AI-driven big data fusion wind farm booster station anti-misoperation locking system

Abstract

This invention discloses an AI-driven big data fusion-based anti-misoperation interlocking system for wind farm booster stations, relating to the field of anti-misoperation interlocking technology for wind farm booster stations. It includes a data acquisition module that collects real-time electrical parameters, operating status, and environmental parameters of the equipment, and uploads them after preprocessing; a big data fusion storage module that categorizes and stores multiple types of data according to a hierarchical strategy, optimizing access and retrieval; an AI intelligent analysis and decision-making module that integrates multi-source data to generate precise interlocking strategies; an anti-misoperation interlocking execution module that executes interlocking in a three-level mode and provides status feedback; a communication interaction module that achieves internal and external data interaction through a dual-mode architecture; a status monitoring module that monitors the operating status of each module and identifies faults; and an emergency handling module that initiates a response mechanism and restores operation after self-checking following fault resolution. This invention relies on AI and big data fusion technology, dynamically adapting interlocking thresholds to equipment and environmental conditions; supporting unattended operation and intelligent maintenance, providing reliable assurance for the safe operation of wind farm booster stations.

Description

Technical Field

[0001] This invention relates to the field of anti-misoperation interlocking technology for wind farm booster stations, and in particular to an AI-driven big data fusion anti-misoperation interlocking system for wind farm booster stations. Background Technology

[0002] As the core hub for power collection, transformation, and transmission, the operational safety of wind farm booster stations directly determines the overall power supply stability and equipment lifespan of the wind farm. Anti-misoperation interlocking systems are crucial technologies for preventing substation malfunctions, eliminating short-circuit faults, and protecting equipment and personnel safety. However, with the increasing capacity of individual wind turbines, the growing complexity of substation equipment, and the frequent occurrence of extreme weather, traditional anti-misoperation interlocking systems have gradually revealed numerous technical bottlenecks. Traditional systems often rely on fixed logic rules for interlocking control, lacking dynamic adjustment capabilities for threshold parameters. This makes them unsuitable for dynamic scenarios such as equipment aging, changing operating conditions, and environmental fluctuations. For example, if the insulation performance of equipment declines after long-term operation, continuing to use the initial interlocking threshold can easily lead to interlocking failure or over-interlocking. Furthermore, traditional systems have limited data acquisition dimensions, often confined to electrical parameters. They lack sufficient integration of multi-source data, including environmental parameters, equipment mechanical status, and operational behavior characteristics. Data processing is limited to simple filtering, failing to uncover potential risk correlations behind the data, resulting in a lack of comprehensive data support for interlocking decisions.

[0003] At the application level of AI technology, while some existing systems incorporate AI models, these models are mostly statically deployed and lack adaptive optimization mechanisms. After long-term operation, they are prone to problems such as decreased inference accuracy and increased false positive rates due to changes in operating conditions and equipment characteristic drift. Furthermore, model updates rely on manual intervention, resulting in delayed responses. In addition, traditional systems often employ a single-device independent interlocking mode, failing to consider the complex electrical connections and operational logic relationships between substation equipment. Even after a single device is interlocked, related devices may still be at risk of misoperation, leading to incomplete interlocking protection. Regarding environmental adaptability, existing systems do not adequately consider the coupling relationship between environmental factors such as wind speed, icing, and lightning and equipment failures. They only trigger simple interlocks in extreme environments, lacking early warning and dynamic protection capabilities, making it difficult to cope with the safety challenges posed by the complex outdoor environment of wind farms. Simultaneously, operational error prevention relies solely on identity authentication and command verification, failing to effectively monitor operator behavior data such as operational trajectories and action characteristics. This makes it impossible to identify hidden risks such as unauthorized personnel imitating operations or authorized personnel making mistakes, further reducing the reliability of the interlocking system.

[0004] Insufficient coordination between equipment operation and maintenance and the interlocking system is also a prominent issue. Traditional systems only focus on real-time interlocking control, failing to integrate equipment lifecycle data. They cannot dynamically optimize interlocking strategies based on equipment health status, nor can they provide targeted suggestions for operation and maintenance. This leads to a disconnect between the interlocking system and equipment operation and maintenance. Safety hazards caused by equipment aging cannot be proactively avoided through interlocking strategy adjustments, while also increasing the blind spots and costs of operation and maintenance. These combined problems result in insufficient safety protection capabilities of traditional anti-misoperation interlocking systems under complex operating conditions, with high false positive and false negative rates, making it difficult to meet the high-precision, intelligent, and comprehensive anti-misoperation interlocking requirements of modern wind farm booster stations. Summary of the Invention

[0005] The present invention proposes an AI-driven big data fusion-based wind farm booster station anti-misoperation interlocking system to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an AI-driven big data fusion-based anti-misoperation interlocking system for wind farm booster stations, comprising the following modules: Data acquisition module: integrates voltage, current, temperature, position sensors and environmental monitoring units to collect electrical parameters, operating status parameters and environmental parameters of key equipment in the substation in real time, and uploads them at a preset frequency after preprocessing using edge computing technology; Big Data Fusion Storage Module: Adopting a distributed storage architecture combined with a hierarchical storage strategy, real-time data, historical operation data, equipment ledger data, and operation record data are classified and stored in corresponding data areas, and storage resource allocation is dynamically adjusted. AI intelligent analysis and decision-making module: Equipped with deep learning, reinforcement learning models and rule reasoning engine, it receives preprocessed data and integrates historical data to analyze equipment operating status, operational compliance and potential risks, and generate permission-based anti-misoperation lockout strategies. Anti-misoperation interlocking execution module: Receives interlocking commands, drives relevant actuators to perform operations in three levels of interlocking modes: emergency, normal, and early warning, and synchronously provides feedback on the execution status; Communication interaction module: Adopts a dual-mode architecture of industrial Ethernet and wireless communication, and can transmit data bidirectionally with external platforms; Status monitoring module: monitors the operating status of each module in real time, identifies faults and anomalies, and marks the location and type of faults; Emergency handling module receives fault signals and emergency risk warnings, activates the emergency response mechanism, executes self-test procedures after troubleshooting, and restores normal system operation after verifying that the function has been restored.

