Fault full-process command assistant based on intelligent engine

Abstract

The invention relates to a fault whole-process command assistant based on an intelligent engine, and the command assistant specifically comprises the steps: multi-source data integration and system integration: carrying out the data structural processing and data missing problem perception of data information obtained from a plurality of systems through the technologies of machine learning, pattern recognition and data mining, and carrying out the data completion through an algorithm; identifying multi-source data feature points, constructing a fault studying and judging system, extracting power grid fault information, constructing a power grid fault analysis rule base, performing real-time operation risk analysis on a fault area, and guiding intelligent generation of a risk plan; building a fault intelligent decision-making platform, carrying out scoring evaluation on the generated scheme decision, and generating a final power supply strategy and an optimal power supply conversion scheme; transparent information transmission of a fault disposal process is realized, and rapid fault disposal is realized based on fault disposal rehearsal and a security check technology. The management benefit is improved, the power grid fault handling efficiency is improved, an efficient intelligent distribution network command decision-making platform is constructed, and quality improvement and efficiency improvement are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent operation and maintenance of electric power equipment, and specifically relates to a full-process fault command assistant based on a smart engine. Background Art

[0002] Traditional fault handling is scattered across multiple systems such as the DAS system, EMS system, OMS system, and service system. The information is complex and needs to be manually sorted out. Information is transmitted over the phone, which results in a large workload and insufficient integration of fault data information. There is insufficient analysis and mining of fault data correlation, which leads to the transmission of distribution network fault handling information not being accurate and intelligent enough, and a large amount of manpower and material resources are wasted unnecessarily.

[0003] At present, there is a lack of application of digital technology in the process of fault diagnosis and treatment of distribution networks. The fault location judgment relies on manual diagnosis, the fault treatment plan relies on manual preparation and the execution of the treatment plan still relies on manual operation. There are inevitably misjudgments and the judgment results are affected by the experience and ability of the judges, which makes it impossible to respond quickly and correctly.

[0004] In the process of distribution network fault detection, there is a problem of insufficient construction of distribution network fault diagnosis and intelligent handling safety control and evaluation system. The comprehensive judgment of safety verification in the fault handling process relies on manual work, which requires high ability and experience of employees and has a high risk of error. The safety assessment and risk analysis of fault handling results are not intelligent, resulting in a complicated evaluation and analysis process. The response to abnormal situations in fault handling is not intelligent enough, resulting in the problem of insufficient response time and speed, and a large amount of unnecessary losses, which has a great impact on users. Summary of the invention

[0005] The technical problem to be solved by the present invention is: intelligently analyzing distribution network faults and making corresponding decisions and solutions, intelligently executing solutions to handle problems and recording and broadcasting fault information. The present invention provides a fault full-process command assistant based on an intelligent engine to overcome the deficiencies of the prior art. The technical solution adopted by the present invention to solve the problems existing in the prior art is:

[0006] Step 1: Multi-source data integration and system integration, including:

[0007] Step 1.1, collect data information from the DAS system, EMS system, OMS system, supply and service system and program-controlled telephone system;

[0008] Step 1.2: The data information is passed through a bidirectional long short-term memory network, conditional random field, support vector machine and According to the principle, data is structured and data missing problems are perceived. If there is a data missing problem, jump to step 1.3. If there is no data missing, jump to step 2.1.

[0009] Step 1.3, import the structured data and use the interpolation algorithm to complete the data;

[0010] Step 2: Identify feature points of multi-source data and build a fault analysis system, including:

[0011] Step 2.1, based on natural language processing technology and knowledge graph technology, extract historical fault information and related information of typical faults, extract key points, analyze the issuance time and sequence of various equipment fault information and the causal relationship with other fault information, and build a power grid fault analysis rule base;

[0012] Step 2.2, after learning the power grid fault analysis rule base based on signal clustering analysis and causal analysis model, the fault information is imported for analysis and judgment;

[0013] Step 2.3: Based on the real-time topological relationship of the power grid, equipment status, fault point information, and the power supply relationship between the upper and lower power grids, the user set is traced starting from the fault point, and the changes in the equipment status before and after the fault are compared to form a user set thematic map;

