Wind turbine monitoring big data analysis system based on multi-mode fusion analysis of voiceprint recognition

Abstract

The application discloses a wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis and belongs to the technical field of power grid monitoring big data analysis; and is used for solving the technical problems of poor data processing and analysis effect and poor diversified early warning effect of the prior art; by integrating multi-mode data such as voiceprints, vibrations and temperatures, the traditional single-sensor monitoring data island phenomenon can be avoided, the data dimension can be improved from 3D to 23D, complete input is provided for fault diagnosis; the first identifier and the second identifier realize accurate association of data and equipment, and ensure the accuracy of subsequent preprocessing and analysis; environmental noise is removed through preprocessing, the signal-to-noise ratio of the voiceprint signal is effectively improved, and the feature extraction accuracy is improved, reliable data can be provided for subsequent fault analysis; based on label classification storage, the data security is improved; through the distributed storage characteristics of the block chain and the fault influence quantitative analysis, slight and serious fault grading early warning can be realized.

Description

Technical Field

[0001] This invention relates to the field of power grid monitoring big data analysis technology, specifically to a wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis. Background Technology

[0002] Big data analysis for wind turbine monitoring based on voiceprint recognition multi-modal fusion analysis is a cutting-edge intelligent operation and maintenance technology. It aims to achieve precise monitoring of the turbine's health status and early warning and diagnosis of faults by collecting and analyzing the sounds emitted during wind turbine operation (voiceprints) and combining them with other multi-source data. Existing big data analysis solutions for wind turbine monitoring based on voiceprint recognition multi-modal fusion analysis typically include data acquisition, preprocessing, feature extraction, multi-modal fusion, fault diagnosis and early warning, and visual management. Acoustic sensors are deployed at key parts of the wind turbine to collect sound signals during operation, while simultaneously acquiring multi-source data such as vibration, temperature, and SCADA operating parameters. After preprocessing such as denoising, framing, and spectrum conversion, features such as MFCC, spectral energy, and zero-crossing rate are extracted and fused with other modal data at the feature layer or decision layer. Machine learning or deep learning models are used to identify and classify typical faults such as bearing wear, gear breakage, and blade damage. Finally, the analysis results are visualized, health assessed, anomaly warnings are provided, and operation and maintenance suggestions are output through a big data platform, forming an intelligent monitoring system.

[0003] However, this type of solution still has many problems and shortcomings in practical applications: First, the wind power site environment is complex, with strong background noise such as wind noise, rain and snow, and electromagnetic interference, resulting in a low signal-to-noise ratio of the acoustic signature signal and making it difficult to extract effective fault features; Second, sound is prone to reflection, attenuation, and distortion when propagating in the tower structure, and the signals received by different sensor locations vary greatly, affecting the stability of the model; In addition, deep learning models are mostly black boxes, lack interpretability, and are difficult to gain the trust of operation and maintenance personnel; Acoustic signature analysis requires a large amount of computation, and the computing power of edge devices is limited, making it difficult to achieve real-time online processing; Currently, there is a lack of unified acoustic signature acquisition standards, feature specifications, and evaluation systems, resulting in poor data processing and analysis effects and poor diversified early warning effects. Summary of the Invention

[0004] The purpose of this invention is to provide a big data analysis system for wind turbine monitoring based on voiceprint recognition and multi-mode fusion analysis, in order to solve the technical problems of poor data processing and analysis effects and poor diversified early warning effects in existing technical solutions.

[0005] The objective of this invention can be achieved through the following technical solutions: A big data analysis system for wind turbine monitoring based on voiceprint recognition and multi-modal fusion analysis includes: Data acquisition and statistics module: Real-time acquisition and statistics of acoustic fingerprint signals and multi-dimensional operation data of wind turbine units, generating standardized operation logs, and uploading them to the data preprocessing and classification module after being classified by wind turbine model; Data preprocessing and classification module: preprocesses the received multimodal data and operation logs and determines the fault type. Based on the determination results, it generates tags and uploads them to the corresponding blockchain to achieve hierarchical data storage. Fault Integration and Collaboration Module: Based on publicly available fault data updated in real time on the public blockchain, the module integrates fault data and constructs a coordinate system, generates a fault fluctuation table, and dynamically pushes it to the wind farm operation and maintenance platform and equipment manufacturers to achieve cross-platform fault collaborative response. Fault handling closed-loop traceability module: verifies the integrity of fields in forms filled in by maintenance personnel, generates processing record IDs, performs multi-dimensional verification of processing results, stores processing records on the blockchain in a hierarchical manner, implements periodic supervision and data processing of existing multi-dimensional verification schemes, analyzes the implementation effect of multi-dimensional verification schemes, and implements targeted dynamic optimization and early warning prompts.

