A guidance method for detecting charging anomalies in intelligent operation and maintenance of charging piles

Abstract

The present invention discloses a guiding method for detecting charging anomalies in intelligent operation and maintenance of charging piles, which can realize efficient data stream processing, and can provide valuable information and insights to support operation and maintenance business decisions and optimize operation and maintenance. The present invention uses ZooKeeper for distributed coordination, Kafka to realize efficient data stream processing, and Java open source high-concurrency non-blocking Socket frameworks such as Netty to build a data communication server specifically serving smart terminal devices and intelligent operation and maintenance models. The server is designed with a data communication interface, which can seamlessly connect to various types of smart terminals, and through precise business logic processing, deeply mine and use data for calculation and analysis, aiming to maximize the value of data and provide strong data support and decision-making basis for intelligent operation and maintenance models. The present invention is applied to the technical field of intelligent operation and maintenance of charging piles.

Description

Technical Field

[0001] The present invention relates to the field of intelligent operation and maintenance of charging piles, and in particular to a guiding method for detecting charging anomalies in intelligent operation and maintenance of charging piles. Background Art

[0002] With the vigorous development of new energy electric vehicle charging infrastructure, the complexity and timeliness of charging pile operation and maintenance have encountered unprecedented challenges. For charging operators, the charging operation and maintenance system plays a vital role. It can not only monitor the status of equipment in real time and efficiently handle maintenance needs, but also effectively reduce the overall operation and maintenance costs. For the user side, the system plays a key response and guidance role. By guiding users to operate correctly and troubleshoot problems by themselves, it significantly improves the operation and maintenance response speed and optimizes the user experience.

[0003] In the operation and maintenance of charging piles, common abnormal maintenance that users can solve by themselves accounts for a considerable proportion of the operation and maintenance work. For example, when the charging pile cannot start charging after being connected to the vehicle, users are often confused and at a loss, and may even choose to leave directly because they cannot determine the next step. This not only leads to the loss of passenger flow and revenue for the operator, but also many times, the root cause of the problem may be simply that the emergency stop button of the charging pile is accidentally touched or there is an abnormality in the user's vehicle itself. Traditional charging operation and maintenance systems usually have two main operation and maintenance solutions: 1. Generally speaking, users will rely on the customer service phone number provided by the charging pile or the operation and maintenance front-end software to contact the manufacturer for technical assistance. With the help of Internet of Things technology, the built-in sensors and monitoring devices of the charging pile can monitor various parameters in real time and detect abnormal conditions immediately. The customer service team uses the information fed back by the operation and maintenance system, combined with the fault code of the charging pile, to quickly identify the type of fault and its potential causes by comparing and analyzing the codes on the display screen, thereby effectively solving the user's charging problem. However, when there is no fault code or it is difficult to accurately locate the problem, the customer service staff has to generate an operation and maintenance work order and assign an operation and maintenance engineer to go to the site for investigation. This process often prolongs the problem solving time and reduces the user's experience satisfaction.

[0004] 2. When users encounter charging pile failures or problems, they submit fault feedback through the operation or operation and maintenance system front-end software. The feedback covers information such as problem description, occurrence time, and user's expected solution. The operation and maintenance system will automatically generate work orders and assign them to operation and maintenance personnel according to preset rules and priorities. After receiving the work order, the operation and maintenance personnel will contact the user through preset communication channels (such as telephone, email, instant messaging tools, etc.) to further confirm the specific situation of the problem. For problems that can be solved through remote guidance, the operation and maintenance personnel will use remote guidance technical means to guide users to operate. For complex faults that require on-site intervention, the operation and maintenance personnel will arrange a trip to the user's site to troubleshoot and repair the problem.

[0005] The problems existing in the prior art are as follows: 1. At present, the charging operation and maintenance system uses the Internet of Things technology to monitor the various parameters of the charging pile in real time through the built-in sensors and monitoring equipment of the charging pile. The data acquisition layer is responsible for collecting real-time data and transmitting it to the application layer for display and simple reminder notifications. It fails to effectively combine the real-time data of the charging pile for operation and maintenance analysis, and there is no intelligent business analysis and processing function in the implementation class.

