Production processing site risk management and control method and device, electronic equipment and storage medium
By combining feedforward neural networks, recurrent neural networks, SVM algorithms and agent reinforcement learning algorithms, feature extraction and risk assessment of production workers' video data is constructed, and the worker's production training model is solved, which is difficult to identify production operations in the existing technology, and the accurate identification and control of production site risks is achieved.
Patent Information
- Application Number
- CN202411683241.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-22
- Publication Date
- 2025-02-28
AI Technical Summary
The prior art is difficult to effectively identify and evaluate the consistency and fluency of production operations in the field of safe production, resulting in the inability to accurately identify safety risks in the entire production process.
Feedforward neural network and recurrent neural network are used to extract feature and process the video data of production workers, identify key action frames and analyze action coherence; combine SVM algorithm and agent reinforcement learning algorithm to evaluate the performance and decision-making capabilities of production operations; by building a worker production training model, determine the risk level at the production site and formulate corresponding risk control plans.
It realizes effective identification of the logical judgment of the coherence of production operations, can accurately identify dangerous operation behaviors such as disordered production operations, missing items, and additional items, quickly evaluate the risk level at the production site, and issue an alarm to ensure safe production.
Smart Images
Figure CN119539493A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of image processing technology, and specifically relates to a production and processing site risk management method, device, electronic equipment and storage medium. Background Art
[0002] In the field of safe production, traditional methods mainly rely on manual monitoring and management, which has many limitations, such as low efficiency, high cost, and difficulty in 24-hour uninterrupted monitoring. Existing recognition algorithms are independent, such as object recognition: identifying key objects in the video, such as workers, tools, equipment, etc. Behavior recognition: analyzing workers' behavior patterns, such as climbing, falling, illegal operations, etc.
[0003] However, in the field of safe production, operations are continuous. Even if a single action is correct, but the order of implementation is wrong, or there are too many or too few actions, there will still be safety risks. For example, a safety operation must be divided into the order of before, during, and after the operation according to the specification. It cannot be random. Only identifying that the actions at a certain stage meet the safety standards does not necessarily prove that the entire process is standardized, and naturally it will not produce accurate safety identification results. Summary of the invention
[0004] The present invention aims at solving the defects of the prior art and provides a method for risk control in a production and processing site.
[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:
[0006] A production and processing site risk management method, the risk management method comprising the following steps:
[0007] Obtain production workers' production action test and operation standard specification video data, use feedforward neural network to extract features from video frames, identify key action frames, use recurrent neural network to process the time series of key frames, analyze the continuity and fluency of actions, identify and evaluate operation behaviors, including production workers' production operation postures and dangerous operation postures, the dangerous operation postures include production operation sequence disorder behavior, production operation missing item behavior, production operation additional item behavior, and generate production operation posture recognition results;
[0008] Based on the production operation posture recognition results, the SVM algorithm is used to simulate the performance of production workers in the production operation scenario, and the production operation performance of production workers in the simulated production operation environment is evaluated, and the adaptability and reaction speed of production workers under sudden production abnormalities are analyzed to generate production operation performance evaluation results;
[0009] Based on the production operation performance evaluation results, the intelligent agent reinforcement learning algorithm is used to simulate the production scenarios in the production operation to evaluate the production operation response capabilities of the production workers, including the coherence logic judgment of the production operation, and generate decision-making ability analysis results;
[0010] Obtain the attribute information of each production equipment, perform anomaly detection on each production equipment respectively, and obtain the anomaly detection results corresponding to each production equipment. During the detection, the attribute information of each production equipment is used to produce the feature matrix results corresponding to each production equipment;
[0011] Based on the production operation posture recognition results, the production operation posture recognition results, the production operation performance evaluation results, and the feature matrix results, a worker production training model is constructed, and then the risk level of the production site is determined through the worker production training model;
[0012] Based on the safety risk level of the production site, a safety risk management and control plan for the production site is determined.
[0013] Furthermore, the worker production training model also includes AI video technology, which is used to re-arrange the timeline of a single action and several actions of the production workers, or to determine missing items and added items.
[0014] Furthermore, the dangerous operating posture also includes climbing behavior, falling behavior, crossing non-working area behavior, illegal equipment operation behavior and / or smoking in dangerous areas.
[0015] Furthermore, a production and processing site risk management method of the present invention also includes a violation behavior recognition algorithm, which collects human behavior characteristics of dangerous operating postures and collects and records channel state information of the human behavior characteristics.