[0014] Step 2.4, based on the N-1 algorithm, analyze the real-time operation risk of the power grid in the fault area, extract key information, and generate a risk plan;

[0015] Step 3: Build an intelligent fault decision-making platform, including:

[0016] Step 3.1, using the strategy tree algorithm in machine learning and the strategy network in the neural network to generate strategies for multi-step power grid operations;

[0017] Step 3.2: Use the value network to evaluate the strategy generated in step 3.1, introduce the influencing parameters through the SHAP value, analyze the power dispatching ontology library based on the AI ​​big model, and generate the adjustment strategy and strategy adjustment results;

[0018] Step 4: Implement one-button sequential control of fault handling solutions based on the DAS system, including:

[0019] Step 4.1: After the adjustment plan corresponding to the adjustment strategy is confirmed by the dispatcher, it is connected to the DAS system, and one-key sequential control is implemented after the secondary verification is correct;

[0020] Step 4.2: Double confirmation of the sequential control process, using the telesignaling and telemetry information of the relevant equipment to automatically check and confirm the operation results;

[0021] Step 5: Transparent information transmission of fault handling process, including:

[0022] Step 5.1: Build a distribution network artificial intelligence dispatching and command system to broadcast information during the entire process of fault analysis, fault impact range analysis, fault decision-making, and fault handling;

[0023] Step 5.2, based on the structured analysis of historical dispatch logs, expert experience and historical fault logs, the extracted fault equipment features, fault handling process and key time points are received, a dispatch log is generated, and the fault log is recorded based on the OMS interface.

[0024] Preferably, the interpolation algorithm in step 1 includes smoothing prediction and single value interpolation.

[0025] Preferably, the influencing parameters described in step 3.2 include the operational convenience of the power grid equipment and the historical status of the equipment.

[0026] Preferably, the specific method for parsing the power dispatching ontology library based on the AI ​​big model described in step 3.2 is:

[0027] The distribution network control data includes control regulations, protection rules, and accident warning rule files. After the rule files are segmented into phrases, data extraction and mapping, data cleaning and conversion are performed to complete the preprocessing of the text of the control regulations; the preprocessed text information is parsed based on natural language processing technology, and the unified vector representation of the control regulations text knowledge is achieved based on word embedding technology, and vectorized data is stored and managed; knowledge representation and modeling are performed based on knowledge graph technology, and a text knowledge base is constructed, that is, a control domain knowledge base is constructed;

[0028] The historical experience data of fault handling in the power dispatching ontology library is extracted, and the historical experience data of fault handling is imported into the control domain knowledge base after behavioral learning based on machine learning to realize the knowledge update of the knowledge base.

[0029] Preferably, the distribution network artificial intelligence dispatching and command system described in step 5.1 includes text conversion, feature point recognition, and disassembly of the expert experience semantics in the data information based on the large model technology, and digital representation and feedback of the semantic knowledge;

[0030] The information broadcasting method described in step 5.1 includes voice broadcasting, SMS notification, and intelligent outbound calling.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] Improve management efficiency and significantly improve the efficiency of power grid fault handling. After access to the distribution network artificial intelligence dispatching command, based on the full-process command system of power grid faults, the handling time of distribution network busbar faults was shortened from 60 minutes to 5 minutes.

[0033] Build an efficient and intelligent distribution network command and decision-making platform. Promote the visual online control of the entire chain of business, focus on the key nodes of the entire process of the command chain, comprehensively sort out the six key business links of fault analysis, fault decision-making, risk control, fault handling, information transmission, and log generation, build a full-process fault handling model, and promote intelligent and closed-loop control of regulation business.