[0006] 2. The current operation and maintenance system has significant limitations when dealing with abnormalities in charging piles. The main problem is that it relies too much on the status information uploaded by the charging piles. Once the charging pile does not trigger the alarm state, even if there is an actual problem of being unable to charge, the operation and maintenance system cannot actively intervene to deal with it. Therefore, the system still needs to manually confirm alarms and abnormal conditions, and cannot automatically and effectively send notifications to users or provide solutions. Summary of the invention

[0007] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a guidance method for intelligent operation and maintenance of charging piles to detect charging anomalies, which can achieve efficient data stream processing and provide valuable information and insights to support operation and maintenance business decisions and optimize operation and maintenance.

[0008] The technical solution adopted by the present invention is: a guiding method for detecting charging anomalies by intelligent operation and maintenance of a charging pile of the present invention, the method is applied in an intelligent charging system, the intelligent charging system includes a data communication server, an intelligent operation and maintenance model, and an intelligent terminal device, the intelligent terminal device independently connects and communicates with the data communication server and the intelligent operation and maintenance model, and the intelligent terminal device, the intelligent operation and maintenance model, and the data communication server provide data collection services and business function realization; The intelligent terminal device includes a charging pile, which is used to initiate a connection request to the data communication server and the intelligent operation and maintenance model, and after the data communication server and the intelligent operation and maintenance model accept the connection request, send the data to be sent to the data communication server and the intelligent operation and maintenance model; The data communication server includes a charging operation and maintenance system for concurrent non-blocking Socket communication connections, data processing and business processing functions; The intelligent operation and maintenance model includes charging operation and maintenance large model training, which is used to conduct in-depth analysis of the collected data and use data analysis and AI algorithms to accurately diagnose the type and location of faults.

[0009] The guidance method of intelligent operation and maintenance of charging piles to detect charging anomalies includes the following steps: Step 1: Monitor the fault status of the charging pile equipment in real time, use the self-check function of the charging pile to further confirm the fault status, and the charging pile equipment sends the real-time fault status type; Step 2: Establish a charging pile fault corpus, compare and identify the fault codes displayed on the charging pile one by one in combination with the fault code manual of the charging pile equipment, and organize the fault type and possible cause QA corpus; Step 3: Build an intelligent operation and maintenance model, strengthen the supervised training of the model's self-iteration, supervise the model's answer results to be close to the answers in the previously sorted fault corpus, and filter out any answer that does not match the answer in the fault corpus; Step 4: Build an intelligent operation and maintenance model and design the main model Actor to train the target language model; Step 5: Build an intelligent operation and maintenance model and design the main model Critic, which serves as the expected return; Step 6: Build an intelligent operation and maintenance model and design the main model Reward to calculate the actual benefits; Step 7: Build an intelligent operation and maintenance model and design the main model Reference to add "constraints" to the language model to prevent the language model from being distorted; Step 8: Execute the learning process of the abnormal charging problem. First, the fault state combination QA corpus of the charging pile is passed to the Actor as input information to form the problem to be solved; the Actor then generates a response. In this process, for each word in the response, the probability p(token|context) of generating the word is calculated according to its context. This probability value is recorded as old_log_prob and stored in the form of a tensor, whose length matches the length of the response. Step 9: Input the original question and the generated answer into Reference. In the original framework, the model will output the probability distribution of each word (token) in the answer based on the context, recorded as ref_log_prob, which is a tensor with the same length as the answer. Then, pass this pair of inputs to Reward, and the model will evaluate and return a score representing the actual benefit. At the same time, the input is also provided to Critic, which calculates an expected benefit for each word (token) in the answer, where the expected benefit of the i-th word is recorded as values[i]. Step 10: Calculate Reward to obtain high-scoring sentences in the QA corpus that meet the token distribution of SFT; Step 11: The actual benefit of the Actor generating the i-th token is the sum of all corpus scores that can be obtained from the generation of the i-th token to the generation of the N-th token. Use return to represent the sentence with the highest score. Step 12: After calculating the actual and expected benefits, strengthen the dominant action. In Actor, generating a word (token) according to the context is an action. If a word (token) is generated in the context and the advantage of this action is high, then the probability of generating the word (token) should be increased, that is, the value of p(token|context) should be increased. Actor models P(toekn|context), designs a loss function for Actor, and optimizes the loss function to achieve the strengthening of dominant actions. Step 13: Critic calculates an expected return for each word (token) in the response. The i-th expected return is recorded as values[i]. Critic is inaccurate, so the loss function MSE is used to measure the gap between the expected return of Critic and the actual return. The Critic is upgraded while the Actor is improved. Step 14: Deploy the trained model to the data communication server, and provide the corresponding SDK or API interface when the charging operation and maintenance system loads the intelligent operation and maintenance model; Step 15: The charging operation and maintenance system is combined with the intelligent operation and maintenance model to analyze the abnormal charging status data when the charging pile is connected to the electric vehicle, and identify specific abnormal problems and their corresponding treatment solutions through model training; Step 16: The charging operation and maintenance system uses a front-end applet or APP to provide charging users with free conversation functions, real-time message push and SMS reminder services; in addition, the system also actively notifies the operation and maintenance team and customer service personnel via SMS so that they can visit users in a timely manner and gain a deeper understanding of charging abnormalities.