[0016] Furthermore, a production and processing site risk management method of the present invention also includes: obtaining production environment parameter information, wherein the production environment parameter information includes temperature, humidity and gas concentration, using an environmental quality index algorithm to simulate the production environment scenario, evaluate the production operation capabilities of production workers, and generate environmental operation performance evaluation results.
[0017] Furthermore, the gas concentration includes any one of the group consisting of oxygen concentration, carbon dioxide concentration, carbon monoxide concentration and hydrogen sulfide concentration.
[0018] Furthermore, when the risk level of the production site is determined, an alarm can be sounded and an alarm text message or email can be sent out, or a dangerous operation alarm phone call can be made.
[0019] A production and processing site risk control device, the risk control device comprising: a production operation posture recognition module, used to identify the production operation posture and dangerous operation posture of production workers; a production operation performance evaluation module, used to evaluate the contingency of production workers in dealing with sudden abnormalities in the production process; a decision-making ability analysis module, used to analyze the production operation ability of production workers; and a feature matrix module; used for attribute information anomaly detection and analysis of each production equipment; wherein the production operation posture recognition module, the production operation performance evaluation module, the decision-making ability analysis module and the feature matrix module construct a worker production training model.
[0020] An electronic device comprises: a memory for storing executable instructions; and a processor for implementing the above-mentioned production and processing site risk management method when running the executable instructions stored in the memory.
[0021] A computer-readable storage medium stores executable instructions, which, when executed by a processor, implement the above-mentioned production and processing site risk management and control method.
[0022] The present invention has significant improvements over the prior art:
[0023] 1. The present invention can make logical judgments on the continuity of production operations and effectively identify whether the production operations are performed in accordance with standard specifications. A feedforward neural network is used to identify the production operation postures and dangerous operation postures of production workers, effectively preventing dangerous operation behaviors such as disordered production operation sequence, missing production operations, and increased production operations during the production operation process.
[0024] 2. The present invention can also evaluate the production operation of workers when sudden abnormal situations occur during the production process, so as to obtain the adaptability and reaction speed of the workers.
[0025] 3. The present invention can effectively detect the attribute information of each production equipment, and use the attribute information of each production equipment to produce a characteristic matrix result corresponding to each production equipment, and quickly obtain the production status information of each production equipment through the characteristic matrix result.
[0026] 4. The present invention can quickly determine the risk level of the production site, and at the same time as issuing an alarm, send an alarm text message or email, or make a dangerous operation alarm call.
[0027] 5. The present invention can collect production environment parameter information so as to know the impact of the production environment on workers' production operations; it can also quickly know the concentration of harmful gases in the production environment so as to quickly take countermeasures to ensure the safety of workers in production. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the prior art description. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.
[0029] Figure 1 The present invention is a schematic diagram of the workflow of a production and processing site risk management method. DETAILED DESCRIPTION
[0030] In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present application.
[0031] Embodiment 1:
[0032] like Figure 1 As shown, a production and processing site risk management method, the risk management method comprises the following steps:
[0033] S1: Obtain production workers' production action test and operation standard specification video data, use feedforward neural network to extract features from video frames, identify key action frames, use recurrent neural network to process the time series of key frames, analyze the continuity and fluency of actions, identify and evaluate operation behaviors, including production workers' production operation postures and dangerous operation postures. The dangerous operation postures include production operation sequence disorder, production operation missing items, and production operation additional items, and generate production operation posture recognition results.
[0034] S2: Based on the production operation posture recognition results, the SVM algorithm is used to simulate the performance of production workers in the production operation scenario. By evaluating the production operation performance of production workers in the simulated production operation environment, the adaptability and reaction speed of production workers under sudden production anomalies are analyzed to generate production operation performance evaluation results.
[0035] S3: Based on the production operation performance evaluation results, using the intelligent agent reinforcement learning algorithm, by simulating the production scenarios in the production operation, evaluate the production operation response capabilities of the production workers, including the coherent logical judgment of the production operation, and generate decision-making ability analysis results.
[0036] S4: Obtain the attribute information of each production equipment, perform anomaly detection on each production equipment respectively, and obtain the anomaly detection result corresponding to each production equipment. During the detection, the attribute information of each production equipment is used to produce the feature matrix result corresponding to each production equipment.
[0037] S5: Based on the production operation posture recognition results, the production operation posture recognition results, the production operation performance evaluation results, and the feature matrix results, a worker production training model is constructed, and then the risk level of the production site is determined through the worker production training model.
