Work high-incidence risk identification and early warning method based on artificial intelligence technology

Abstract

The invention provides an operation high-incidence risk identification and early warning method based on an artificial intelligence technology, and belongs to the field of safety production. The method comprises the steps of improving the accuracy of a risk prediction model by combining a machine learning technology on the basis of operation historical data of a risk control card; continuously optimizing the risk prediction model by adopting a machine learning technology based on actual operation data of the risk control card to form a closed-loop risk management process; and supervised learning, unsupervised learning and data analysis and data mining technologies are utilized to continuously improve the accuracy of the risk prediction model. According to the method, the capability of distinguishing risks in the operation process is high, and the accuracy of risk early warning is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of safe production technology, and in particular to an artificial intelligence-based high-risk operation identification and early warning method. Background Art

[0002] The digital and intelligent transformation of power systems has been an important development trend in the power industry in recent years. From the initial manual operation to automation and informatization, and then to the current digitalization and intelligence, every technological upgrade of the power system has greatly improved production efficiency and reduced safety risks.

[0003] In the existing power production process, production safety is becoming more and more important. On the one hand, more and more laws and regulations have made clear provisions for law-based public security, requiring the implementation of more stringent and standardized safety production standards. In the face of new situations and new requirements, it is necessary to promote law-based public security and ensure that safety production management and behavior are in compliance with the law. On the other hand, the existing production site operation risk control capabilities are not strong, the on-site operation safety level is not high, and the production site operation risk control system is not sound. It is impossible to strengthen source prevention and hierarchical control, and it is difficult to effectively improve the level of production site operation risk control. At the same time, the ability to identify and warn of high-risk operations is insufficient, making it difficult to ensure the safe and stable operation of the power system. Summary of the invention

[0004] The purpose of the present invention is to provide a method for identifying and warning of high-risk operations based on artificial intelligence technology. The method has a strong ability to identify risks in the operation process and greatly improves the accuracy of risk warning.

[0005] In order to achieve the above object, the present invention provides a method for identifying and warning high-risk operations based on artificial intelligence technology, the method comprising:

[0006] Combine machine learning technology with historical data of risk control card operation to improve the accuracy of risk prediction model;

[0007] Based on the actual data of risk control card operation, machine learning technology is used to continuously optimize the risk prediction model to form a closed-loop risk management process;

[0008] Use supervised learning and unsupervised learning as well as data analysis and data mining technology to continuously improve the accuracy of the risk prediction model, generate risk control cards based on weekly plans, extract job types and job risk levels as keyword information, generate work tickets, extract equipment names and main work content information as keyword information, search the risk information database and historical work violation information based on keywords, generate risk point warnings and risk prevention and control measures information, and push them to mobile terminals and read receipts.

[0009] Preferably, based on the historical data of the risk control card operation, the accuracy of the risk prediction model is improved by combining machine learning technology, including:

[0010] Collect historical data of risk control cards and perform preprocessing. Obtain historical data of risk events of multiple scenarios covering different professions and different work teams from the risk control card system. Clean the collected data to remove invalid, duplicate or erroneous data records, and mark the characteristic fields related to risk prediction in the data.

[0011] Select models and improve training. Choose logistic regression, decision tree, neural network or deep learning algorithm according to the characteristics of data and prediction requirements. Continuously train and optimize models, divide data into standard data, test data and validation data for model training, validation and testing, and optimize and modify model parameters through cross-validation, grid search and random search.

[0012] Complete early warning mechanisms and response measures, set reasonable early warning thresholds based on model prediction results, generate risk warning information and send it to relevant personnel through notifications, and have on-site managers formulate targeted response measures and plans based on the risk level and risk factor of the early warning information.

[0013] Preferably, based on the actual data of the risk control card operation, the risk prediction model is continuously optimized using machine learning technology to form a closed-loop risk management process including:

[0014] Collect and process actual data, including multi-dimensional data on people, machines, materials, methods, and processes. Data processing includes deduplication, processing missing values, detecting and processing outliers, and data normalization and standardization.

