Deep learning-based offshore wind turbine generator two-ticket intelligent checking method

Abstract

The invention belongs to the technical field of two-ticket checking, and particularly provides an offshore wind turbine generator two-ticket intelligent checking method based on deep learning, and the method comprises the steps: 1, data collection and preprocessing; step 2, performing data checking on the data preprocessed in the step 1; step 3, face value content abnormity detection; step 4, performing electronic signature and identity verification; and 5, synchronizing check data to a wind power plant management system, and updating archive data. According to the method, a deep learning technology is utilized, a work ticket and an operation ticket of an offshore wind turbine generator are automatically checked, and data comparison, anomaly detection, video analysis and the like are carried out, so that the accuracy, integrity and consistency of ticket surface content are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of two-ticket verification, and specifically relates to an intelligent two-ticket verification method for offshore wind turbines based on deep learning. Background Art

[0002] In the wind power industry, especially offshore wind power, work permits and operation permits (also known as "two permits") are crucial for ensuring safe production and equipment management. However, the current inspection and verification of these two permits for offshore wind turbines relies heavily on manual labor, which presents the following problems: 1. Inefficiency: Manual verification requires checking a large number of work tickets and operation tickets, and often involves tedious verification steps, which is time-consuming and inefficient.

[0003] 2. Error-prone: Human factors often lead to omissions and errors, especially when complex operational processes are involved, and absolute accuracy cannot be guaranteed.

[0004] 3. Data dispersion and difficulty in tracing: The two-vote data of many wind farms are scattered and stored in different systems and formats, making it difficult to effectively integrate them, posing challenges to management and auditing.

[0005] 4. Lack of real-time performance: During the two-ticket verification and management process, the feedback time is long, and problems cannot be discovered in time, affecting the safety and efficiency of subsequent operations.

[0006] Therefore, an intelligent and automated method is needed to solve these problems, improve the efficiency and accuracy of the two-vote verification, and reduce human omissions. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide an intelligent verification method for two tickets of offshore wind turbines based on deep learning, which automatically verifies the work ticket and operation ticket of the offshore wind turbine, performs data comparison, anomaly detection, video analysis, etc. to ensure the accuracy, completeness and consistency of the ticket content.

[0008] To solve the above technical problems, the technical solution adopted by the present invention is: a two-ticket intelligent verification method for offshore wind turbines based on deep learning, comprising the following steps: Step 1: Data collection and preprocessing; Step 2: Perform data verification on the data preprocessed in step 1; Step 3: Detect abnormalities in ticket content; Step 4: Electronic signature and identity verification; Step 5: Synchronize the verification data to the wind farm management system and update the archive data.

[0009] In a preferred solution, in step 1, the collected data includes structured data and unstructured data obtained from the wind farm management system.

[0010] In a preferred solution, the structured data includes work tickets and operation tickets in Excel and CSV formats; the unstructured data includes PDF files, scanned documents and videos.

[0011] In a preferred solution, in step one, data preprocessing includes formatting, deduplication, and filling in missing items of the collected data, and then standardizing the content of the ticket.

[0012] In a preferred solution, in step 2, data verification includes text verification, image and video analysis, and time verification.

[0013] In a preferred solution, the text verification method is: using natural language processing technology to parse the text content in the work ticket and operation ticket to check whether there is any missing or non-standard content.

[0014] In a preferred embodiment, the image and video analysis includes computer vision analysis of work videos, operation videos, and safety briefing videos, using convolutional neural networks to identify equipment, operating steps, and tools in the video, and determine whether the video content is consistent with the description on the ticket.

[0015] In a preferred solution, the time verification is to use a time series analysis method to verify whether the time records in the work ticket and the operation ticket are consistent with the actual operation time, and automatically identify time sequence problems or filling errors.

[0016] In the preferred solution, in step three, an anomaly detection algorithm based on deep learning automatically identifies potential problems with the content on the ticket.

[0017] In the preferred solution, in step 4, the signature column of the work ticket and the operation ticket is authenticated by face recognition or fingerprint recognition technology; if there is a missed signature or an irregular signature, timely feedback is given and an additional signature is required.

[0018] The present invention provides a deep learning-based intelligent two-ticket verification method for offshore wind turbines, which has the following beneficial effects: 1. Improve verification efficiency: Through deep learning and automated analysis technology, manual review time is greatly reduced and the efficiency of two-ticket verification is improved.