[0006] Preferably, when classifying and labeling data, the wind turbine model is used as the first identifier, and the unique code of the wind farm is used as the second identifier.

[0007] Preferably, when preprocessing the voiceprint signal, environmental noise is removed by low-pass filtering, the short-time energy and spectral peak features of the signal are extracted, and the data are aligned with vibration and temperature data by timestamp to form fused data of fan ID-timestamp-voiceprint feature-vibration value-temperature value; Update the corresponding wind turbine's operation log according to the second identifier to ensure that the data is associated with each device. When determining the fault type and generating tags, serious faults are preset as public standard types and stored in the system's local database.

[0008] Preferably, the fault type in the operation log is obtained and matched with a preset standard type; for example, a broken gear in the gearbox is matched with a severe fault. If a match is found, the data is marked as public behavior and a public label is generated; If the match fails, it is marked as a privacy behavior and a privacy tag is generated; The judgment result is determined by processing the obtained public or private tags.

[0009] Preferably, the corresponding blockchain storage allocation includes uploading public tag data and uploading privacy tag data; When uploading public tag data, the judgment results are traversed, and the fault data corresponding to the public tag is uploaded to the public chain through the first identifier and the second identifier. The fault data includes the faulty component, the time of occurrence, the duration, and the vibration amplitude. When uploading privacy tag data, the normal operation data and minor fault records corresponding to the privacy tag are uploaded to the private chain through the first identifier and the second identifier.

[0010] Preferably, when a public link receives new public tag data, an integration instruction is automatically generated to trigger the data integration process; when performing fault frequency interval matching, a fault frequency interval table is preset, the number of faults for each wind turbine model is matched with the fault frequency interval table, the interval corresponding to the fault count is obtained, and the total number of occurrences in that interval is incremented by 1 and updated.

[0011] Preferably, for the first identifier, the total number of occurrences of each faulty component is counted, and the standard fault threshold corresponding to the faulty component is obtained. The total number of occurrences of each faulty component is then compared with the corresponding standard fault threshold. If the total number is less than the standard fault threshold, then associate it with a normal component tag; Conversely, associate the abnormal component label; Sort by total number of components in descending order to obtain data on the local impact of the fault.

[0012] Preferably, when calculating the fault impact, the median of the total number of occurrences of each faulty component in the corresponding interval of the fault impact coordinate system is obtained in sequence. The median obtained is multiplied by the corresponding total number of occurrences to obtain the local impact. The local impact of each faulty component is summed and the summed value is set as the fault impact. The impact of the fault is compared and analyzed with the preset fault impact threshold. If the impact of the fault is less than the preset fault impact threshold, then associate it with the overall normal impact tag; Conversely, the overall abnormality will affect the label; Set the first identifier, local fault impact data, overall fault impact degree, and overall early warning label as fault fluctuation items and combine them to generate a fault fluctuation table; The fault fluctuation table is synchronously pushed to the equipment manufacturer's system, the wind farm operation and maintenance platform, and the regional dispatch center through the first identifier, and an alarm is triggered.

[0013] Preferably, within a preset regulatory period, the total number of data verifications that passed, the total number of data verifications that failed, the total number of data verifications that passed manual review, and the total number of data verifications that failed manual review are statistically analyzed. The values ​​of each statistical data item are extracted, and the first verification value corresponding to the data verification is calculated using a formula. and the second verification value reviewed by a human. Wherein, N1, N2, N3, and N4 represent the total number of data verifications that passed, the total number of data verifications that failed, the total number of data verifications that passed manual review, and the total number of data verifications that failed manual review, respectively. This is the first verification standard value. This is the second verification standard value.