[0010] As a further optimization, in step 2, not only the abnormal status of the charging pile and its corresponding processing methods are systematically sorted out, but also a practical and effective solution is provided for the abnormal processing of the charging pile.

[0011] As a further optimization, in step 3, operation and maintenance analysis is performed by integrating the real-time data of charging piles, realizing the intelligent capabilities that were previously lacking in business analysis and processing functions.

[0012] As a further optimization, in step 14, the intelligence of the charging operation and maintenance system is optimized, and a universal loading and docking mechanism is implemented to ensure that various docking requirements can be met at any time.

[0013] As a further optimization, in step 15, the abnormal alarm processing capability of the charging operation and maintenance system is enhanced, so that it can automatically generate effective processing solutions for abnormal problems that were previously difficult to foresee.

[0014] As a further optimization, in step 16, the charging users are effectively guided to identify abnormal problems, and the charging operation and maintenance knowledge of the charging users is cultivated so that the charging users can quickly respond to and resolve abnormalities.

[0015] As a further optimization, the intelligent terminal device includes an AC charging pile and / or a DC charging pile. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a communication diagram of a Socket connection, SDK or API based on the TCP / IP protocol, a data communication server, an intelligent terminal device, and an intelligent operation and maintenance model provided by the implementation of the present invention; Figure 2 It is a communication flow chart of the intelligent terminal device, intelligent operation and maintenance model and data communication server provided by the implementation of the present invention; Figure 3 It is a data processing flow chart of a data communication server provided by the present invention; Figure 4 It is a business flow chart of a data communication server for charging users provided by the implementation of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation methods described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] 1. Introduction to intelligent terminal devices, intelligent operation and maintenance models, and data communication server communication methods Figure 1It is a communication diagram of the data communication server, intelligent terminal device and intelligent operation and maintenance model of the present invention, such as Figure 1 As shown, including: 101-103, intelligent terminal equipment; 104, intelligent operation and maintenance model; 105, data communication server; 106-108, charging users; In steps 101 to 103, one or more DC charging piles and AC charging piles are represented; in step 104, an intelligent operation and maintenance model is represented; in step 105, a Java open source high-concurrency non-blocking Socket gateway data communication server based on zookeeper, kafka, Netty, etc. is represented; in steps 106 to 108, one or more front-end charging users using mini programs are represented; 2. Communication business flow chart of intelligent terminal equipment, intelligent operation and maintenance model and data communication server Figure 2 It is a communication service flow chart of the intelligent terminal device, intelligent operation and maintenance model and data communication server provided by the implementation of the present invention, including: 201. Establish a charging pile fault corpus, fault type and possible cause QA corpus; 202. Build an intelligent operation and maintenance model and strengthen supervised training of model self-iteration; 203, create Actor model, Reference model, Reward model, and Critic model for intensive training; 204, substitute the test data into the formula and find suitable questions for question training. Input questions to the Actor model and generate responses. Given the text in the previous part, the probability of generating the next word p(token|context) is recorded as old_log_prob, which is a tensor whose length is the length of the response; 205, the question and the generated answer are used as input to the Reference model, the Reward model, the Critic model, etc. for calculation.