[0015] Perform feature engineering to extract useful features from raw data so that the model can better understand and predict risks, including:

[0016] Select the most relevant and influential features from the original features;

[0017] Use statistical methods or machine learning algorithms to transform raw data into more meaningful and interpretable features;

[0018] Extract new features by combining or exponentiating the original features;

[0019] Use performance evaluation indicators to evaluate the model, continuously optimize the model by adjusting model parameters, feature selection and algorithm selection, continuously monitor the performance of the model after deployment and make adjustments based on actual conditions, and continuously optimize the parameters and features of the model by collecting new risk control card actual data and feedback;

[0020] Send risk prediction information to professional systems, and confirm the effectiveness of warnings based on whether the professional systems have updated plans and the updated content. At the same time, improve the hit rate of risk warnings and promote an increase in the feedback ratio of professional systems to form a closed loop of risk management.

[0021] Preferably, the feature selection method includes correlation coefficient and variance threshold, and the feature extraction method includes principal component analysis PCA and linear discriminant analysis LDA.

[0022] Preferably, feature expansion includes dividing or multiplying two features to obtain a new feature.

[0023] Preferably, using supervised learning and unsupervised learning as well as data analysis and data mining techniques to continuously improve the accuracy of the risk prediction model includes:

[0024] An artificial intelligence big data model is introduced to construct a multi-dimensional operational risk adaptive quantitative assessment model through multiple static factors and an adaptive weighting method that integrates subjective and objective dynamics. At the same time, based on the PDCA theory, from the five dimensions of man, machine, material, method, and environment, the artificial intelligence big data model is used to analyze the influencing factors of risk assessment level determination in various operational scenarios, and an early warning mechanism is established based on the comprehensive risk assessment results.

[0025] Preferably, the multiple static factors include voltage level, operation scope, operation content, equipment type, maintenance type and whether it is a core component.

[0026] Preferably, after receiving the warning push, the operating conditions are checked and comprehensively judged, including:

[0027] If the result of human judgment is the same as the warning information, the business specialist will adjust the job plan information in a timely manner; if the result of human judgment is different from the warning information of the robot model, the business specialist will feed back the different information to the model, and the model will analyze the feedback information, find the structure and rules of the data, tune and iterate the model, and perform unsupervised learning of the model; at the same time, use the machine model training set data and validation set data to evaluate the model performance, optimize the model through hyperparameter tuning, and perform supervised learning; and through reciprocating learning, the prediction results of the training model are infinitely close to the actual results.

[0028] Through the above technical solution, on the basis of the historical data of risk control card operation, combined with machine learning technology, the accuracy of the risk prediction model is improved; based on the actual data of risk control card operation, the risk prediction model is continuously optimized using machine learning technology to form a closed-loop risk management process; at the same time, supervised learning and unsupervised learning as well as data analysis and data mining technology are used to continuously improve the accuracy of the risk prediction model; in this way, this method not only enhances the ability to identify risks in the operation process, but also greatly improves the accuracy of risk warning.

[0029] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0031] Figure 1 It is a risk warning idea diagram in the high-risk identification and warning method of operations based on artificial intelligence technology provided by the present invention;

[0032] Figure 2 It is a comparison diagram of algorithms in the high-risk identification and early warning method for operations based on artificial intelligence technology provided by the present invention. DETAILED DESCRIPTION

[0033] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0034] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0035] See also Figure 1 and Figure 2 The present invention provides a method for identifying and warning high-risk operations based on artificial intelligence technology, the method comprising:

[0036] Based on the historical data of risk control card operation, combined with machine learning technology, the accuracy of risk prediction model is improved;

[0037] Based on the actual data of risk control card operation, the risk prediction model is continuously optimized using machine learning technology to form a closed-loop risk management process;