[0014] Preferably, the calculated first verification value and the second verification value are analyzed. If both the first verification value and the second verification value are greater than 0, an overall valid verification label is generated, and the existing verification scheme is maintained. If either the first verification value or the second verification value is not greater than 0, a locally valid verification label is generated, and a local upgrade and optimization warning is issued for the verification scheme corresponding to the verification value not greater than 0. If both the first and second verification values ​​are not greater than 0, an overall verification anomaly label will be generated, and both the existing automatic verification scheme and the manual review verification scheme will be upgraded, optimized, and given an early warning.

[0015] Compared to existing solutions, the beneficial effects achieved by this invention are: This invention integrates multimodal data such as acoustic signature, vibration, and temperature, avoiding the data silos of traditional single-sensor monitoring. It can increase the data dimension from 3D to 23D, providing complete input for fault diagnosis. By using a first identifier and a second identifier, it achieves precise association between data and equipment, ensuring the accuracy of subsequent preprocessing and analysis. In addition, it can support data acquisition from different models of wind turbines, adapt to complex environments with different wind speeds, and improve data acquisition coverage.

[0016] This invention removes environmental noise through preprocessing, which can effectively improve the signal-to-noise ratio of the acoustic signature and the accuracy of feature extraction, providing reliable data for subsequent fault analysis. Based on tag-based classification and storage, it can ensure the public sharing of key fault data while protecting the technical privacy of manufacturers, thus improving data security. Through the distributed storage characteristics of blockchain, it can effectively shorten the data query response time and support the rapid filtering of historical data by wind turbine model and fault type.

[0017] This invention enables graded early warning of minor and serious faults through quantitative analysis of fault impact, effectively reducing the false alarm rate. Through cross-platform collaboration, it can establish data links between ventilation and power plants, manufacturers, and dispatch centers, effectively shortening emergency fault response time.

[0018] This invention achieves full-process visualization of fault handling through a four-step process of recording, verification, evidence storage, and traceability. It can solve the problem of no follow-up after warning in traditional systems and effectively improve the closed-loop rate of fault handling. By using blockchain evidence storage and timestamp solidification, it can ensure that the processing records are tamper-proof. Combined with dual verification by data and human verification, it can effectively improve the credibility of fault handling results and provide manufacturers with real data basis for design improvement. By automatically summarizing effective handling measures for high-frequency faults, a reusable operation and maintenance knowledge base can be formed, which can effectively improve the fault handling efficiency of new operation and maintenance personnel. Attached Figure Description

[0019] The present invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a flowchart illustrating the operation of the wind turbine monitoring big data analysis system based on voiceprint recognition and multi-modal fusion analysis according to the present invention.

[0021] Figure 2 This is a block diagram of the wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis of the present invention.

[0022] Figure 3 A schematic diagram of the structure of a computer device for implementing embodiments of the present invention. Detailed Implementation

[0023] 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.

[0024] Example 1: As Figures 1 to 2 As shown, the present invention is a big data analysis system for wind turbine monitoring based on voiceprint recognition multi-mode fusion analysis, including a data acquisition and statistics module, a data preprocessing and classification module, a fault integration and coordination module, as well as private chain and public chain storage units; The multimodal data and operation logs output by the data acquisition module serve as inputs to the data preprocessing and classification module. The preprocessed classified data is stored in the corresponding blockchain. Finally, the fault integration and collaboration module calls the on-chain data to generate a fault fluctuation table and pushes it to the relevant parties.

[0025] Data Acquisition and Statistics Module: This module collects and statistically analyzes acoustic signature signals and multi-dimensional operational data from wind turbine generators in real time, generates standardized operation logs, and uploads them to the data preprocessing and classification module after categorizing them by wind turbine model. Specifically: When conducting multi-source data acquisition, the following sensors and data acquisition units are deployed for different types of wind turbines, such as 2MW onshore wind turbines and 3MW offshore wind turbines: Acoustic sensors: installed in gearboxes, generators, and blade roots respectively, to collect acoustic signals during operation, such as gear meshing sounds, blade airflow noise, and bearing noise. Auxiliary sensors: synchronously collect vibration data, temperature data, current data, and environmental parameters; All sensors are connected to the local data acquisition terminal via industrial Ethernet. The data is stored in the format of fan ID-timestamp-data type-value; for example: WTG-001-20251027083000-acoustic print-gearbox signal-120dB. When generating and uploading runtime logs, the process includes data classification and labeling, log field definition, and data uploading. When classifying and labeling data, the wind turbine model is used as the first identifier, and the unique code of the wind farm is used as the second identifier. When defining log fields, the operation log includes fault type, fault component, occurrence time, duration, and processing status. Fault type is divided into critical fault, minor fault, and normal operation; fault components include gearbox, generator, blades, bearings, etc. Serious faults, such as broken gear teeth or cracked blades; minor faults, such as abnormal bearing noise or temperature fluctuations. When uploading data, multimodal data and operation logs with the same first identifier are packaged and uploaded to the data preprocessing and classification module in real-time Beijing time order, accurate to the second. For example, when the vibration amplitude of the gearbox of the first identifier 2MW-Model A exceeds the threshold, the log record is: "Fault type: serious fault; faulty component: gearbox high-speed shaft; occurrence time: 2025-10-27 08:30:00; duration: 30s; processing status: unprocessed".