[0007] Furthermore, it also includes a dynamic interlocking threshold optimization module, which dynamically adjusts the interlocking judgment threshold based on the real-time operating status of the equipment and environmental changes. The calculation expression is as follows: ,in This is the dynamic locking threshold. This is the weighting coefficient for equipment health. This provides a real-time health measurement value for the device. For environmental impact weighting coefficients, To comprehensively quantify environmental factors, This is the operation frequency weighting coefficient. This is the normalized value of the recent operation frequency of the equipment. + + =1.

[0008] Furthermore, it also includes an operational risk assessment module, which quantitatively assesses the operational risk level by combining equipment status, operation type, environmental conditions, and historical fault data. The calculation expression is as follows: ,in This is the operational risk level value. This is a weighting coefficient for operational complexity. For quantification values ​​of complex operation types, This is the equipment defect weighting coefficient. This represents the cumulative quantitative value of the current defects in the equipment. For environmental risk weighting coefficients, Quantification of environmental risk level, + + =1, based on the risk level value, it is divided into three levels: low risk, medium risk, and high risk, corresponding to different lockout response strategies.

[0009] Furthermore, it also includes a permission management module, which adopts a role-based hierarchical authorization mechanism to divide system operation permissions into administrator permissions, operation and maintenance personnel permissions, monitoring personnel permissions, and temporary authorization permissions. Different permissions are assigned to different roles, and the identity of the operator is confirmed through a combination of identity authentication, password verification, and biometric identification. All permission-related operation behavior logs are recorded.

[0010] Furthermore, it also includes a data encryption module, which encrypts sensitive data during internal data transmission and external data interaction. It uses a combination of symmetric and asymmetric encryption algorithms to encrypt the core data of the device, generate encrypted data packets, and attach data verification codes.

[0011] Furthermore, it also includes an AI model adaptive adjustment module, which monitors in real time the three core performance indicators of the deep learning model and reinforcement learning model in the AI ​​intelligent analysis and decision-making module: inference accuracy, decision response time, and fault identification accuracy. It calculates the model confidence quantification value, and when the confidence is lower than the preset threshold, it automatically extracts recent high-value data, including abnormal working condition data, misjudgment correction data, and new scenario data, to construct an incremental training set. It uses transfer learning algorithms to retrain the model in a lightweight manner, dynamically adjusts the number of network layers, neurons, and weight parameters of the model, and retains a backup of the historical best model parameters.

[0012] Furthermore, it also includes a cross-device linkage interlocking module, which constructs a correlation map of key equipment in the booster station, clarifies the electrical connection relationships and operational logic dependencies between equipment, establishes a linkage interlocking rule base, and when a single device triggers interlocking, it simultaneously analyzes the operating status and operational intent of the associated devices, determines the scope of linkage interlocking according to the principles of main device priority, safety criticality priority, and operation sequence priority, drives the associated devices to perform collaborative interlocking operations, and supports custom configuration of linkage logic.

[0013] Furthermore, it also includes an environment and equipment coupling early warning module, which establishes a coupling correlation model between environmental parameters and equipment failures, integrates five types of environmental data (wind speed, humidity, temperature, icing thickness, lightning activity) with historical correlation data of equipment insulation performance, mechanical characteristics, and electrical parameters, and uses a gradient boosting algorithm to mine coupling patterns and calculate the environment and equipment coupling risk coefficient in real time.

[0014] Furthermore, it also includes an operation trajectory error prevention and verification module, which integrates a visual acquisition unit and a motion sensing unit to collect the operator's operation trajectory, action sequence, operation force characteristics, and operation command input rhythm in real time. It extracts operation behavior feature vectors through AI algorithms and compares them with the operator's historical compliant operation feature library and the system's preset standard operation feature templates to calculate the behavior similarity. When the similarity is lower than a set threshold, it is judged as an abnormal operation trajectory and triggers a secondary verification mechanism, including operation command review, identity secondary authentication, and operation intent confirmation. At the same time, it automatically updates the operator behavior feature library.

[0015] Furthermore, it also includes a full lifecycle health management module, which integrates equipment operation data, fault records, maintenance records, and lockout event records to establish a full lifecycle health assessment model for equipment. It uses a multi-dimensional weighted scoring method to calculate the real-time health of the equipment, and combined with equipment parameters, predicts the remaining safe operating time of the equipment. It dynamically optimizes the lockout strategy based on the health decay trend, and generates equipment maintenance suggestions including maintenance timing, maintenance items, and spare parts replacement warnings, thus deeply integrating the lockout system with equipment operation and maintenance.

[0016] Compared with existing technologies, the beneficial effects of this invention are: The AI-driven big data fusion wind farm booster station anti-misoperation interlocking system of the present invention achieves a comprehensive technical upgrade by addressing the technical pain points of traditional systems, significantly improving the intelligence level, safety reliability and environmental adaptability of the anti-misoperation interlocking, and providing a strong guarantee for the safe and stable operation of wind farm booster stations.

[0017] Regarding the accuracy of interlocking decisions, the dynamic interlocking threshold optimization module dynamically adjusts the threshold based on multiple dimensions, including equipment health, environmental factors, and operation frequency. This ensures that interlocking decisions better align with the actual operating status and environmental conditions of the equipment, avoiding interlocking failures or over-interlocking caused by fixed thresholds. The AI ​​model adaptive adjustment module monitors model performance in real time and dynamically optimizes the model through incremental training and transfer learning. This ensures the model maintains high inference accuracy and a low false positive rate over the long term, improving the intelligence and timeliness of interlocking decisions. The application of multi-source data fusion and big data processing technologies integrates multi-dimensional data such as electrical parameters, environmental parameters, equipment status, and operational behavior. It also uncovers data-related risks, providing comprehensive and in-depth data support for interlocking decisions and significantly improving the scientific nature of the decisions.