[0019] 206, obtain the probability distribution ref_log_prob output under the original model and the probability distribution old_log_prob output under the Actor model.

[0020] 207, calculate the total amount that can be obtained from the generation of the i-th token to the generation of the N-th token.

[0021] 208, use return to indicate the actual answer, generate a high-priority token in the context, and increase the probability of generating the token.

[0022] 209, the data communication server loads the intelligent operation and maintenance model; 210, the intelligent terminal device sends a fault status message; 211, the data communication server stores and processes the intelligent operation and maintenance model feedback information; 3. Data communication server data processing flow chart Figure 3 This is a data processing flow chart of a data communication server provided by the present invention. The intelligent operation and maintenance model provides the source of analysis data, and the data communication server performs business processing on the data. It includes: 301, acquiring data of the intelligent terminal device in real time, and when the fault status data of the intelligent terminal device is acquired, sending the fault status data to the intelligent operation and maintenance model; 302, under the premise of step 301, obtain the current fault processing status information fed back by the intelligent operation and maintenance model, and record the time point of the fault status data, pile information and corresponding processing details; 303, classify and process the acquired fault status, pile information and processing details information, and perform statistical analysis on the data; 304, the data communication server creates a data primary key, sets data associations such as notification method, notification frequency, and notification group, and configures the business process of charging abnormality guidance; 4. Data communication server service flow chart for charging users Figure 4 The data communication server provided by the present invention is a service flow chart for charging users, including: 401, the data communication server sends the key abnormal status parameters to the data communication server through the charging pile specific communication protocol (such as CAN bus, Modbus, TCP / IP, etc.) based on the Socket connection of the TCP / IP protocol. The data communication server stores the abnormal status information in the MySQL database and pushes the abnormal status information to other interfaces.

[0023] 402, the data communication server calls the intelligent operation and maintenance model through the SDK or API interface to push the abnormal status information of the charging pile. The model is called through the interface to predict the abnormal processing method. The data communication server stores the processing method in the MySQL database. The front-end applet obtains the processing method pushed by the data communication server and creates a free dialogue page to display the abnormal processing method.

[0024] 403, the front-end applet displays the charging pile abnormal status processing method and asks the charging user to feedback the help situation. When the charging user selects help, the front-end applet calls the POST interface to record the charging user's abnormal status processing process data. When the charging user selects no help, the input box of the free dialogue page is released for the user to enter a detailed description.

[0025] 404, the front-end applet collects the user input data and sends it to the data communication server. After the intelligent operation and maintenance model makes a prediction, the result is returned to the front-end applet through the interface for feedback.

[0026] 405, the charging user closes the free dialogue page of the front-end applet, ends the charging pile exception processing, and the front-end applet prompts whether help is still needed for this exception feedback.

[0027] 406, the data communication server will call the SMS module sendTemplateSMS interface to send a SMS message to notify the operation and maintenance personnel and customer service personnel, and the relevant personnel will manually assist the charging user to handle the abnormal status.