[0026] In this embodiment of the invention, by integrating multimodal data such as soundprint, vibration, and temperature, the data silo phenomenon of traditional single-sensor monitoring can be avoided, and the data dimension can be increased from 3D to 23D, providing complete input for fault diagnosis. The first and second identifiers enable precise association between data and equipment, ensuring the accuracy of subsequent preprocessing and analysis. In addition, it can support data acquisition from different models of wind turbines, adapt to complex environments with different wind speeds, and improve data acquisition coverage.

[0027] Data preprocessing and classification module: This module preprocesses and classifies received multimodal data and operation logs, determines fault types, generates tags based on the determination results, and uploads them to the corresponding blockchain to achieve hierarchical data storage. Specifically: When preprocessing the voiceprint signal, environmental noise is removed by low-pass filtering, and the short-time energy and spectral peak features of the signal are extracted. These features are then aligned with vibration and temperature data by timestamp to form fused data of fan ID-timestamp-voiceprint feature-vibration value-temperature value. The preprocessing of the voiceprint signal is a conventional technique, and the specific implementation steps will not be elaborated here. Update the corresponding fan's operation log according to the second identifier to ensure that the data is associated with each device. For example, update the historical log of the fan with the newly collected gearbox temperature data through the second identifier of WTG-001. When determining the fault type and generating tags, severe faults are preset as public standard types and stored in the system's local database; Retrieve the fault type from the operation log and match it with a preset standard type; for example, match a broken gear in the gearbox with a severe fault. If a match is found, the data is marked as public behavior and a public label is generated; the label format is public-faulty component-occurrence time. If a match fails, for example, if the fault type is a minor abnormal noise or normal operation, it is marked as a privacy behavior and a privacy tag is generated; the tag format is privacy-faulty component-occurrence time; The judgment result is determined by processing the obtained public or private tags; When allocating corresponding blockchain storage, it includes uploading public label data and uploading privacy label data; When uploading public tag data, the judgment results are traversed, and the fault data corresponding to the public tag is uploaded to the public chain through the first identifier and the second identifier. The fault data includes the faulty component, the time of occurrence, the duration, and the vibration amplitude. When uploading privacy tag data, the normal operation data and minor fault records corresponding to the privacy tag are uploaded to the private chain through the first identifier and the second identifier; In this embodiment of the invention, by removing environmental noise through preprocessing, the signal-to-noise ratio of the acoustic signature signal and the accuracy of feature extraction can be effectively improved, providing reliable data for subsequent fault analysis; based on tag-based classification storage, it can ensure the public sharing of key fault data and protect the technical privacy of manufacturers, thus improving data security; through the distributed storage characteristics of blockchain, the data query response time can be effectively shortened, and historical data can be quickly filtered by wind turbine model-fault type.