[0018] The comprehensiveness and reliability of security protection are significantly enhanced. The cross-device linkage interlocking module constructs a device association map, enabling coordinated interlocking of related devices and eliminating protection vulnerabilities caused by independent interlocking of single devices, forming a comprehensive interlocking protection network. The environment-device coupling early warning module uncovers the coupling patterns between the environment and equipment failures, upgrading from passive interlocking to proactive early warning, and proactively avoiding security risks brought about by complex environments. The operation trajectory error prevention and verification module monitors operational behavior characteristics to identify and perform secondary verification of abnormal operations, effectively preventing hidden risks such as misoperation and operational errors after successful identity authentication, further strengthening the operational safety defense line.

[0019] In terms of operation and maintenance collaboration and cost-effectiveness, the full lifecycle health management module integrates equipment lifecycle data to achieve dynamic assessment of equipment health and prediction of remaining operating time. It dynamically optimizes interlocking strategies while generating targeted operation and maintenance suggestions, deeply integrating the interlocking system with equipment operation and maintenance. This not only proactively avoids safety hazards caused by equipment aging but also reduces blind operation and maintenance costs. The access control and data encryption modules ensure operational compliance and data security, preventing unauthorized operations and data security risks, and ensuring stable and reliable system operation. Overall, the system achieves an integrated intelligent upgrade of interlocking control, risk warning, and operation and maintenance collaboration through AI-driven and big data integration. This results in faster interlocking response, more accurate decision-making, more comprehensive protection, and more efficient operation and maintenance, fully meeting the anti-misoperation interlocking needs of modern wind farm booster stations under complex operating conditions. Attached Figure Description

[0020] Figure 1 This is a schematic block diagram of the AI-driven big data fusion wind farm booster station anti-misoperation interlocking system proposed in this invention; Figure 2 Line graph showing the change of dynamic blocking threshold of onshore substations under different ambient temperatures; Figure 3 A bar chart showing the recognition rate for different fault types; Figure 4 Grouped bar charts showing the interlocking malfunction rate for different years of operation; Figure 5 A pie chart showing the breakdown of operation and maintenance costs for wind farm booster stations. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0023] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0024] Reference Figures 1 to 5 A wind farm booster station anti-misoperation interlocking system driven by AI and big data integration, comprising the following modules: The data acquisition module integrates voltage sensors, current sensors, temperature sensors, position sensors, and environmental monitoring units. It collects electrical parameters, operating status parameters, and environmental parameters such as wind speed, humidity, and temperature of key equipment such as the main transformer, circuit breaker, disconnector, and grounding switch of the substation in real time. It uses edge computing technology to perform noise reduction, filtering, and format standardization preprocessing on the raw data and uploads the data at a preset frequency. The big data fusion storage module adopts a distributed storage architecture combined with a hierarchical storage strategy to classify and store the collected real-time data, historical operation data, equipment ledger data, and operation record data into high-speed cache area, hot data area, and cold data area. It dynamically adjusts the allocation of storage resources based on data popularity, optimizes read and write caching strategies, and establishes multi-dimensional data indexes to achieve efficient data retrieval. The AI ​​intelligent analysis and decision-making module is equipped with a deep learning model, a reinforcement learning model, and a rule reasoning engine. It receives pre-processed data from the data acquisition module in real time, integrates historical data from the big data fusion storage module, analyzes the equipment operating status, compliance of operation instructions, and potential risks, and generates accurate anti-misoperation interlocking strategies, including decision results such as allowing operation, prohibiting operation, delaying operation, and hierarchical interlocking. The anti-misoperation locking execution module receives the locking command from the AI ​​intelligent analysis and decision module, drives the electromagnetic lock, mechanical lock, intelligent measurement and control unit and other execution mechanisms to perform locking operations in three levels: emergency locking, normal locking and early warning locking, and synchronously feeds back the execution status to the AI ​​intelligent analysis and decision module; The communication interaction module adopts a dual-mode architecture of industrial Ethernet and wireless communication to realize internal data interaction between the data acquisition module, big data fusion storage module, AI intelligent analysis and decision-making module, and anti-misoperation interlocking execution module. At the same time, it can perform bidirectional data transmission with the wind farm monitoring center, equipment operation and maintenance platform, and remote control terminal to ensure real-time data synchronization and efficient transmission of instructions. The status monitoring module monitors the operating status of each module in the system in real time, including data acquisition accuracy, storage device health, communication link stability, AI model running efficiency, and interlocking actuator response speed, and identifies module faults and anomalies and marks the fault location and type. The emergency response module receives fault signals from the status monitoring module and emergency risk warnings from the AI ​​intelligent analysis and decision-making module, and activates the corresponding emergency response mechanism, including fault module isolation, backup module switching, emergency unlocking, tiered alarms, and data backup. After troubleshooting, it executes the system self-test program to verify that the functions of each module have returned to normal before resuming normal system operation.

[0025] This invention also includes a dynamic locking threshold optimization module, which dynamically adjusts the locking judgment threshold based on the real-time operating status of the equipment and environmental changes. It achieves precise threshold adaptation through multi-factor coupled calculation, and the calculation expression is as follows: ,in This is the dynamic locking threshold. This is the weighting coefficient for equipment health. The real-time health metric values ​​for the device range from 0 to 1. For environmental impact weighting coefficients, The comprehensive quantitative value of environmental factors ranges from 0 to 1. This is the operation frequency weighting coefficient. The normalized value for the recent operation frequency of the equipment ranges from 0 to 1. + + =1. This calculation allows the interlocking threshold to be dynamically adjusted as equipment ages, the environment deteriorates, and the frequency of operation increases, thereby improving the accuracy and adaptability of interlocking determination.

