use the training set data to train the model; Use the training set to train the constructed neural network. Usually, the training of the neural network is carried out through the backpropagation algorithm, with the goal of minimizing the loss function (such as mean squared error, cross-entropy loss, etc.); By calculating the fitness value of the model on the training set (such as the accuracy and loss value of the validation set), select several model configurations with the best performance. The higher the fitness value, the better the performance of the neural network on the training data; Select several groups of model configurations with the highest fitness values as candidate models. These models may achieve better training results under different hyperparameter settings; For the remaining candidate models, continue to adjust their hyperparameters through random search. Random search is a method of exploring the model configuration space by randomly selecting hyperparameters, which can help find better model configurations; The new model configuration after each random search will be trained on the training set, calculate its fitness value, and evaluate its performance on the validation set; According to the evaluation results, repeat step S334 and step S335 for multiple rounds of optimization until the maximum number of iterations is reached. After each round of optimization, select the configuration with the best fitness and gradually approach the optimal solution; After multiple rounds of optimization, select the neural network configuration with the best performance. These optimal configurations have the best generalization ability and prediction effect; Use the finally optimized model to predict the test set and output the safety performance trends of each region at future times.

[0148] Specifically, the present invention uses the bee algorithm to optimize the key parameters of the risk prediction model. The bee algorithm is an optimization algorithm inspired by the foraging behavior of bees in nature. In the algorithm, the position of the food source represents a set of solutions to the optimization problem. A small search area formed by the position of the food source and its neighborhood is called a flower cluster. During the process of searching for food, the bee colony first sends out scout bees to search. The scout bees continuously fly from one food source to another, searching for all possible food sources. When the profitability of the food source found (such as the sugar content) exceeds a certain amount, the scout bee returns to the hive, unloads the honey and performs a magical "waggle dance" on the "dancing board". This "waggle dance" is a necessary tool for information exchange within the bee colony, enabling the entire bee colony to understand the orientation, distance to the hive, and profitability of the discovered food source. After dancing, the scout bee leads the follower bees waiting outside the hive back to the corresponding food source to collect honey, and the food source with a higher profitability will recruit more follower bees to collect honey. During the honey collection process, the forager bees calculate the profitability of the food source they are collecting honey from and return to the hive to perform the "waggle dance", transmitting information about the current profitability of the food source to the bee colony. If the profitability of the food source is still high, more follower bees will be recruited to collect honey; if the profitability of the food source is low, the current food source will be abandoned.

[0149] According to another embodiment of the present invention, as Figure 2 shown, there is also provided a safety risk zoning management system for safety supervision of chemical enterprises, which system includes: A data acquisition module 1, configured to acquire safety monitoring data of each area within a chemical enterprise and perform feature extraction on the safety monitoring data to obtain safety performance feature data of each area; A risk analysis module 2, configured to analyze the safety performance feature data by using a risk analysis algorithm to obtain key risk factors affecting the safety performance of each area; A model prediction module 3, configured to establish a risk prediction model based on the key risk factors and use the risk prediction model to predict the safety performance trend of each area at a future moment; A safety management module 4, configured to evaluate the safety performance status of each area based on the safety performance trend and take safety management measures according to the evaluation results.

[0150] In summary, by means of the above technical solutions of the present invention, the present invention conducts correlation analysis through a risk analysis algorithm, and can identify key risk factors affecting the safety performance of the region from a large amount of data. The identification of these key factors helps chemical enterprises to early warn potential safety problems. By constructing frequent pattern trees and association rules, the mutual relationships between different risk factors can be discovered, so as to provide forward-looking risk prediction for enterprises. By calculating the safety risk index of each region and combining with key risk factors, the main factors affecting safety can be quickly determined in case of anomalies, and a clear basis for safety decision-making can be provided, enabling enterprises to promptly activate the emergency response mechanism and avoid the spread of risks. By accurately dividing the safety management requirements of regions with different risk levels, chemical enterprises can then more reasonably allocate safety resources. The construction and optimization of the risk prediction model based on key risk factors of the present invention provide strong technical support for the safety management of chemical enterprises. Through methods such as data standardization, intrinsic mode extraction, and neural network optimization, the future safety performance trends can be accurately identified and predicted, and potential risks can be discovered in advance. It not only improves the safety management efficiency of chemical enterprises, but also enhances the emergency response ability, reduces the possibility of accidents, and ensures the safe operation of enterprises.

[0151] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A safety risk zoning management method for safety supervision of chemical enterprises, characterized in that: The method comprises the following steps: S1. Obtain the safety monitoring data of each area in the chemical enterprise, and extract the features of the safety monitoring data to obtain the safety performance feature data of each area; S2. Analyze the safety performance characteristic data using risk analysis algorithms to obtain key risk factors that affect the safety performance of each area; S3. Based on key risk factors, establish a risk prediction model and use the risk prediction model to predict the safety performance trend of each area in the future; S4. Based on the safety performance trend, evaluate the safety performance status of each area and take safety management measures according to the evaluation results.

2. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 1, characterized in that: The step of obtaining the safety monitoring data of each area in the chemical enterprise and extracting features from the safety monitoring data to obtain the safety performance feature data of each area comprises the following steps: S11. Collect safety monitoring data from various areas within the chemical enterprise and use the safety performance equation to calculate key characteristic indicators related to regional safety; S12, using a data processing method to process the safety monitoring data of each area and extract the safety performance signal of each area; S13, demodulating the extracted safety performance signal using a generalized demodulation algorithm, and performing frequency analysis on the demodulated safety performance signal to extract a safety performance spectrum; S14. Combine the peak characteristics in the safety performance spectrum with other safety performance indicators to extract safety performance characteristic data related to the safety performance of each area.

3. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 1, characterized in that: The method of analyzing the safety performance characteristic data using the risk analysis algorithm to obtain the key risk factors affecting the safety performance of each area includes the following steps: S21, standardizing the collected safety performance characteristic data of each area; S22. Based on the entropy weight method, weights of various indicators are allocated to the safety performance characteristic data of each area, and the safety risk index of each area is calculated; S23. Based on the security risk index of each area, a clustering algorithm is used to perform a level analysis on the security risk of each area to obtain areas with different risk levels; S24. Use risk analysis algorithms to conduct correlation analysis on areas with different risk levels and identify key risk factors that affect the safety performance of each area.

4. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 3, characterized in that: The method of allocating weights of various indicators to the safety performance characteristic data of each area based on the entropy weight method and calculating the safety risk index of each area includes the following steps: S221. Analyze the safety performance characteristics of each area based on the expert scoring method and the feature vector method, and determine the preliminary weight of each criterion layer; S222. For each criterion layer, the weight information entropy of each indicator in each criterion is calculated by the entropy weight method; S223, using weight information entropy to correct the weight vectors between criterion layers, and normalizing the corrected weights; S224. Combine the safety performance characteristic data of each area, perform weighted summation based on the weights corrected by the weight information entropy, and calculate the safety risk index of each area through the weighted score.

5. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 4, characterized in that: The formula for calculating the weighted information entropy of each indicator in each criterion is: ； Where X(t) represents the weighted information entropy of the t-th criterion layer; ω(t, j) represents the weight of the j-th safety performance indicator of the t-th criterion layer; n(t) represents the total number of safety performance indicators in the t-th criterion layer.

6. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 5, characterized in that: The formula for calculating the safety risk index of each area by weighted score is: ； In the formula, S i represents the safety risk index of the ith region; ω j represents the weight of the jth safety performance indicator; x ij represents the standardized value of the i-th area on the j-th safety performance index; n represents the total number of safety performance indicators.

7. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 6, characterized in that: The method of using the risk analysis algorithm to perform correlation analysis on areas of different risk levels and identifying key risk factors affecting the safety performance of each area includes the following steps: S231, inputting a sampling data set including the safety performance of each area and the corresponding risk level, and setting the minimum support and minimum confidence as the thresholds for association analysis; S232, analyzing the safety performance data set of each area, searching for safety risk factors with a frequency greater than or equal to the minimum support, and forming the safety risk factors into a frequent risk factor set, and arranging them in descending order according to their support from large to small; S233, based on the frequent risk factor set, gradually establish branches for the header of the frequent pattern tree, and arrange each branch in descending order according to the support of the frequent items, and gradually insert them to construct the frequent pattern tree; S234, starting from the bottom of the frequent pattern tree, traverse each node step by step, and recursively generate all frequent risk factor combinations that meet the minimum support requirement; S235. Calculate the confidence between frequent risk factors, retain the association rules whose confidence is greater than the minimum confidence threshold, and determine the safety performance risk level of each area; S236. Store the association rules between the screened frequent risk factors and their corresponding safety performance risk levels, and identify the key risk factors that affect the safety performance of each area.

8. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 1, characterized in that: The method of establishing a risk prediction model based on key risk factors and using the risk prediction model to predict the safety performance trend of each area at a future time includes the following steps: S31. Input the identified key risk factor set that affects the regional safety performance, decompose the input key risk factor set, and extract a number of intrinsic mode functions; S32, dividing each intrinsic mode function into a training set and a test set according to the time series, and performing standardization processing; S33. Build a risk prediction model based on a neural network, optimize the key parameters of the risk prediction model, use the optimized risk prediction model to predict the test set, and generate the safety performance trend of each area at future times.

9. A safety risk zoning management method for safety supervision of chemical enterprises according to claim 8, characterized in that: The method of building a risk prediction model based on a neural network, optimizing key parameters of the risk prediction model, predicting a test set using the optimized risk prediction model, and generating safety performance trends of various regions at future moments includes the following steps: S331, setting the hyperparameter search range of the neural network model and the feature subset that affects the model performance; S332, randomly initializing the parameter configuration and feature subset of the neural network, and constructing a search space; S333, constructing and training a neural network model according to the parameter configuration and feature subset of the initialized neural network; S334. Selecting several candidate neural network model configurations with the best performance according to the fitness value of the risk prediction model; S335, performing a random search for the remaining candidate neural network model configurations, and calculating the fitness value of the new neural network model configuration; S336. Repeat steps S334 to S335, and continue multiple rounds of optimization until the maximum number of iterations is reached, and select the neural network configuration with the best performance as the optimized risk prediction model; and use the optimized risk prediction model to predict the test set and output the safety performance trend of each area at future times.

10. A safety risk zoning management system for safety supervision of chemical enterprises, used to implement the safety risk zoning management method for safety supervision of chemical enterprises according to any one of claims 1 to 9, characterized in that: The system includes: The data acquisition module is used to obtain the safety monitoring data of each area in the chemical enterprise, and extract the characteristics of the safety monitoring data to obtain the safety performance characteristic data of each area; The risk analysis module is used to analyze the safety performance characteristic data using the risk analysis algorithm to obtain the key risk factors that affect the safety performance of each area; Model prediction module, which is used to establish a risk prediction model based on key risk factors, and use the risk prediction model to predict the safety performance trend of each area at future moments; The safety management module is used to evaluate the safety performance status of each area based on the safety performance trend, and take safety management measures according to the evaluation results.

Images