[0028] Fault Integration and Collaboration Module: Based on publicly available fault data updated in real-time on the public blockchain, this module integrates fault data and constructs a coordinate system. It then generates a fault fluctuation table, which is dynamically pushed to the wind farm operation and maintenance platform and equipment manufacturers, enabling cross-platform collaborative fault response. Specifically: When integrating fault data and constructing a coordinate system, an integration instruction is automatically generated and the data integration process is triggered when a public link receives new public tag data. When performing fault frequency range matching, a fault frequency range table is preset, which specifically includes the ranges: [0-5), [5-10), [10-15), and ≥15. Match the number of failures for each wind turbine model with the failure frequency interval table, obtain the interval corresponding to the number of failures, and increment the total number of occurrences in that interval by 1 and update it; When constructing the fault impact coordinate system, the fault frequency range is used as the horizontal axis and sorted from smallest to largest: [0-5), [5-10), [10-15), ≥15; Construct a two-dimensional coordinate system with the total number of occurrences in the interval as the vertical axis and an interval of 5. For the first identifier, the total number of occurrences of each faulty component is counted, and the standard fault threshold corresponding to the faulty component is obtained. The total number of occurrences of each faulty component is then compared with the corresponding standard fault threshold. The standard fault threshold can be determined based on existing industry standard data, or it can be customized according to the actual application requirements of the actual application scenario. The specific value is not limited. If the total number is less than the standard fault threshold, then associate it with a normal component tag; Conversely, associate the abnormal component label; Sort by total number of components in descending order to obtain data on the local impact of the fault; When calculating the impact of a fault, the median of the total number of occurrences of each faulty component in the corresponding interval of the fault impact coordinate system is obtained in turn. The median obtained is multiplied by the corresponding total number of occurrences to obtain the local impact. The local impact of each faulty component is summed and the sum is set as the fault impact. The fault impact is compared and analyzed with the preset fault impact threshold. The fault impact threshold can be determined based on existing industry standard data or customized according to the actual application requirements of the actual application scenario. The specific value is not limited. If the impact of the fault is less than the preset fault impact threshold, then associate it with the overall normal impact tag; Conversely, the overall abnormality will affect the label; Set the first identifier, local fault impact data, overall fault impact degree, and overall early warning label as fault fluctuation items and combine them to generate a fault fluctuation table; The fault fluctuation table is synchronously pushed to the equipment manufacturer's system, the wind farm operation and maintenance platform, and the regional dispatch center through the first identifier, and an alarm is triggered. When a third-party platform receives the first identifier input by the user through the platform, it generates a service instruction; the third-party platform is, for example, a wind power operation and maintenance APP; the user is, for example, an operation and maintenance personnel. According to the service instructions, a request is sent to the public blockchain to obtain the fault fluctuation table corresponding to the model, and displayed through the APP interface; for example: the overall status of the 2MW-model A wind turbine is currently normal, but the number of gearbox failures has reached the threshold, and it is recommended to focus on checking it during the next maintenance, so as to realize cross-platform sharing and collaborative response of monitoring data.

[0029] In this embodiment of the invention, through quantitative analysis of the impact of faults, it is possible to achieve graded early warning of minor and serious faults, which can effectively reduce the false alarm rate; through cross-platform collaboration, it is possible to connect the data links of the power plant, the manufacturer, and the dispatch center, which can effectively shorten the emergency fault response time.