[0026] This invention also includes an operational risk assessment module, which quantitatively assesses the operational risk level by combining equipment status, operation type, environmental conditions, and historical fault data, providing support for interlocking decisions. The calculation expression is as follows: ,in The operational risk level value ranges from 0 to 10. This is a weighting coefficient for operational complexity. The quantification value for operation type complexity ranges from 1 to 5, corresponding to different operation types. This is the equipment defect weighting coefficient. The cumulative quantitative value of the current defects of the equipment ranges from 0 to 3. For environmental risk weighting coefficients, The environmental risk level quantification value ranges from 0 to 2. + + =1, based on the risk level value, it is divided into three levels: low risk, medium risk, and high risk, corresponding to different lockout response strategies.

[0027] This invention also includes a permission management module, which adopts a role-based hierarchical authorization mechanism to divide system operation permissions into administrator permissions, operation and maintenance personnel permissions, monitoring personnel permissions, and temporary authorization permissions. Different permissions such as data viewing, operation execution, parameter configuration, and fault handling are assigned to different roles. The identity of the operator is confirmed through a combination of identity authentication, password verification, and biometric identification. All permission-related operation behavior logs are recorded to achieve operation traceability.

[0028] This invention also includes a data encryption module, which encrypts sensitive data during internal data transmission and external data interaction. It employs a combination of symmetric and asymmetric encryption algorithms to encrypt core data such as equipment operating parameters, interlocking commands, and operation records, generating encrypted data packets and attaching data verification codes. The receiver decrypts the data after verifying the verification codes, thus preventing data tampering, theft, and leakage.

[0029] This invention also includes an AI model adaptive adjustment module, which monitors in real time the three core performance indicators of the deep learning model and reinforcement learning model in the AI ​​intelligent analysis and decision-making module: inference accuracy, decision response time, and fault identification accuracy. It calculates the model confidence quantification value. When the confidence is lower than a preset threshold, it automatically extracts recent high-value data, including abnormal working condition data, misjudgment correction data, and new scenario data, to construct an incremental training set. It uses a transfer learning algorithm to retrain the model in a lightweight manner, dynamically adjusting the number of network layers, neurons, and weight parameters of the model, while retaining a backup of the historical best model parameters. This allows the model to maintain high decision accuracy under complex working condition changes and equipment aging scenarios, reducing the lock-in misjudgment caused by model solidification.

[0030] This invention also includes a cross-equipment linkage interlocking module, which constructs a correlation diagram of key equipment in the substation, clarifies the electrical connection relationships and operational logic dependencies between equipment such as main transformers, circuit breakers, disconnect switches, and grounding switches, and establishes a linkage interlocking rule base. When a single device triggers interlocking, the operating status and operational intent of the associated devices are analyzed simultaneously. The scope of linkage interlocking is determined according to the principles of main device priority, safety criticality priority, and operation sequence priority, driving the associated devices to perform coordinated interlocking operations. This reduces the risk of system short circuits and equipment damage caused by incomplete interlocking of a single device. At the same time, it supports custom configuration of linkage logic to adapt to the equipment layout and operating mode of different substations.

[0031] This invention also includes an environment-equipment coupling early warning module, which establishes a coupling correlation model between environmental parameters and equipment faults. It integrates five types of environmental data—wind speed, humidity, temperature, icing thickness, and lightning activity—with historical correlation data of equipment insulation performance, mechanical characteristics, and electrical parameters. It uses a gradient boosting algorithm to mine coupling patterns and calculates the environment-equipment coupling risk coefficient in real time. When the coefficient exceeds the early warning threshold, it adjusts the interlocking judgment sensitivity in advance and strengthens the operation verification logic under high-risk environments. For example, it raises the interlocking threshold for disconnecting switch operation in icing weather and prohibits non-emergency operations during lightning activity, thus achieving an upgrade from passive interlocking to active early warning.

[0032] This invention also includes an operation trajectory error prevention and verification module, which integrates a visual acquisition unit and a motion sensing unit to collect the operator's operation trajectory, action sequence, operation force characteristics, and operation command input rhythm in real time. The module extracts operation behavior feature vectors through AI algorithms and compares them with the operator's historical compliant operation feature library and the system's preset standard operation feature templates to calculate the behavior similarity. When the similarity is lower than a set threshold, it is determined to be an abnormal operation trajectory, triggering a secondary verification mechanism including operation command review, identity secondary authentication, and operation intent confirmation. This reduces the lockout failure caused by misoperation, unauthorized operation, and operation error, while automatically updating the operator behavior feature library to adapt to reasonable changes in operation habits.

[0033] This invention also includes a full lifecycle health management module, which integrates equipment operation data, fault records, maintenance records, and interlock event records to establish a full lifecycle health assessment model for the equipment. It uses a multi-dimensional weighted scoring method to calculate the real-time health of the equipment, and combines parameters such as the equipment's design life, operating years, and accumulated environmental stress to predict the remaining safe operating time. Based on the health degradation trend, it dynamically optimizes the interlock strategy; for example, it appropriately reduces the allowable operation frequency and increases interlock sensitivity for aging equipment, and uses a gradual adjustment of the interlock threshold during the break-in period for new equipment. Simultaneously, it generates equipment maintenance suggestions, including maintenance timing, maintenance items, and spare parts replacement warnings, deeply integrating the interlock system with equipment operation and maintenance, extending equipment lifespan, and reducing operation and maintenance costs.