[0019] 2. Enhanced accuracy: Automated text parsing, video analysis, and anomaly detection can significantly improve verification accuracy and avoid omissions and errors in manual inspections.

[0020] 3. Real-time feedback and early warning: During the two-ticket review process, the system can detect problems in real time and issue alarms to ensure the safety and compliance of operations.

[0021] 4. Intelligent signature and identity verification: Through technologies such as face recognition and fingerprint recognition, the integrity and legitimacy of the signature field are ensured, and the reliability of the data is improved.

[0022] 5. Integration and traceability: All data are automatically uploaded, integrated and synchronized to the wind farm management system for easy management and subsequent traceability. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] Example 1: A deep learning-based intelligent two-ticket verification method for offshore wind turbines includes the following steps: Step 1: Data collection and preprocessing.

[0025] Structured and unstructured data obtained from wind farm management systems.

[0026] The structured data includes work tickets and operation tickets in Excel and CSV formats; the unstructured data includes PDF files, scanned documents, videos, etc.

[0027] Data preprocessing includes formatting, deduplication, and filling in missing items of the collected data, and then standardizing the content of the ticket to ensure the uniformity and accuracy of the data.

[0028] Specifically, formatting includes converting structured data into JSON or database format, and extracting key fields such as ticket number, equipment number, operation steps, time information, etc.; unstructured data is converted into structured text through OCR recognition.

[0029] To fill missing items, rule-based and statistical methods are used, such as using the mean of similar historical data to fill missing fields, or using machine learning models (such as KNN and decision trees) for predictive completion. Standardization includes unifying terminology, field formats, and time formats. For example, the time format is standardized to YYYY-MM-DD HH:mm:ss, and the terminology on the bill is mapped to the State Grid or enterprise standards.

[0030] Step 2: perform data verification on the data preprocessed in step 1.

[0031] Data verification includes text verification, image and video analysis, and time verification.

[0032] The text verification method is to use natural language processing technology (NLP) to parse the text content in the work ticket and operation ticket to check whether there is any missing or non-standard content, for example, whether the safety measures, operation steps, equipment names, etc. are accurate.

[0033] Specific implementation methods include: using word segmentation (such as Jieba) to segment the ticket content, combining named entity recognition (NER) to identify key entities (such as equipment name, time, and operator); then building a rule library or using deep learning models (such as BERT) to perform semantic comparison and normative judgment to determine whether there is any missing key information, inappropriate wording, or description that does not comply with the standards.

[0034] The image and video analysis includes computer vision analysis of work videos, operation videos and safety briefing videos, using convolutional neural networks to identify equipment, operation steps and tools in the video, and determine whether the video content is consistent with the description on the ticket.

[0035] The specific implementation method is: use convolutional neural networks (such as ResNet, YOLOv5) to perform image target recognition on the video frame by frame, extract key features such as equipment category, tool type, and personnel action, and analyze the operation sequence through a timing model (such as LSTM). The actual execution steps of the video are compared with the content of the ticket to determine whether there are inconsistencies or missed operations.

[0036] Use computer vision technology to analyze work videos and safety briefing videos to ensure that the video content is consistent with the descriptions on work tickets and operation tickets, thereby avoiding false reporting.

[0037] The time verification is to use the time series analysis method to verify whether the time records in the work ticket and the operation ticket are consistent with the actual operation time, and automatically identify time sequence problems or filling errors.

[0038] Specific methods include: building a time series model (such as ARIMA, Prophet) to model the time intervals of each step in the operation ticket; at the same time, introducing logical rules, such as "the operation start time must be earlier than the operation end time", "the video length should cover all operation steps", etc., through model prediction and rule comparison to determine whether the time field is reasonable. If there are anomalies such as time inversion and time span abnormalities, the system will automatically mark and prompt for correction.

[0039] Step 3: Detect abnormalities in ticket content.

[0040] The deep learning-based anomaly detection algorithm automatically identifies potential problems with the content on the ticket, such as incorrect voltage levels, missing signatures, and unrecorded operation times for key equipment, and provides corresponding correction suggestions.

[0041] Provide real-time feedback during the verification process. If any problems are found, warnings will be automatically issued and problem reports will be generated to remind operators to make corrections.