[0030] Example 2: It also includes a fault handling closed-loop traceability module: verifying the integrity of fields in forms filled out by maintenance personnel, generating processing record IDs, performing multi-dimensional verification of processing results, and storing processing records on a blockchain with hierarchical notarization. It implements periodic monitoring and data processing of existing multi-dimensional verification schemes, analyzes the implementation effectiveness of multi-dimensional verification schemes, and implements targeted dynamic optimization and early warning prompts. Specific steps include: The system retrieves forms filled out by maintenance personnel through a third-party platform, and verifies the integrity of the fields in real time. If the processing result is selected as fault elimination, then the verification method and verification time must be filled in. If the maintenance type is component replacement, then the QR code scan record of the replaced component must be uploaded to ensure the component model is accurate. After submission, a processing record ID is generated, in the format of RID-fan ID-fault number, which serves as a unique identifier for subsequent tracing. The system has a built-in fault handling record form, which includes preset fields for basic information, processing procedure, and result verification. The basic information area includes the wind turbine ID, faulty component, fault number, and the ID of the person handling the issue. The processing section includes the maintenance type, replacement part model, tools used, and processing time. The results verification area includes the processing results, verification methods, and verification time; among which, the processing results include fault elimination, fault mitigation, and unresolved; the verification methods include sensor data verification and / or manual visual inspection verification. When performing multi-dimensional verification of the processing results, it includes automatic data verification and manual review verification; During automatic data verification, the system calls upon real-time monitoring data from the data acquisition and statistics module to automatically verify the processing results. Within K hours after the processing of the object is completed, corresponding monitoring data is collected and matched with corresponding standard data to determine whether the verification requirements are met. For example: If the fault is abnormal noise from the gearbox, collect the soundprint signal within 1 hour after the problem is resolved and compare it with the preset normal operation soundprint template. For example, the normal meshing sound spectrum range of the gearbox is 500Hz~2kHz. If the spectrum matching degree is ≥90%, the data verification is passed; otherwise, the data verification fails. If the fault is excessive bearing vibration, continuously monitor the vibration amplitude for 24 hours after handling. If the value is ≤6mm / s for 3 consecutive times, the data verification is passed; otherwise, the data verification fails. During manual review and verification, records that fail data verification or whose processing results are unresolved are automatically pushed to the manufacturer's technical personnel for review. Auditors can view the original fault data and processing records through a private chain. The original fault data includes vibration waveform diagrams and acoustic spectrum diagrams. They can fill in audit comments, such as inappropriate handling measures or the need to replace the bearing clearance. Once approved, a manual verification label will be generated, in the format: Manual - Passed - Reviewer ID; otherwise, it will be marked as Manual - Failed - Rectification Suggestions. When processing and recording blockchain hierarchical evidence storage, evidence storage labels are generated based on the processing results and verification status, including public evidence storage labels and private evidence storage labels; the classification logic of the above-mentioned public labels can be used as a basis. Publicly available evidence label: The processing result is that the fault has been eliminated and the data has been verified; Privacy Evidence Label: The processing result is unresolved or failed manual review; When allocating blockchain storage, the public blockchain stores records with publicly available evidence tags. The uploaded fields include the first identifier, faulty component, processing time, processing result, and data verification conclusion, which can be queried by wind farms and dispatch centers. Private blockchain storage carries records with privacy-preserving tags, and additionally uploads manual review comments, rectification suggestions, and component procurement costs, which are only accessible to manufacturers; All records are automatically appended with a timestamp from the National Time Service Center when uploaded, accurate to the second, to ensure that the storage time cannot be tampered with; Furthermore, within the preset regulatory period (unit: days, specific values ​​are not limited), the total number of data verifications that passed, failed, passed manual review, and failed manual review will be statistically analyzed. The values ​​of each statistical data item will be extracted, and the first verification value corresponding to the data verification will be calculated using a formula. and the second verification value reviewed by a human. Wherein, N1, N2, N3, and N4 represent the total number of data verifications that passed, the total number of data verifications that failed, the total number of data verifications that passed manual review, and the total number of data verifications that failed manual review, respectively. This is the first verification standard value. The first and second verification standard values ​​can be determined based on existing industry standard data or customized according to the application requirements of actual application scenarios. The specific values ​​are not limited. Data analysis is performed on the first and second verification values ​​obtained from the calculation. If both the first and second verification values ​​are greater than 0, a valid overall verification label is generated, and the existing verification scheme is maintained. If either the first verification value or the second verification value is not greater than 0, a locally valid verification label is generated, and a local upgrade and optimization warning is issued for the verification scheme corresponding to the verification value not greater than 0. If both the first and second verification values ​​are not greater than 0, an overall verification anomaly label will be generated, and both the existing automatic verification scheme and the manual review verification scheme will be upgraded, optimized, and given an early warning.

[0031] It should be noted that, unlike existing technical solutions that cannot proactively monitor and optimize verification schemes, this invention, through digital processing and analysis of the implementation effects of different verification schemes, and by providing dynamic optimization and early warning prompts for existing multidimensional verification schemes based on the analysis results, can effectively improve the accuracy and reliability of multidimensional verification scheme implementation.

[0032] In this embodiment of the invention, a four-step process of recording, verifying, storing evidence, and tracing is used to achieve full-process visualization of fault handling, which can solve the problem of no follow-up after the traditional system issues an alert and can effectively improve the closed-loop rate of fault handling. By using blockchain evidence storage and timestamp solidification, the processing records can be ensured to be tamper-proof. Combined with dual verification by data and manual verification, the credibility of fault handling results can be effectively improved, providing manufacturers with real data basis for design improvement. By automatically summarizing effective handling measures for high-frequency faults, a reusable operation and maintenance knowledge base can be formed, which can effectively improve the fault handling efficiency of new operation and maintenance personnel.