[0034] The following two examples further illustrate the specific implementation of this system: Example 1: Application of booster station in large onshore wind farms This embodiment is applied to a 150MW large-scale onshore wind farm booster station, adapted to a 220kV voltage level, and includes key equipment such as 2 main transformers, 8 circuit breakers, 16 disconnect switches, and 12 grounding switches. It faces environmental challenges such as complex terrain wind shear, extreme low temperatures, and icing. All system modules and technical solutions have been fully implemented to achieve high-precision anti-misoperation interlocking and intelligent operation and maintenance collaboration.

[0035] The data acquisition module integrates 32 voltage sensors, 24 current sensors, 18 temperature sensors, 20 position sensors, and 6 environmental monitoring units. Voltage sensors collect three-phase voltage and zero-sequence voltage from each device; current sensors monitor charging / discharging current and leakage current; temperature sensors are located on key components such as device contacts and windings; position sensors provide real-time feedback on switch open / closed status; and environmental monitoring units collect data on wind speed, temperature, humidity, icing thickness, and air pressure. Edge computing technology is used to perform third-order Butterworth filtering for noise reduction and Z-score normalization on the raw data. The sampling frequency is set to 50Hz, and the pre-processed data is uploaded to the big data fusion storage module at a frequency of once per second.

[0036] The big data converged storage module adopts a distributed cluster architecture, combining a tiered strategy of SSD high-speed caching, SAS hot data storage, and SATA cold data storage. Real-time data is stored in the SSD high-speed cache, retaining 24-hour data; historical operating data and operation records for the past 3 months are stored in the hot data area; and equipment ledger data and annual fault records are stored in the cold data area. Storage resources are dynamically adjusted based on data popularity, with popular data automatically migrated to high-speed storage nodes. A multi-dimensional index is established based on device number, timestamp, and data type, and data retrieval response time is controlled within 50 milliseconds.

[0037] The AI-powered intelligent analysis and decision-making module is equipped with a CNN-LSTM deep learning model, a DQN reinforcement learning model, and a rule-based reasoning engine. The CNN-LSTM model processes time-series data to identify abnormal equipment states, the DQN model optimizes the interlocking decision strategy, and the rule-based reasoning engine integrates 120 substation operation specifications. After receiving pre-processed data in real time, it integrates nearly one year of historical operating data to analyze equipment insulation performance, compliance of operating command logic, and environmental coupling risks, generating decision results such as permitted operation and prohibited operation, with a decision response time of no more than 100 milliseconds.

[0038] After receiving the decision command, the anti-misoperation interlocking execution module drives 16 sets of electromagnetic locks, 24 sets of mechanical locks, and 8 intelligent monitoring and control units to perform operations in a three-level mode. Emergency interlocking, in case of short-circuit risk, cuts off the operating circuit and locks within 10 milliseconds; regular interlocking, in case of operational logic error, executes the interlock within 50 milliseconds; early warning interlocking, in case of impending environmental risk, issues an alarm and restricts operating permissions, while simultaneously feeding back the execution status through the communication module.

[0039] The communication interaction module adopts a dual-mode architecture of industrial Ethernet and 5G wireless communication. The industrial Ethernet enables gigabit-level data interaction between modules within the station, while the 5G module establishes a low-latency link with the wind farm monitoring center and remote operation and maintenance platform. Core data is synchronized every 500 milliseconds, and the transmission latency of interlocking commands does not exceed 20 milliseconds. The status monitoring module monitors the operating parameters of each module in real time, including data acquisition accuracy error ≤ ±0.5%, storage device IO utilization, communication link packet loss rate, AI model inference time, and interlocking actuator response speed. After identifying module faults, it marks the location and type and uploads the information to the emergency handling module.

[0040] Upon receiving a fault signal, the emergency response module will initiate fault module isolation within 300 milliseconds, switch to the backup module, and trigger tiered alarms. For minor faults, such as data acquisition accuracy drift, a text alarm will be sent; for moderate faults, such as a single communication link interruption, automatic switch to the backup link will be initiated and maintenance will be notified; for severe faults, such as AI module failure, an emergency unlocking plan will be activated and critical data will be backed up. After troubleshooting, 12 self-test items will be performed, including sensor calibration, model accuracy verification, and interlocking execution function testing. Normal operation will resume once all tests are passed.

[0041] The dynamic locking threshold optimization module is based on the formula. Calculate the threshold and set it. =0.4 =0.35 =0.25. When the main transformer health... =0.85, Comprehensive Environmental Quantitative Value =0.7 Normalized value of operating frequency at -10 degrees Celsius When =0.6, =0.4×0.85+0.35×0.7+0.25×0.6=0.34+0.245+0.15=0.735, corresponding to a 15% reduction in the locking threshold compared to the base value. The operational risk assessment module uses the formula... Calculation, setting =0.5 =0.3 =0.2, disconnector switch closing operation =3. Equipment defects =1. Environmental Risks =1, =0.5×3+0.3×1+0.2×1=1.5+0.3+0.2=2.0, which is determined to be low risk, and a routine interlocking check is performed.

[0042] The access control module is divided into four roles: administrators have full permissions, operations and maintenance personnel can perform operations and handle faults, and monitoring personnel can only view data. Temporary authorization permissions are valid for 24 hours. Identity is confirmed through fingerprint biometrics and dynamic password verification, and all operation logs are recorded and retained for one year. The data encryption module uses AES-256 symmetric encryption to process operational data, RSA-2048 asymmetric encryption to transmit locking commands, and adds a CRC32 checksum to ensure data integrity. The AI ​​model adaptive adjustment module detects that when the model confidence drops to 0.85, it extracts 500 recent abnormal operating condition data points to construct an incremental training set, uses transfer learning for lightweight retraining, adjusts the model network layers from 12 to 10, and optimizes the number of neurons to 800.