[0038] And use supervised learning and unsupervised learning as well as data analysis and data mining techniques to continuously improve the accuracy of risk prediction models; specifically including:

[0039] Step 1: Generate risk control cards synchronously based on the weekly plan;

[0040] Step 2: extracting the operation type and operation risk level as keyword information;

[0041] Step 3: Generate a work ticket;

[0042] Step 4: Extract the equipment name and main information of the work content as keyword information;

[0043] Step 5: Search the risk information database and historical operation violation information based on keywords;

[0044] Step 6: Generate risk point warning and risk prevention and control measures information and push it to the mobile terminal;

[0045] Step 7. Read the receipt.

[0046] In this embodiment, it is preferred to combine machine learning technology with the historical data of risk control card operation to improve the accuracy of the risk prediction model, including:

[0047] Collection and preprocessing of historical data of risk control cards: First, the historical data obtained from the risk control card system should cover multiple types of scenario risk events of different professions and different work teams to ensure the integrity of the data; second, the collected data should be cleaned to remove invalid, duplicate or erroneous data records to ensure data quality; finally, the data should be marked, and the operation type, operation time, operation location and feature fields related to risk prediction should be marked from the collected data to reduce the complexity of the model and improve the prediction efficiency;

[0048] Model selection and training improvement: First, select the appropriate machine learning algorithm based on the characteristics of the data and prediction requirements; second, continuously train and optimize the model, divide the data into standard data, test data, and validation data for model training, validation, and testing, and optimize and modify the model parameters through cross-validation, grid search, and random search to optimize model performance and improve the accuracy of model prediction;

[0049] Complete early warning mechanism and response measures: First, according to the model prediction results, set a reasonable early warning threshold, generate risk warning information, and send it to relevant personnel via SMS, email and system notification; second, on-site managers formulate targeted response measures and plans based on the risk level and risk factor of the early warning information.

[0050] Among them, the above-mentioned appropriate machine learning algorithms include logistic regression, decision tree, neural network and deep learning algorithm.

[0051] In this embodiment, it is preferred to continuously optimize the risk prediction model based on the actual data of the risk control card operation using machine learning technology to form a closed-loop risk management process. Specifically, the following steps are included:

[0052] Step A: Data collection and processing:

[0053] In risk prediction, the quality of the actual data of the risk control card plays a vital role in the accuracy of the final model. First, the actual data needs to be collected and processed. The collected data includes multi-dimensional data of people, machines, materials, methods, and environments. Data processing includes deduplication, processing missing values, detecting and processing outliers, and data normalization or standardization to ensure the quality and integrity of the data;

[0054] Step B: Perform feature engineering:

[0055] By extracting useful features from raw data, the model can better understand and predict risks. Specifically, in feature engineering, commonly used methods include:

[0056] (1) Feature selection: selecting the most relevant and influential features from the original features; commonly used feature selection methods include correlation coefficient, variance threshold, etc.

[0057] (2) Feature extraction: using statistical methods or machine learning algorithms to transform raw data into more meaningful and interpretable features; common feature extraction methods include principal component analysis (PCA) and linear discriminant analysis (LDA).

[0058] (3) Feature expansion: extracting new features by combining or exponentiating the original features. For example, dividing or multiplying two features to obtain new features.

[0059] Step C: Model evaluation and optimization:

[0060] First, the model is evaluated using performance evaluation indicators such as AUC value. The model is continuously optimized by adjusting model parameters, feature selection, and algorithm selection to improve prediction accuracy. For example, the Information Value statistic can be used to screen variable features, and the variable features can be re-encoded using woe (weight of evidence) coding to improve the variable's ability to distinguish between the two types of users.

[0061] Secondly, continuous monitoring and adjustment: After the model is deployed, its performance is continuously monitored and adjusted according to actual conditions; by collecting actual data and feedback from new risk control cards, the parameters and characteristics of the model are continuously optimized to ensure the accuracy and effectiveness of the model.