[0033] Example 3: As Figure 3 The diagram shown is a structural schematic of a computer device for implementing a wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis, as provided in an embodiment of the present invention.

[0034] Computer devices may include processors, memory, and buses, and may also include computer programs stored in memory and capable of running on the processor, such as a big data analysis program for wind turbine monitoring based on voiceprint recognition multi-mode fusion analysis.

[0035] The memory includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory can be an internal storage unit of a computer device, such as a portable hard drive. In other embodiments, the memory can be an external storage device of a computer device, such as a plug-in portable hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory can include both internal and external storage units of the computer device. The memory can be used not only to store application software and various types of data installed on the computer device, such as the code of a wind turbine monitoring big data analysis program based on voiceprint recognition multi-modal fusion analysis, but also to temporarily store data that has been output or will be output.

[0036] In some embodiments, the processor may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits packaged with the same or different functions. This includes combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processing units, and various control chips. The processor is the control unit of the computer device, connecting various components of the entire computer device through various interfaces and lines. It executes programs or modules stored in memory (e.g., a big data analysis program for wind turbine monitoring based on voiceprint recognition multi-mode fusion analysis) and calls data stored in memory to perform various functions of the computer device and process data.

[0037] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. The bus is configured to enable communication between the memory and at least one processor, etc.

[0038] Figure 3 Only computer equipment with components is shown; those skilled in the art will understand that... Figure 3 The structure shown does not constitute a limitation on the computer device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0039] For example, although not shown, the computer device may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to at least one processor via a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power sources, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The computer device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be elaborated further here.

[0040] Furthermore, the computer device may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the computer device and other computer devices.

[0041] Optionally, the computer device may further include a user interface, which may be a display, an input unit (such as a keyboard), and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the computer device and to display a visual user interface.

[0042] It should be understood that the above embodiments are for illustrative purposes only and are not limited to this structure in the scope of patent applications.

[0043] The big data analysis program for wind turbine monitoring based on voiceprint recognition multi-mode fusion analysis, stored in the memory of a computer device, is a combination of multiple instructions.

[0044] Specifically, the processor's implementation method for the above instructions can be found in [reference needed]. Figures 1 to 2 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0045] Furthermore, if the modules / units integrated into a computer device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, a computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0046] The present invention also provides a computer-readable storage medium storing a computer program that is executed by a processor of a computer device.

[0047] In the several embodiments provided by this invention, it should be understood that the disclosed methods can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is merely a logical functional division, and there may be other division methods in actual implementation.

[0048] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0049] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0050] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A wind turbine monitoring big data analysis system based on voiceprint recognition and multi-modal fusion analysis, characterized in that, include: Data acquisition and statistics module: Real-time acquisition and statistics of acoustic fingerprint signals and multi-dimensional operation data of wind turbine units, generating standardized operation logs, and uploading them to the data preprocessing and classification module after being classified by wind turbine model; Data preprocessing and classification module: preprocesses the received multimodal data and operation logs and determines the fault type. Based on the determination results, it generates tags and uploads them to the corresponding blockchain to achieve hierarchical data storage. Fault Integration and Collaboration Module: Based on publicly available fault data updated in real time on the public blockchain, the module integrates fault data and constructs a coordinate system, generates a fault fluctuation table, and dynamically pushes it to the wind farm operation and maintenance platform and equipment manufacturers to achieve cross-platform fault collaborative response. Fault handling closed-loop traceability module: verifies the integrity of fields in forms filled in by maintenance personnel, generates processing record IDs, performs multi-dimensional verification of processing results, stores processing records on the blockchain in a hierarchical manner, implements periodic supervision and data processing of existing multi-dimensional verification schemes, analyzes the implementation effect of multi-dimensional verification schemes, and implements targeted dynamic optimization and early warning prompts.

2. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis according to claim 1, characterized in that, When classifying and labeling data, the wind turbine model is used as the first identifier, and the unique code of the wind farm is used as the second identifier.

3. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis according to claim 2, characterized in that, When preprocessing the acoustic signature signal, environmental noise is removed by low-pass filtering, and the short-time energy and spectral peak features of the signal are extracted. These features are then aligned with vibration and temperature data by timestamp to form fused data of fan ID-timestamp-acoustic signature features-vibration value-temperature value. Update the corresponding wind turbine's operation log according to the second identifier to ensure that the data is associated with each device. When determining the fault type and generating tags, serious faults are preset as public standard types and stored in the system's local database.

4. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-modal fusion analysis according to claim 3, characterized in that, Retrieve the fault type from the operation log and match it with a preset standard type; for example, match a broken gear in the gearbox with a severe fault. If a match is found, the data is marked as public behavior and a public label is generated; If the match fails, it is marked as a privacy behavior and a privacy tag is generated; The judgment result is determined by processing the obtained public or private tags.

5. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-modal fusion analysis according to claim 4, characterized in that, When allocating corresponding blockchain storage, it includes uploading public label data and uploading privacy label data; When uploading public tag data, the judgment results are traversed, and the fault data corresponding to the public tag is uploaded to the public chain through the first identifier and the second identifier. The fault data includes the faulty component, the time of occurrence, the duration, and the vibration amplitude. When uploading privacy tag data, the normal operation data and minor fault records corresponding to the privacy tag are uploaded to the private chain through the first identifier and the second identifier.

6. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-modal fusion analysis according to claim 5, characterized in that, When a public link receives new public tag data, an integration instruction is automatically generated, triggering the data integration process. When performing fault frequency range matching, a fault frequency range table is preset. The number of faults for each wind turbine model is matched with the fault frequency range table to obtain the range corresponding to the number of faults. The total number of occurrences in that range is incremented by 1 and updated.

7. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis according to claim 6, characterized in that, For the first identifier, count the total number of occurrences of each faulty component and obtain the standard fault threshold corresponding to the faulty component. Then, compare the total number of occurrences of each faulty component with the corresponding standard fault threshold. If the total number is less than the standard fault threshold, then associate it with a normal component tag; Conversely, associate the abnormal component label; Sort by total number of components in descending order to obtain data on the local impact of the fault.

8. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-modal fusion analysis according to claim 7, characterized in that, When calculating the impact of a fault, the median of the total number of occurrences of each faulty component in the corresponding interval of the fault impact coordinate system is obtained in turn. The median obtained is multiplied by the corresponding total number of occurrences to obtain the local impact. The local impact of each faulty component is summed and the sum is set as the fault impact. The impact of the fault is compared and analyzed with the preset fault impact threshold. If the impact of the fault is less than the preset fault impact threshold, then associate it with the overall normal impact tag; Conversely, the overall abnormality will affect the label; Set the first identifier, local fault impact data, overall fault impact degree, and overall early warning label as fault fluctuation items and combine them to generate a fault fluctuation table; The fault fluctuation table is synchronously pushed to the equipment manufacturer's system, the wind farm operation and maintenance platform, and the regional dispatch center through the first identifier, and an alarm is triggered.

9. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-mode fusion analysis according to claim 8, characterized in that, Within a pre-defined regulatory period, the total number of data verifications that passed, the total number of data verifications that failed, the total number of data verifications that passed manual review, and the total number of data verifications that failed manual review are statistically analyzed. The values ​​of each statistical data item are extracted, and the first verification value corresponding to the data verification is calculated using a formula. and the second verification value reviewed by a human. Wherein, N1, N2, N3, and N4 represent the total number of data verifications that passed, the total number of data verifications that failed, the total number of data verifications that passed manual review, and the total number of data verifications that failed manual review, respectively. This is the first verification standard value. This is the second verification standard value.

10. The wind turbine monitoring big data analysis system based on voiceprint recognition multi-modal fusion analysis according to claim 9, characterized in that, Data analysis is performed on the first and second verification values ​​obtained from the calculation. If both the first and second verification values ​​are greater than 0, a valid overall verification label is generated, and the existing verification scheme is maintained. If either the first verification value or the second verification value is not greater than 0, a locally valid verification label is generated, and a local upgrade and optimization warning is issued for the verification scheme corresponding to the verification value not greater than 0. If both the first and second verification values ​​are not greater than 0, an overall verification anomaly label will be generated, and both the existing automatic verification scheme and the manual review verification scheme will be upgraded, optimized, and given an early warning.

Images