[0043] The cross-equipment linkage interlocking module constructs an equipment association map based on the electrical wiring diagram. A strong association is established between the main transformer and the circuit breakers and disconnect switches on both sides. When the main transformer is under maintenance and interlocked, the associated six devices are simultaneously interlocked. The environment-equipment coupling early warning module integrates five types of environmental data. When the icing thickness exceeds 10mm, the disconnect switch interlocking threshold is increased by 20%, prohibiting closing operations. The operation trajectory error prevention and verification module collects operator behavior through high-definition cameras and motion sensors, extracting a 30-dimensional feature vector. When the similarity is below 0.7, command verification and secondary identity authentication are triggered. The full lifecycle health management module integrates three years of equipment operation data, calculates the main transformer's health level to be 0.82, predicts the remaining safe operating time to be 5.2 years, and generates maintenance recommendations for annual insulation testing.

[0044] Table 1. Performance Comparison of Anti-misoperation Interlocking Systems for Substations in Onshore Wind Farms Table 1 clearly demonstrates the technical advantages of the system of this invention. Traditional systems suffer from high interlocking misjudgment rates, cannot provide early warnings of risks, rely heavily on manual experience for operation and maintenance, and have many blind spots in fault protection. This invention, through AI model optimization, multi-source data fusion, and linkage interlocking design, reduces the interlocking misjudgment rate to 0.8%, provides early warning of environmental coupling risks 15-30 minutes in advance, reduces operation and maintenance costs by 28%, compresses operation response latency to 80 milliseconds, and achieves a fault protection coverage rate of 99.2%, fully meeting the high-precision and high-reliability anti-misinterlocking requirements of large-scale onshore wind farm booster stations.

[0045] Example 2: Application of offshore wind farm booster station This embodiment is applied to a 200MW offshore wind farm booster station, adapted to a 330kV voltage level, and adopts an unattended design. It includes 3 main transformers, 10 circuit breakers, 20 disconnect switches, and 15 grounding switches. Facing harsh marine environments such as high salt spray, high humidity, strong typhoons, and surges, the system is reinforced with anti-corrosion and anti-interference design, and all technical solutions are fully implemented to achieve unattended intelligent anti-misoperation interlocking.

[0046] The data acquisition module integrates 40 voltage sensors, 30 current sensors, 24 temperature sensors, 25 position sensors, and 8 marine environmental monitoring units. All sensors are encapsulated to resist salt spray corrosion; the voltage and current sensors have an accuracy class of 0.2, and the position sensors have an IP68 protection rating. The environmental monitoring units additionally collect data on salt spray concentration, wave height, and tide level. Edge computing technology is used for Kalman filtering noise reduction and normalization preprocessing. The sampling frequency is 60Hz, and data is uploaded every 500 milliseconds. Raw data is cached locally for 2 hours to prevent communication interruptions.

[0047] The big data converged storage module adopts a distributed redundant storage architecture, with SSD high-speed cache, SAS hot data area, and SATA cold data area configured with capacities of 1TB, 10TB, and 50TB, respectively. Real-time data and critical operation records are stored with dual copies, and cold data is compressed and encrypted for storage. Multi-dimensional indexes support fast retrieval by device, time, and fault type, with data access latency ≤40 milliseconds, meeting the data traceability requirements in unattended scenarios.

[0048] The AI-powered intelligent analysis and decision-making module is equipped with an improved CNN-LSTM model, a PPO reinforcement learning model, and an extended rule-based reasoning engine. The rule base includes 150 specific operational guidelines for offshore substations. It integrates nearly two years of historical operational data, equipment corrosion status data, and marine environmental data to analyze equipment electrical performance, operational compliance, and salt spray corrosion risks. The decision response time is ≤90 milliseconds, and it automatically strengthens interlocking strategies for scenarios such as typhoon warnings.

[0049] The anti-misoperation interlocking execution module drives 20 sets of electromagnetic locks, 30 sets of mechanical locks, and 10 intelligent measurement and control units. The actuator is treated with anti-salt spray coating. The emergency interlocking response time is ≤8 milliseconds, the normal interlocking time is ≤40 milliseconds, the early warning interlocking synchronously triggers remote alarms, and the execution status is transmitted back to the shore-based monitoring center in real time.

[0050] The communication module adopts a dual-mode architecture of industrial Ethernet and 5G private network. The industrial Ethernet enables 10 Gigabit-level data interaction between modules within the station, while the 5G private network establishes a low-packet-loss link with the shore-based platform through the offshore communication base station. Core data is synchronized every 300 milliseconds, and the transmission latency of interlocking commands is ≤15 milliseconds, supporting seamless switching of communication links. The status monitoring module focuses on monitoring the corrosion resistance of sensors, the redundancy status of storage devices, and the anti-interference performance of communication. After identifying a fault, it marks the location, type, and scope of impact, and automatically triggers an emergency response.

[0051] Upon receiving a fault signal, the emergency response module isolates the faulty module within 200 milliseconds, and the backup module automatically switches over with a switching delay of ≤200 milliseconds. In the event of a communication interruption, a local autonomous operation mode is initiated, executing interlocking operations according to a preset strategy. Data is synchronized after communication is restored. In the event of a severe fault, a shore-based audible and visual alarm and SMS notification are triggered. After fault resolution, 15 self-test items are performed, including sensor calibration, model accuracy verification, and interlocking mechanism function testing. Normal operation resumes after all tests are passed.