[0038] S6: Determine a safety risk management and control plan for the production site based on the safety risk level of the production site.
[0039] The SVM algorithm, also known as support vector machine, is a binary classification model that is good at handling small sample classification problems. It has strong generalization capabilities and can handle nonlinear classification problems. It can also be used to handle multi-classification problems through the combination of multiple SVMs.
[0040] It should be noted that SVM is used to analyze the linearly separable case. For the linearly inseparable case, the samples in the low-dimensional input space are mapped to the high-dimensional feature space by using a nonlinear mapping algorithm to make them linearly separable, so that the high-dimensional feature space of the safety production site violation identification algorithm can use a linear algorithm to perform linear analysis on the nonlinear features of the samples. The safety production site violation identification algorithm constructs the optimal segmentation hyperplane in the feature space based on the structural risk minimization theory, so that the learner is globally optimized, and the expected risk in the entire sample space meets a certain upper bound with a certain probability.
[0041] In order to further determine whether the workers' production operations are compliant, the worker production training model also includes AI video technology, which is used to re-sort the timeline of single actions and multiple actions of production workers, or to determine missing items or added items.
[0042] Dangerous operating postures also include climbing, falling, crossing non-working areas, illegal operation of equipment and / or smoking in dangerous areas.
[0043] In some optional embodiments, in order to effectively identify illegal operating behaviors, an illegal behavior recognition algorithm is added, and the human behavior characteristics of dangerous operating postures are collected through the illegal behavior recognition algorithm, and channel state information that records the human behavior characteristics is collected.
[0044] It should be noted that the channel state information (CSI) presents the amplitude and phase of multipath propagation at different frequencies, thereby more accurately characterizing the channel with frequency selective fading characteristics of the safety production operation site violation identification algorithm.
[0045] The advantages of channel state information are: 1. Improving the adaptability and reliability of the communication system. 2. Improving the efficiency and accuracy of data transmission. 3. Supporting high-precision positioning and tracking.
[0046] In some optional embodiments, in order to obtain production environment information in a timely manner, a production and processing site risk management method of the present invention also includes: obtaining production environment parameter information, wherein the production environment parameter information includes temperature, humidity and gas concentration, and using an environmental quality index algorithm to simulate the production environment scenario, evaluate the production operation capabilities of production workers, and generate environmental operation performance evaluation results.
[0047] The temperature of the production environment can generally be controlled at 20℃~25℃. It can be selected according to the temperature of the specific production industry.
[0048] The humidity of the production environment can generally be controlled at 20% to 70%, which can be selected according to the temperature of the specific production industry.
[0049] In some optional embodiments, a structure of gas concentration is provided, wherein the gas concentration includes any one of the group consisting of oxygen concentration, carbon dioxide concentration, carbon monoxide concentration and hydrogen sulfide concentration.
[0050] The gas concentration may be an oxygen concentration, a carbon dioxide concentration, a carbon monoxide concentration, or a hydrogen sulfide concentration.
[0051] Another structure of the gas concentration may include oxygen concentration, carbon dioxide concentration, carbon monoxide concentration and hydrogen sulfide concentration.
[0052] Another structure of the gas concentration may include oxygen concentration, carbon dioxide concentration, and carbon monoxide concentration.
[0053] It is understandable that the gas concentration to be monitored can be selected according to actual production conditions.
[0054] Oxygen (O2): The oxygen content in the air of a confined space workplace should be 19.5%-21%. If the oxygen content in the air is lower than 19.5%, ventilation measures should be taken.
[0055] Carbon monoxide (CO): time-weighted average allowable concentration is 20 mg / m 3 (17.5ppm), the permissible concentration for short-term contact is 30mg / m 3 (26ppm).
[0056] Carbon dioxide (CO2): time-weighted average allowable concentration is 9000 mg / m 3 (5000ppm), the permissible concentration for short-term contact is 18000mg / m 3 (10000ppm).
[0057] In some optional embodiments, in order to facilitate timely notification of supervisors, when the risk level of the production site is determined, an alarm can be issued and an alarm text message or email can be sent out, or a dangerous operation alarm phone number can be dialed, so that the supervisors can receive the information in time.
[0058] In some optional embodiments, a production and processing site risk control device is provided, and the risk control device includes: a production operation posture recognition module, which is used to identify the production operation posture and dangerous operation posture of production workers; a production operation performance evaluation module, which is used to evaluate the contingency of production workers to deal with sudden abnormalities in the production process; a decision-making ability analysis module, which is used to analyze the production operation ability of production workers; and a feature matrix module; which is used for attribute information anomaly detection and analysis of each production equipment; wherein the production operation posture recognition module, production operation performance evaluation module, decision-making ability analysis module and feature matrix module construct a worker production training model.