[0034] Integrate artificial intelligence technology to improve quality and efficiency. Use artificial intelligence applications to reduce the workload of dispatchers on duty, achieve manual substitution and improve the quality and efficiency of distribution network control, intelligently push power grid fault events, intelligent outbound calls, and fully perceive the handling process, freeing dispatchers from repetitive non-technical work and greatly improving the operating efficiency of the team. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0036] Figure 1 This is the overall flow chart of the fault full-process command assistant system based on the intelligent engine of the present invention

[0037] Figure 2 This is the overall structure diagram of the fault full-process command assistant system based on the intelligent engine of the present invention

[0038] Figure 3 This is a flowchart of the intelligent decision-making process of power grid fault in the embodiment of the fault full process command assistant based on the smart engine of the present invention DETAILED DESCRIPTION

[0039] For example, certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that hardware manufacturers may use different terms to refer to the same component. This specification and claims do not use differences in names as a way to distinguish components, but use differences in the functions of components as the criteria for distinction. For example, "including" mentioned throughout the specification and claims is an open term, so it should be interpreted as "including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve the technical problem within a certain error range and basically achieve the technical effect.

[0040] In the description of the present invention, it is necessary to understand that the terms "up", "down", "front", "back", "left", "right", "horizontal", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0041] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0042] The following is a further detailed description of the intelligent engine-based full-process fault assistant of the present invention in conjunction with the accompanying drawings, but it is not intended to limit the present invention.

[0043] like Figure 1 As shown, the specific implementation method of the fault full-process command assistant based on the intelligent engine includes steps 1-5.

[0044] Step 1: Multi-source data integration and system integration, including:

[0045] Collect data information from the DAS system, EMS system, OMS system, supply and service system and program-controlled telephone system to achieve information integration of multi-platform systems and improve platform compatibility.

[0046] Since there are a lot of unstructured and semi-structured data in each platform, it is necessary to transform the data information through bidirectional long short-term memory network, conditional random field, support vector machine and The principle is to carry out data structuring and perceive data missing problems.

[0047] If there is a problem of missing data, import the structured data and use the smooth prediction algorithm to complete the data to achieve multi-source data fusion and system integration.

[0048] Step 2: Identify feature points of multi-source data and build a fault analysis system, including:

[0049] Based on natural language processing technology and knowledge graph technology, extract historical fault information and related information of typical faults, extract key points, analyze the issuance time and sequence of various equipment fault information and the causal relationship with other fault information, and build a power grid fault analysis rule base;

[0050] After learning the power grid fault analysis rule base based on signal clustering analysis and causal analysis models, the fault information is imported for analysis and judgment;

[0051] Based on the user set tracing algorithm, the impact range of power grid faults can be determined. Based on the real-time topological relationship of the power grid, equipment status, fault point information, and the power supply relationship between the upper and lower power grids, the user set is traced starting from the fault point, and the changes in the equipment status before and after the fault are compared to form a user set thematic map;

[0052] Analyze the real-time operation risk of the power grid in the fault area based on the N-1 algorithm and extract key risk information;

[0053] Step 3: Build an intelligent fault decision-making platform, including:

[0054] Based on the decision tree algorithm, intelligent decision-making for power grid faults is achieved. The strategy tree algorithm in machine learning and the strategy network in the neural network are used to generate strategies for multi-step power grid operations. The generated strategies are scored and evaluated using the value network, and the influencing parameters are introduced through the SHAP value. The parameters include the operational convenience of power grid equipment and the historical status of the equipment. The distribution network control rules are analyzed based on the AI ​​big model to generate adjustment strategies and strategy adjustment results.

[0055] The specific method for parsing the power dispatching ontology library based on the AI ​​big model is:

[0056] The distribution network control data includes control regulations, protection rules, and accident warning rule files. After the rule files are segmented into phrases, data extraction and mapping, data cleaning and conversion are performed to complete the preprocessing of the text of the control regulations; the preprocessed text information is parsed based on natural language processing technology, and the unified vector representation of the control regulations text knowledge is achieved based on word embedding technology, and vectorized data is stored and managed; knowledge representation and modeling are performed based on knowledge graph technology, and a text knowledge base is constructed, that is, a control domain knowledge base is constructed;

[0057] The historical experience data of fault handling in the power dispatching ontology library is extracted, and the historical experience data of fault handling is imported into the control domain knowledge base after behavioral learning based on machine learning to realize the knowledge update of the knowledge base.