[0052] The dynamic locking threshold optimization module is based on the formula. Calculate the threshold and set it. =0.45 =0.4 =0.15. When the circuit breaker health level is... =0.78, Comprehensive quantitative value of marine environment =0.65 Normalized value of high salt spray concentration and operation frequency When =0.5, =0.45×0.78+0.4×0.65+0.15×0.5=0.351+0.26+0.075=0.686, the lockout threshold is reduced by 22% compared to the base value. The operational risk assessment module uses the formula... Calculation, setting =0.55、 =0.25、 =0.2, disconnecting switch tripping operation =4. Equipment defects =2. Environmental Risks =2, =0.55×4+0.25×2+0.2×2=2.2+0.5+0.4=3.1, which is determined to be medium risk, and enhanced verification and flow limiting blocking are initiated.

[0053] The access control module employs a four-level authorization mechanism. Onshore administrators have the highest privileges, remote maintenance personnel can execute operational commands, and monitoring personnel can only view data. Emergency authorization is valid for 12 hours, and identity is verified through facial biometrics and hardware keys. All operation logs are encrypted and stored for two years. The data encryption module uses a combination of AES-256-GCM symmetric encryption and RSA-4096 asymmetric encryption to prevent data theft and tampering. The AI ​​model adaptive adjustment module detects that when the model confidence drops to 0.83, it extracts 800 recent marine special-condition data points to construct an incremental training set. After transfer learning retraining, the model accuracy improves by 12%.

[0054] The cross-equipment linkage interlocking module constructs an equipment association map, establishing strong associations between the main transformer and auxiliary equipment, as well as submarine cable connection equipment. During typhoon warnings, 12 associated devices are simultaneously interlocked. The environment-equipment coupling early warning module integrates six types of environmental data, including salt spray concentration. When salt spray concentration exceeds the standard, the equipment operation interlocking threshold is increased by 30%, prohibiting non-emergency operations. The operation trajectory error prevention verification module collects operation command characteristics through a remote control terminal; a secondary verification is triggered when the similarity is below 0.65. The full lifecycle health management module integrates five years of equipment operation data, calculates a circuit breaker health level of 0.75, predicts a remaining safe operating time of 4.8 years, and generates a recommendation for corrosion prevention maintenance every six months.

[0055] Table 2. Performance Comparison of Anti-misoperation Interlocking Systems for Offshore Wind Farm Substations Table 2 highlights the advantages of this invention's system in adapting to marine environments. Traditional systems are ill-suited to harsh marine environments, suffer from high interlocking misjudgment rates, lack unattended operation support, rely on manual fault recovery, and incur high maintenance costs. This invention, through corrosion-resistant design, AI-powered intelligent optimization, linked interlocking, and an autonomous emergency mechanism, reduces the interlocking misjudgment rate to 0.6%, fully adapting to harsh marine environments and unattended operation requirements. It achieves a 92% automatic fault recovery rate and reduces annual maintenance costs by 35%, providing a safe, reliable, intelligent, and efficient anti-misinterlocking solution for offshore wind farm booster stations.

[0056] Reference Figure 2 This diagram visually illustrates the core value of the dynamic interlocking threshold optimization module. Traditional systems use a fixed threshold of 0.8, which cannot adapt to the actual operating conditions of equipment under different environments such as low and high temperatures. At low temperatures, over-interlocking can easily affect normal equipment operation, while at high temperatures, insufficient thresholding can easily lead to safety risks. This invention dynamically calculates the threshold using the formula T=α×H+β×E+γ×F, precisely adjusting it according to temperature changes. In low-temperature environments, the threshold is appropriately lowered to enhance protection, while in normal and high-temperature environments, the threshold is reasonably raised to adapt to equipment operating conditions. This ensures that the interlocking judgment always closely matches the actual state of the equipment, significantly improving the interlocking accuracy of onshore substations under different temperature conditions.

[0057] Reference Figure 3 This diagram clearly illustrates the technological advantages of the AI-powered intelligent analysis and decision-making module. Traditional systems rely solely on single electrical parameters to diagnose faults, exhibiting weak capabilities in identifying complex faults such as environmental coupling risks and sensor failures, with recognition rates generally below 70%. This vulnerability can easily lead to safety accidents due to missed fault detection. This invention, through multi-source data fusion and deep learning models, integrates multi-dimensional data from electrical, environmental, and equipment status dimensions to uncover correlation patterns in fault characteristics. The recognition rate for various fault types has been improved to over 90%, with a nearly 50% increase in the recognition rate for hidden faults such as environmental coupling risks. This significantly reduces the probability of missed fault detection and ensures the safe operation of the booster station.

[0058] Reference Figure 4 This diagram highlights the synergistic value of full lifecycle health management and adaptive adjustment of AI models. Traditional systems, lacking mechanisms for adapting to equipment health status, experience a rapid increase in malfunction rate with increasing equipment age, reaching 16.8% after 5 years, severely impacting normal equipment operation. This invention dynamically optimizes the interlocking strategy based on full lifecycle equipment data, combined with adaptive adjustment of the AI ​​model to maintain high inference accuracy. Even after 5 years of operation, the malfunction rate remains at a low level of only 1.5%. This feature effectively adapts to performance changes caused by equipment aging, extends the effective operating cycle of equipment, and reduces the frequency of maintenance interventions.

[0059] Reference Figure 5This diagram visually demonstrates the optimization effect of this invention on operation and maintenance costs. Traditional systems rely heavily on manual inspections to troubleshoot faults, with manual inspection and repair costs accounting for up to 75% of total costs. Equipment replacement costs are also high, resulting in low operation and maintenance efficiency and high costs. This invention, through intelligent monitoring, precise early warning, and automatic fault recovery, significantly reduces the costs of manual inspections and repairs, which now account for only 23% of total costs. Although the proportion of other operation and maintenance costs increases, the overall total operation and maintenance cost is reduced by more than 30%. This optimized cost structure reflects the system's shift from "passive maintenance" to "proactive early warning and intelligent operation and maintenance," effectively improving the economic efficiency of wind farm operation and maintenance.