[0062] Step D: Risk management closed loop:

[0063] First, the risk prediction information is sent to the professional system, and the effectiveness of the warning is confirmed based on whether the professional system has updated the plan and the updated content;

[0064] Secondly, conduct professional system feedback to improve the hit rate of risk warnings and promote an increase in the proportion of professional system feedback to form a closed loop of risk management.

[0065] Through the above steps and methods, machine learning technology can be effectively used to continuously optimize the risk prediction model, improve the accuracy and efficiency of the prediction, and form a closed-loop risk management process.

[0066] In the traditional mode, business specialists manually split the voltage level, power outage scope, and operation content according to the maintenance plan, match the risk classification description based on experience, and integrate the judgment to give the operation risk rating. And according to the operation content, split the risk factors, and upload the plan. In this embodiment, it is preferred to use supervised learning and unsupervised learning as well as data analysis and data mining technology to continuously improve the accuracy of the risk prediction model, including:

[0067] The artificial intelligence big model is introduced to build a multi-dimensional operation risk adaptive quantitative assessment model through multiple static factors such as voltage level, operation scope, operation content, equipment type, maintenance type, whether it is a core component, etc., using the adaptive weighting method of dynamic fusion of subjective and objective factors to achieve accurate assessment of risk levels in different operation scenarios;

[0068] At the same time, the artificial intelligence big model is based on the PDCA theory, starting from the five dimensions of people, machines, materials, methods, and environment, to analyze the influencing factors of risk assessment levels in various operation scenarios, and establishes an early warning mechanism based on the comprehensive risk assessment results. It provides data support for the modification and determination of operation risk levels, so as to achieve accurate assessment and closed-loop control of various operation risks.

[0069] Specifically, after receiving the early warning push, the on-site personnel will check and make a comprehensive judgment on the operating conditions:

[0070] If the result of human judgment is the same as the warning information, the business specialist will adjust the operation plan information in a timely manner;

[0071] If the human judgment result is different from the warning information of the robot model, the business specialist will feed back the different information to the model. The model will perform data analysis and mining based on the feedback information, find the structure and rules of the data, and tune and iterate the model. This process is the unsupervised learning of the model. At the same time, the machine model can use the training set data and the validation set to evaluate the model performance and optimize the model through hyperparameter tuning. This process is supervised learning.

[0072] In this way, the machine model learns repeatedly in this way, so that the prediction results of the training model are infinitely close to the actual results, achieving the accuracy of the early warning.

[0073] In summary, the present invention studies the identification and early warning methods of high-risk operations based on big data, and studies the early warning information and digital implementation plans of control measures for different risk levels of production operations. Research measures to improve the risk control of the entire production operation process, and propose optimization and improvement measures for major risks in substation maintenance, transmission line maintenance, and distribution maintenance operations. At the same time, based on the historical operation data of the risk control card, analyze the frequency of various risks in the operation links of different professions and different work teams, combine risk prediction models and typical risk libraries, design a set of high-risk identification methods for operations based on information systems, and design different early warning methods for risk levels and risk factors. Through the data obtained from the operation risk control card, extract the risk level and early warning information of the production operation, formulate reasonable response measures and plans, and ensure the safe and stable operation of the power system.

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

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

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

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

[0078] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0079] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0080] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0081] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0082] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for identifying and warning high-risk operations based on artificial intelligence technology, characterized in that: The method comprises: Combine machine learning technology with historical data of risk control card operation to improve the accuracy of risk prediction model; Based on the actual data of risk control card operation, machine learning technology is used to continuously optimize the risk prediction model to form a closed-loop risk management process; Use supervised learning and unsupervised learning as well as data analysis and data mining technology to continuously improve the accuracy of the risk prediction model, generate risk control cards based on weekly plans, extract job types and job risk levels as keyword information, generate work tickets, extract equipment names and main work content information as keyword information, search the risk information database and historical work violation information based on keywords, generate risk point warnings and risk prevention and control measures information, and push them to mobile terminals and read receipts.

2. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 1 is characterized in that: Based on the historical data of risk control card operation, combined with machine learning technology, the accuracy of risk prediction model is improved, including: Collect historical data of risk control cards and perform preprocessing. Obtain historical data of risk events of multiple scenarios covering different professions and different work teams from the risk control card system. Clean the collected data to remove invalid, duplicate or erroneous data records, and mark the characteristic fields related to risk prediction in the data. Select models and improve training. Choose logistic regression, decision tree, neural network or deep learning algorithm according to the characteristics of data and prediction requirements. Continuously train and optimize models, divide data into standard data, test data and validation data for model training, validation and testing, and optimize and modify model parameters through cross-validation, grid search and random search. Complete early warning mechanisms and response measures, set reasonable early warning thresholds based on model prediction results, generate risk warning information and send it to relevant personnel through notifications, and have on-site managers formulate targeted response measures and plans based on the risk level and risk factor of the early warning information.

3. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 1 is characterized in that: Based on the actual data of the risk control card operation, the risk prediction model is continuously optimized using machine learning technology to form a closed-loop risk management process including: Collect and process actual data, including multi-dimensional data on people, machines, materials, methods, and processes. Data processing includes deduplication, processing missing values, detecting and processing outliers, and data normalization and standardization. Perform feature engineering to extract useful features from raw data so that the model can better understand and predict risks, including: Select the most relevant and influential features from the original features; Use statistical methods or machine learning algorithms to transform raw data into more meaningful and interpretable features; Extract new features by combining or exponentiating the original features; Use performance evaluation indicators to evaluate the model, continuously optimize the model by adjusting model parameters, feature selection and algorithm selection, continuously monitor the performance of the model after deployment and make adjustments based on actual conditions, and continuously optimize the parameters and features of the model by collecting new risk control card actual data and feedback; Send risk prediction information to professional systems, and confirm the effectiveness of warnings based on whether the professional systems have updated plans and the updated content. At the same time, improve the hit rate of risk warnings and promote an increase in the feedback ratio of professional systems to form a closed loop of risk management.

4. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 3 is characterized in that: Feature selection methods include correlation coefficient and variance threshold, and feature extraction methods include principal component analysis (PCA) and linear discriminant analysis (LDA).

5. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 3 or 4, characterized in that: Feature expansion involves dividing or multiplying two features to obtain a new feature.

6. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 1 is characterized in that: The use of supervised and unsupervised learning as well as data analysis and data mining techniques to continuously improve the accuracy of risk prediction models includes: An artificial intelligence big data model is introduced to construct a multi-dimensional operational risk adaptive quantitative assessment model through multiple static factors and an adaptive weighting method that integrates subjective and objective dynamics. At the same time, based on the PDCA theory, from the five dimensions of man, machine, material, method, and environment, the artificial intelligence big data model is used to analyze the influencing factors of risk assessment level determination in various operational scenarios, and an early warning mechanism is established based on the comprehensive risk assessment results.

7. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 6 is characterized in that: Multiple static factors include voltage level, operation scope, operation content, equipment type, maintenance type, and whether it is a core component.

8. The method for identifying and warning high-risk operations based on artificial intelligence technology according to claim 6 or 7, characterized in that: After receiving the early warning push, the operating conditions are checked and comprehensively judged, including: If the result of human judgment is the same as the warning information, the business specialist will adjust the job plan information in a timely manner; if the result of human judgment is different from the warning information of the robot model, the business specialist will feed back the different information to the model, and the model will analyze the feedback information, find the structure and rules of the data, tune and iterate the model, and perform unsupervised learning of the model; at the same time, use the machine model training set data and validation set data to evaluate the model performance, optimize the model through hyperparameter tuning, and perform supervised learning; and through reciprocating learning, the prediction results of the training model are infinitely close to the actual results.

Images