[0058] The specific method for parsing the distribution network control rules based on the AI ​​big model is to perform data extraction and mapping, data cleaning and conversion after phrase segmentation of the control procedure text to complete the preprocessing of the control procedure text; parse the preprocessed text information based on natural language processing technology, realize the unified vector representation of the control procedure text knowledge based on word embedding technology, and store and manage vectorized data; perform knowledge representation and modeling based on knowledge graph technology, and build a text knowledge base, that is, build a control field knowledge base.

[0059] Step 4: Implement one-button sequential control of fault handling solutions based on the DAS system, including:

[0060] After the adjustment plan corresponding to the adjustment strategy is confirmed by the dispatcher, it will be connected with the DAS system, and one-key sequential control will be realized after the second verification. Double confirmation of the sequential control process is carried out, and the operation results are automatically verified and confirmed using the telesignaling and telemetry information of related equipment.

[0061] Step 5: Transparent information transmission of fault handling process, including:

[0062] Build an artificial intelligence dispatching and command system for distribution networks to broadcast information throughout the entire process of fault assessment, fault impact range analysis, fault decision-making, and fault handling.

[0063] Based on the structured analysis of historical dispatch logs, expert experience and historical fault logs, the extracted fault equipment features, fault handling process and key time points are received, the dispatch log is generated, and the fault log is recorded based on the OMS interface.

[0064] like Figure 2 The figure shows the application architecture of the fault full-process command assistant based on the intelligent engine of the present invention, which includes five parts: application, function, module, technology and data.

[0065] The embodiments are introduced to more clearly illustrate the technical solution of the present invention.

[0066] like Figure 3 As shown in the figure, after extracting key information from the data information integrated from the multi-source platform, it is concluded that the fault is load power outage, equipment overload, and there is a risk of power outage for important users. The real-time topology relationship of the power grid, equipment status information, fault point information, and the common point relationship between the upper and lower power grids are read to trace the user set and form a user set thematic map.

[0067] The N-1 algorithm is used to analyze the operational risk of the fault area and form risk parameters. The strategy tree algorithm and strategy network are used to generate preliminary strategies. The busbar grounding data and historical disposal process, that is, the convenience of implementing the operation and the historical status of the equipment are considered, and the final strategy solution is generated by importing it into the decision network.

[0068] For load power outage faults, decisions are generated to restore power supply. For heavy overload faults, strategies are generated to control loads. For the risk of power outages for important users, strategies are generated to restore the dual power supply mode. The generated strategies are transmitted to the dispatcher port, and after confirmation by the operator, they are connected to the DAS system for secondary verification. After the secondary verification is correct, one-button sequential control is implemented. The operation results are automatically verified and confirmed using the telesignaling and telemetry information of related equipment.

[0069] The artificial intelligence dispatching and command system broadcasts information on the entire process of fault analysis, fault impact range analysis, fault decision-making, and fault handling. Fault information is sent to relevant personnel through the intelligent outbound call system in the form of notification text messages. Faults that can be automatically controlled to restore power supply and other repair operations are automatically repaired. For parts that require manual repair, the system pushes fault information to the fault repair unit for fault repair. The system dynamically monitors the maintenance progress, playing a role in supervising and warning the fault maintenance unit.

[0070] The entire fault handling process is pushed to the command system platform through the artificial intelligence dispatching platform, which can monitor the entire process, make the entire process transparent, and timely discover system problems to improve system security. It can also intelligently link fault outage and power restoration information to the national official system platform for timely information broadcasting.

[0071] Based on the structured analysis of historical dispatch logs, expert experience and historical fault logs, the extracted fault equipment features, fault handling process and key time points are received, the dispatch log is generated, and the fault log is recorded based on the OMS interface.

[0072] The embodiments of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A full-process fault command assistant based on an intelligent engine, characterized by: The specific steps include: Step 1, multi-source data integration and system integration, including: Step 1.1, collect data information from the DAS system, EMS system, OMS system, supply and service system and program-controlled telephone system; Step 1.2: The data information is passed through a bidirectional long short-term memory network, conditional random field, support vector machine and According to the principle, data is structured and data missing problems are perceived. If there is a data missing problem, jump to step 1.