[0059] In the electronic device provided in the embodiment of the present application, the memory in the electronic device is used to store executable instructions; when the processor is used to run the executable instructions stored in the memory, the above-mentioned production and processing site risk control method is implemented. Its specific implementation method is consistent with the above-mentioned method implementation method, and can achieve the same beneficial effects, which will not be repeated here.
[0060] An embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the methods provided in the above embodiments are implemented.
[0061] It is understandable that the logic commands in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several commands to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.
[0062] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the protection scope of the invention.
Claims
1. A production and processing site risk management method, characterized in that: The risk management method comprises the following steps: Obtain production workers' production action test and operation standard specification video data, use feedforward neural network to extract features from video frames, identify key action frames, use recurrent neural network to process the time series of key frames, analyze the continuity and fluency of actions, identify and evaluate operation behaviors, including production workers' production operation postures and dangerous operation postures, the dangerous operation postures include production operation sequence disorder behavior, production operation missing item behavior, production operation additional item behavior, and generate production operation posture recognition results; Based on the production operation posture recognition results, the SVM algorithm is used to simulate the performance of production workers in the production operation scenario, and the production operation performance of production workers in the simulated production operation environment is evaluated, and the adaptability and reaction speed of production workers under sudden production abnormalities are analyzed to generate production operation performance evaluation results; Based on the production operation performance evaluation results, the intelligent agent reinforcement learning algorithm is used to simulate the production scenarios in the production operation to evaluate the production operation response capabilities of the production workers, including the coherence logic judgment of the production operation, and generate decision-making ability analysis results; Obtain the attribute information of each production equipment, perform anomaly detection on each production equipment respectively, and obtain the anomaly detection results corresponding to each production equipment. During the detection, the attribute information of each production equipment is used to produce the feature matrix results corresponding to each production equipment; Based on the production operation posture recognition results, the production operation posture recognition results, the production operation performance evaluation results, and the feature matrix results, a worker production training model is constructed, and then the risk level of the production site is determined through the worker production training model; Based on the safety risk level of the production site, a safety risk management and control plan for the production site is determined.
2. The production and processing site risk management method according to claim 1 is characterized in that: The worker production training model also includes AI video technology, which is used to re-arrange the timeline of a single action or several actions of a production worker, or to determine missing items or added items.
3. The production and processing site risk management method according to claim 1 is characterized in that: The dangerous operating postures also include climbing, falling, crossing non-working areas, illegal equipment operation and / or smoking in dangerous areas.
4. The production and processing site risk management method according to any one of claims 1 to 3, characterized in that: It also includes an illegal behavior recognition algorithm, which collects human behavior characteristics of dangerous operating postures and collects and records channel state information of the human behavior characteristics.
5. The production and processing site risk management method according to claim 1 is characterized in that: Also includes: Acquire production environment parameter information, which includes temperature, humidity and gas concentration, and use an environmental quality index algorithm to simulate production environment scenarios, evaluate the production operation capabilities of production workers, and generate environmental operation performance evaluation results.
6. The production and processing site risk management method according to claim 5 is characterized in that: The gas concentration includes any one of the group consisting of oxygen concentration, carbon dioxide concentration, carbon monoxide concentration and hydrogen sulfide concentration.
7. The production and processing site risk management method according to claim 1 is characterized in that: When determining the risk level of the production site, an alarm can be sounded and an alarm text message or email can be sent out, or a dangerous operation alarm phone call can be made.
8. A production and processing site risk control device, characterized in that: The risk management and control device comprises: Production operation posture recognition module, used to identify production workers' production operation posture and dangerous operation posture; The production operation performance evaluation module is used to evaluate the response of production workers to sudden abnormalities in the production process; Decision-making capability analysis module, used to analyze the production operation capability of production workers; and Feature matrix module; used for abnormal detection and analysis of attribute information of each production equipment; Among them, the production operation posture recognition module, production operation performance evaluation module, decision-making ability analysis module and feature matrix module construct a worker production training model.
9. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable instructions; The processor is used to implement the production and processing site risk management method described in any one of claims 1 to 7 when running the executable instructions stored in the memory.
10. A computer-readable storage medium storing executable instructions, characterized in that: When the executable instructions are executed by the processor, the production and processing site risk management method described in any one of claims 1 to 7 is implemented.