[0028] 407, the front-end applet initiates a POST interface request to call the back-end service to record the abnormal conditions encountered by the charging user during the charging process and the corresponding processing data. These data cover a variety of abnormal conditions such as charging interruption, current and voltage abnormalities, equipment failure, etc. The collected data is transmitted to the intelligent operation and maintenance model in real time for in-depth analysis and intensive training. It continuously learns from actual cases and optimizes its decision-making logic, so that it can more accurately predict potential problems and automatically adjust the operation and maintenance strategy.

[0029] The present invention has the following beneficial effects: 1. The present invention uses ZooKeeper for distributed coordination, Kafka for efficient data stream processing, and Java open source high-concurrency non-blocking Socket frameworks such as Netty to build a data communication server specifically serving smart terminal devices and smart operation and maintenance models. The server is designed with a data communication interface that can seamlessly connect to various smart terminals. Through precise business logic processing, it deeply mines and uses data for calculation and analysis, aiming to maximize the value of data and provide strong data support and decision-making basis for smart operation and maintenance models.

[0030] 2. When an electric vehicle encounters a problem with the charging plug, it is often difficult to quickly determine whether it is a vehicle fault or a charging pile problem. To this end, the operation and maintenance system and intelligent operation and maintenance model are introduced. They not only help charging users to intuitively understand the problem, but also assist in fault analysis, significantly reducing the difficulty of operation and maintenance and shortening the processing time. In addition, the addition of a charging pile self-check function for charging abnormalities allows users to more clearly understand the specific causes and troubleshooting steps of charging abnormalities, thereby effectively reducing the workload of operation and maintenance and reducing operation and maintenance costs.

[0031] 3. Call the front-end interface to start the business process, and push messages to users through SMS and subscription message interfaces to optimize operation and maintenance and improve services. At the same time, the intelligent operation and maintenance model calls the front-end free dialogue page to handle abnormal status of smart terminal devices through intelligent manual assistance. Through insights and support for manual processing of charging users, we can maximize the potential of data and provide valuable information and insights to support operation and maintenance business decisions and optimize operation and maintenance.

[0032] Although the embodiments of the present invention are described with practical solutions, they do not constitute limitations on the meaning of the present invention. For those skilled in the art, it is obvious to modify the implementation scheme and combine it with other solutions based on this description.