3. If there is no data missing, jump to step 2. Step 1.3, import the structured data and use the interpolation algorithm to complete the data; Step 2: Identify feature points of multi-source data and build a fault analysis system, including: Step 2.1, based on natural language processing technology and knowledge graph technology, extract historical fault information and related information of typical faults, extract key points, analyze the issuance time and sequence of various equipment fault information and the causal relationship with other fault information, and build a power grid fault analysis rule base; Step 2.2, after learning the power grid fault analysis rule base based on signal clustering analysis and causal analysis model, the fault information is imported for analysis and judgment; Step 2.3: Based on the real-time topological relationship of the power grid, equipment status, fault point information, and the power supply relationship between the upper and lower power grids, the user set is traced starting from the fault point, and the changes in the equipment status before and after the fault are compared to form a user set thematic map; Step 2.4, based on the N-1 algorithm, analyze the real-time operation risk of the power grid in the fault area, extract key information, and form a risk warning; Step 3: Build an intelligent fault decision-making platform, including: Step 3.1, using the strategy tree algorithm in machine learning and the strategy network in the neural network to generate strategies for multi-step power grid operations; Step 3.2: Use the value network to evaluate the strategy generated in step 3.1, introduce the influencing parameters through the SHAP value, analyze the power dispatching ontology library based on the AI ​​big model, and generate the adjustment strategy and strategy adjustment results; Step 4: Implement one-button sequential control of the fault handling solution based on the DAS system, including: Step 4.1: After the adjustment plan corresponding to the adjustment strategy is confirmed by the dispatcher, it is connected with the DAS system for secondary verification. After the secondary verification is correct, one-key sequential control is implemented; Step 4.2, check the operation results, and use the telesignaling and telemetry information of related equipment to automatically check and confirm the operation results; Step 5: Transparent information transmission of fault handling process, including: Step 5.1: Build a distribution network artificial intelligence dispatching and command system to broadcast information during the entire process of fault analysis, fault impact range analysis, fault decision-making, and fault handling; Step 5.2, based on the structured analysis of historical dispatch logs, expert experience and historical fault logs, the extracted fault equipment features, fault handling process and key time points are received, a dispatch log is generated, and the fault log is recorded based on the OMS interface.

2. The intelligent engine-based fault full-process command assistant according to claim 1, characterized in that: The interpolation algorithm described in step 1 includes smoothing prediction and single value interpolation.

3. The intelligent engine-based fault full-process command assistant according to claim 1, characterized in that: The influencing parameters described in step 3.2 include the operational convenience of the power grid equipment and the historical status of the equipment.

4. The intelligent engine-based fault full-process command assistant according to claim 1, characterized in that: The specific method for parsing the power dispatching ontology library based on the AI ​​big model described in step 3.2 is: The distribution network control data includes control regulations, protection rules, and accident warning rule files. After the rule files are segmented into phrases, data extraction and mapping, data cleaning and conversion are performed to complete the preprocessing of the control regulations text; Parse the preprocessed text information based on natural language processing technology, realize the unified vector representation of regulatory regulations text knowledge based on word embedding technology, and store and manage vectorized data; perform knowledge representation and modeling based on knowledge graph technology, and build a text knowledge base, that is, a regulatory domain knowledge base; The historical experience data of fault handling in the power dispatching ontology library is extracted, and the historical experience data of fault handling is imported into the control domain knowledge base after behavioral learning based on machine learning to realize the knowledge update of the knowledge base.

5. The intelligent engine-based fault full-process command assistant according to claim 1, characterized in that: The distribution network artificial intelligence dispatching and command system described in step 5.1 includes text conversion, feature point recognition, and disassembly of the expert experience semantics in the data information based on the large model technology, and digital representation and feedback of the semantic knowledge; The information broadcasting method described in step 5.1 includes voice broadcasting, SMS notification, and intelligent outbound calling.

Images