[0042] This anomaly detection algorithm builds an anomaly recognition system based on an autoencoder or isolation forest model. This system trains standardized bill data into normal samples, performs reconstruction error analysis or feature space partitioning on the input data, and automatically identifies data items that deviate from the standard. Alternatively, a graph neural network (GNN) model can be used to model the relationships between bill fields and identify potential anomaly patterns.

[0043] Step 4: Electronic signature and identity verification.

[0044] For the signature column of work tickets and operation tickets, identity verification is performed through face recognition or fingerprint recognition technology to ensure the legitimacy and integrity of the signature.

[0045] Authenticate signatures through facial or fingerprint recognition technology, automatically check signature integrity and compliance, and avoid missed or fake signatures.

[0046] If there is any missing signature or irregular signature, timely feedback will be given and additional signature will be requested.

[0047] Step 5: Synchronize the verification data to the wind farm management system and update the archive data.

[0048] After the verification is completed, the system automatically generates a detailed verification report, which includes the verification content, problems found, correction suggestions and related operation records.

[0049] All verification data and feedback results are automatically synchronized to the wind farm management system to ensure real-time updating and accuracy of archival data.

[0050] The system automatically generates problem reports and provides feedback during the verification process, ensuring that operators can correct problems in a timely manner and reduce safety hazards.

[0051] The present invention applies deep learning and NLP technology to the text verification of two tickets, automatically detecting omissions and irregular items in the ticket content, and automatically checking the consistency with the actual operation process.

[0052] Through time series analysis and deep learning anomaly detection methods, operation time, equipment operation records, etc. are automatically verified to ensure that the time on the ticket is consistent with the actual operation and avoid filling errors.

[0053] Example 2: Taking the "two tickets" data of a certain offshore wind farm in the fourth quarter of 2024 as an example, the two-ticket intelligent verification method of offshore wind turbines based on deep learning proposed in this invention is used for system deployment and testing.

[0054] The test data includes: 1) Structured data: a total of 1,200 work tickets and operation tickets, all in Excel / CSV format; 2) Unstructured data: including 870 workflow PDF documents, 265 scanned images, and 180 video clips; 3) User identity data: Covers facial images and fingerprint samples of 76 operators in total.

[0055] Compared with manual verification, the efficiency comparison table is shown in Table 1.

[0056]

[0057] From the above data analysis, the verification time of the technical solution of the present invention is shortened to 0.9 minutes per bill, and the error rate is also greatly reduced, which proves the operability of this method in the intelligent verification of two bills.

[0058] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A deep learning-based intelligent two-ticket verification method for offshore wind turbines, characterized in that: The following steps are involved: Step 1: Data collection and preprocessing; Step 2: Perform data verification on the data preprocessed in step 1; Step 3: Detect abnormalities in ticket content; Step 4: Electronic signature and identity verification; Step 5: Synchronize the verification data to the wind farm management system and update the archive data.

2. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 1 is characterized in that: In the step 1, the collected data includes structured data and unstructured data obtained from the wind farm management system.

3. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 2 is characterized in that: The structured data includes work tickets and operation tickets in Excel and CSV formats; the unstructured data includes PDF files, scanned documents and videos.

4. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 1 is characterized in that: In the step 1, data preprocessing includes formatting, removing duplicates, filling in missing items of the collected data, and then standardizing the content of the ticket.

5. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 1 is characterized in that: In the step 2, data verification includes text verification, image and video analysis, and time verification.

6. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 5 is characterized in that: The text verification method is: using natural language processing technology to parse the text content in the work ticket and operation ticket to check whether there is any missing or non-standard content.

7. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 5 is characterized in that: The image and video analysis includes computer vision analysis of work videos, operation videos and safety briefing videos, using convolutional neural networks to identify equipment, operation steps and tools in the video, and determine whether the video content is consistent with the description on the ticket.

8. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 5 is characterized in that: The time verification is to use the time series analysis method to verify whether the time records in the work ticket and the operation ticket are consistent with the actual operation time, and automatically identify time sequence problems or filling errors.

9. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 1 is characterized in that: In step three, a deep learning-based anomaly detection algorithm automatically identifies potential problems with the ticket content.

10. The method for intelligent verification of two tickets of offshore wind turbines based on deep learning according to claim 1, characterized in that: In step 4, the signature column of the work ticket and the operation ticket is authenticated by using facial recognition or fingerprint recognition technology; if there is a missing signature or an irregular signature, timely feedback is provided and additional signatures are required.

Images