Claims

1. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies, the method is applied in an intelligent charging system, the intelligent charging system includes a data communication server, an intelligent operation and maintenance model, and an intelligent terminal device, the intelligent terminal device independently connects and communicates with the data communication server and the intelligent operation and maintenance model, and the intelligent terminal device, the intelligent operation and maintenance model, and the data communication server provide data collection services and business function implementation; The intelligent terminal device includes a charging pile, which is used to initiate a connection request to the data communication server and the intelligent operation and maintenance model, and after the data communication server and the intelligent operation and maintenance model accept the connection request, send the data to be sent to the data communication server and the intelligent operation and maintenance model; The data communication server includes a charging operation and maintenance system for concurrent non-blocking Socket communication connections, data processing and business processing functions; The intelligent operation and maintenance model includes charging operation and maintenance large model training, which is used to conduct in-depth analysis of the collected data and use data analysis and AI algorithms to accurately diagnose the type and location of faults; The characteristic is that: the guidance method of charging pile intelligent operation and maintenance to detect charging anomalies comprises the following steps: Step 1: Monitor the fault status of the charging pile equipment in real time, use the self-check function of the charging pile to further confirm the fault status, and the charging pile equipment sends the real-time fault status type; Step 2: Establish a charging pile fault corpus, compare and identify the fault codes displayed on the charging pile one by one in combination with the fault code manual of the charging pile equipment, and organize the fault type and possible cause QA corpus; Step 3: Build an intelligent operation and maintenance model, strengthen the supervised training of the model's self-iteration, supervise the model's answer results to be close to the answers in the previously sorted fault corpus, and filter out any answer that does not match the answer in the fault corpus; Step 4: Build an intelligent operation and maintenance model and design the main model Actor to train the target language model; Step 5: Build an intelligent operation and maintenance model and design the main model Critic, which serves as the expected return; Step 6: Build an intelligent operation and maintenance model and design the main model Reward to calculate the actual benefits; Step 7: Build an intelligent operation and maintenance model and design the main model reference to add "constraints" to the language model to prevent the language model from being distorted; Step 8: Execute the learning process of the abnormal charging problem. First, the fault state combination QA corpus of the charging pile is passed to the Actor as input information to form the problem to be solved; the Actor then generates a response. In this process, for each word in the response, the probability p(token|context) of generating the word is calculated according to its context. This probability value is recorded as old_log_prob and stored in the form of a tensor, whose length matches the length of the response. Step 9: Input the original question and the generated answer into Reference. In the original framework, the model will output the probability distribution of each word (token) in the answer based on the context, recorded as ref_log_prob, which is a tensor with the same length as the answer. Then, pass this pair of inputs to Reward, and the model will evaluate and return a score representing the actual benefit. At the same time, the input is also provided to Critic, which calculates an expected benefit for each word (token) in the answer, where the expected benefit of the i-th word is recorded as values[i]. Step 10: Calculate Reward to obtain high-scoring sentences in the QA corpus that meet the token distribution of SFT; Step 11: The actual benefit of the Actor generating the i-th token is the sum of all corpus scores that can be obtained from the generation of the i-th token to the generation of the N-th token. Use return to represent the sentence with the highest score. Step 12: After calculating the actual and expected benefits, strengthen the dominant action. In Actor, generating a word (token) according to the context is an action. If a word (token) is generated in the context and the advantage of this action is high, then the probability of generating the word (token) should be increased, that is, the value of p(token|context) should be increased. Actor models P(toekn|context), designs a loss function for Actor, and optimizes the loss function to achieve the strengthening of dominant actions. Step 13: Critic calculates an expected return for each word (token) in the response. The i-th expected return is recorded as values[i]. Critic is inaccurate, so the loss function MSE is used to measure the gap between the expected return of Critic and the actual return. The Critic is upgraded while the Actor is improved. Step 14: Deploy the trained model to the data communication server, and provide the corresponding SDK or API interface when the charging operation and maintenance system loads the intelligent operation and maintenance model; Step 15: The charging operation and maintenance system is combined with the intelligent operation and maintenance model to analyze the abnormal charging status data when the charging pile is connected to the electric vehicle, and identify specific abnormal problems and their corresponding treatment solutions through model training; Step 16: The charging operation and maintenance system uses a front-end applet or APP to provide charging users with free conversation functions, real-time message push and SMS reminder services; in addition, the system also actively notifies the operation and maintenance team and customer service personnel via SMS so that they can visit users in a timely manner and gain a deeper understanding of charging abnormalities.

2. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: In step 2, not only the abnormal status of the charging pile and its corresponding processing methods are systematically sorted out, but also a practical and effective solution is provided for the abnormal processing of the charging pile.

3. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: In step 3, operation and maintenance analysis is performed by integrating the real-time data of charging piles, realizing the intelligent capabilities that were previously lacking in business analysis and processing functions.

4. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: In step 14, the intelligence of the charging operation and maintenance system is optimized, and a universal loading and docking mechanism is implemented to ensure that various docking requirements can be met at any time.

5. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: In step 15, the abnormal alarm processing capability of the charging operation and maintenance system is enhanced, so that it can automatically generate effective processing solutions for abnormal problems that were previously difficult to foresee.

6. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: In step 16, the charging users are effectively guided to identify abnormal problems, and the charging operation and maintenance knowledge of the charging users is cultivated so that the charging users can quickly respond to and resolve abnormalities.

7. A method for guiding charging pile intelligent operation and maintenance to detect charging anomalies according to claim 1, characterized in that: The intelligent terminal device includes an AC charging pile and / or a DC charging pile.

Images