[0060] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An AI-driven big data fusion wind farm booster station anti-misoperation locking system, characterized in that, Comprise the following modules: Data acquisition module: integrate voltage, current, temperature, position sensor and environmental monitoring unit, real-time acquisition of key equipment of booster station electrical parameters, operating state parameters and environmental parameters, using edge computing technology preprocessing according to the preset frequency upload; Big data fusion storage module: using distributed storage architecture combined with hierarchical storage strategy, real-time data, historical operation data, equipment account data, operation record data are classified and stored in corresponding data area, dynamic adjustment of storage resource allocation; AI intelligent analysis and decision module: equipped with deep learning, reinforcement learning model and rule reasoning engine, receive preprocessed data and fuse historical data, analyze device running state, operation compliance and potential risk, generate allow anti-misoperation locking strategy; Anti-misoperation locking execution module: receive locking instruction, drive related execution mechanism according to emergency, conventional, early warning locking three level mode, synchronous feedback execution state; Communication interaction module: using industrial Ethernet and wireless communication dual mode architecture, bidirectional transmission of data with external platform at the same time; State monitoring module: real-time monitoring of each module running state, identify fault and mark fault location and type; emergency treatment module receives fault signal and emergency risk warning, starts emergency response mechanism, executes self-checking program after fault elimination, restores system normal operation after verifying function recovery.

2. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, Also includes a dynamic lockout threshold optimization module, based on the device real-time running state and environmental changes dynamic adjustment lockout decision threshold, the expression is calculated as Wherein is a dynamic lockout threshold, is a device health degree weight coefficient, is a device real-time health degree quantitative value, is an environmental influence weight coefficient, is an environmental factor comprehensive quantification, is an operation frequency weight coefficient, is a device recent operation frequency normalized value, + + =1. 3.The wind farm booster station anti-misoperation locking system driven by AI and big data fusion of claim 1, characterized in that, Also includes an operation risk assessment module, combined with device state, operation type, environmental conditions, historical failure data quantitative assessment of operation risk level, calculation expression is , wherein is the operation risk level value, is the operation complexity weight coefficient, is the operation type complexity quantitative value, is the device defect weight coefficient, is the device current defect cumulative quantitative value, is the environmental risk weight coefficient, is the environmental risk level quantitative value, + + = 1, based on the risk level value is divided into low risk, medium risk, high risk three levels, corresponding to different lock response strategy.

4. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, Also includes permission management module, using role hierarchical authorization mechanism, the system operation permission is divided into administrator permission, operation and maintenance personnel permission, monitoring personnel permission, temporary authorization permission, different permissions are allocated for different roles, the identity of the operator is confirmed by the verification method combining identity authentication, password verification and biological recognition, and the operation behavior log related to all permissions is recorded.

5. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, Also includes data encryption module, sensitive data in the process of internal data transmission and external data interaction of the system is encrypted, using the encryption method combining symmetric encryption algorithm and asymmetric encryption algorithm, encrypting the core data of the device, generating encrypted data packet and adding data check code.

6. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, Also includes AI model adaptive adjustment module, real-time monitoring of the inference accuracy, decision response time and fault identification accuracy of deep learning model and reinforcement learning model in AI intelligent analysis and decision module, calculating the model confidence quantitative value, when the confidence is lower than the preset threshold, automatically extracting recent high value data including abnormal working condition data, misjudgment correction data and new scene data to construct incremental training set, using transfer learning algorithm to retrain the model, dynamically adjusting the number of network layers, number of neurons and weight parameters of the model, while keeping the backup of the historical optimal model parameters.

7. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, Also includes cross device linkage locking module, builds the correlation graph of key equipment of booster station, clarifies the electrical connection relationship and operation logic dependency relationship between devices, establishes linkage locking rule library, when a single device triggers locking, synchronously analyzes the running state and operation intention of associated devices, determines the linkage locking range according to the principle of priority of main device, priority of safety critical and priority of operation time sequence, drives associated devices to execute cooperative locking operation, while supporting custom configuration of linkage logic. 8.The wind farm booster station anti-misoperation locking system driven by AI and big data fusion of claim 1, characterized in that, It also includes an environment and equipment coupling early warning module, which establishes a coupling correlation model between environmental parameters and equipment failures, integrates five types of environmental data (wind speed, humidity, temperature, icing thickness, lightning activity) with historical correlation data of equipment insulation performance, mechanical characteristics, and electrical parameters, and uses a gradient boosting algorithm to mine coupling patterns and calculate the environment and equipment coupling risk coefficient in real time. 9.The wind farm booster station anti-misoperation locking system driven by AI and big data fusion of claim 1, wherein, It also includes an operation trajectory error prevention and verification module, which integrates a visual acquisition unit and a motion sensing unit to collect the operator's operation trajectory, action sequence, operation force characteristics and operation command input rhythm in real time. It extracts operation behavior feature vectors through AI algorithms and compares them with the operator's historical compliant operation feature library and the system's preset standard operation feature templates to calculate the behavior similarity. When the similarity is lower than the set threshold, it is judged as an abnormal operation trajectory and triggers a secondary verification mechanism, including operation command review, identity secondary authentication, and operation intent confirmation. At the same time, it automatically updates the operator behavior feature library.

10. The AI-driven big data fusion wind farm booster station anti-misoperation locking system according to claim 1, characterized in that, It also includes a full lifecycle health management module, which integrates equipment operation data, fault records, maintenance records, and lockout event records to establish a full lifecycle health assessment model for equipment. It uses a multi-dimensional weighted scoring method to calculate the real-time health of the equipment, combines equipment parameters to predict the remaining safe operating time of the equipment, dynamically optimizes the lockout strategy based on the health decay trend, and generates equipment maintenance suggestions including maintenance timing, maintenance items, and spare parts replacement warnings, thus deeply integrating the lockout system with equipment operation and maintenance.

Images