CN-121995166-A - MMC fault detection method and system based on acousto-optic-electric fusion in wind power generation

CN121995166ACN 121995166 ACN121995166 ACN 121995166ACN-121995166-A

Abstract

The invention provides an MMC fault detection method and system based on acousto-optic-electric fusion in wind power generation. The system comprises an acoustic monitoring unit, an optical monitoring unit, an electric signal acquisition unit and an AI algorithm processing module. The acoustic monitoring unit adopts a non-contact acoustic emission and ultrasonic sensor to capture acoustic signals of partial discharge and mechanical vibration of the power module; the optical monitoring unit acquires the temperature field and discharge radiation information of the power device through infrared thermal imaging, ultraviolet detection and fiber bragg grating sensing, the electric acquisition unit acquires the voltage, current and high-frequency electromagnetic signals of the submodule, and the AI algorithm processing module utilizes the deep learning model to conduct fusion analysis on the multi-mode data, identify the fault type and evaluate the health state. The invention has the advantages of non-invasive and high-sensitivity on-line monitoring, can realize earlier discovery and more accurate positioning of the early failure of the MMC converter valve under the strong interference offshore environment, and remarkably improves the safety and self-healing capacity of the system operation.

Inventors

XU MINGCHUN
LV CAN
XIONG GAOXIN
FANG ZHONGLEI
CHEN JUN
TANG LUPING
Shen Macheng

Assignees

华能国际电力股份有限公司湖南清洁能源分公司
中国电建集团江西省水电工程局有限公司

Dates

Publication Date: 20260508
Application Date: 20260115

Claims (10)

1. The fault detection system based on acousto-optic-electric fusion in the wind power MMC system is characterized by comprising an acoustic monitoring unit, an optical monitoring unit, an electric acquisition unit and an AI algorithm processing module; The acoustic monitoring unit is used for collecting acoustic emission signals and ultrasonic signals of the MMC converter valve internal power module; the optical monitoring unit is used for collecting an infrared temperature field and an insulation discharge ultraviolet radiation signal of the power module; the electric acquisition unit is used for acquiring voltage, current and partial discharge electromagnetic signals of the power module; The AI algorithm processing module is used for carrying out fusion analysis on the multi-mode data from the acoustic monitoring unit, the optical monitoring unit and the electrical acquisition unit to identify faults, and controlling the MMC system according to the diagnosis result of the AI algorithm processing module.
2. The fault detection system based on acousto-optic fusion in the wind power MMC system according to claim 1, wherein the AI algorithm processing module comprises a sudden fault detection module and an intelligent feature analysis module, and performs fusion analysis on multi-mode data from an acoustic monitoring unit, an optical monitoring unit and an electric acquisition unit to identify faults, wherein the multi-mode data is output in two paths, one path directly enters the sudden fault detection module and is used for identifying short-time high-frequency anomalies, and the other path enters the intelligent feature analysis module and is used as input of deep feature extraction.
3. The fault detection system based on acousto-optic fusion in a wind power MMC system according to claim 1, wherein the acoustic monitoring unit comprises one or more acoustic emission sensors and ultrasonic sensors, which are installed on or near the power module packaging surface of the MMC converter valve for detecting acoustic signals generated by cracks, partial discharges inside the power device and vibration of the cooling component.
4. The fault detection system based on acousto-optic fusion in the wind power MMC system according to claim 1 is characterized in that the optical monitoring unit comprises an infrared thermal imager, an ultraviolet light sensor and an optical fiber Bragg grating sensor, wherein the infrared thermal imager is used for monitoring surface temperature distribution of an IGBT chip of an MMC submodule and a power connecting piece so as to detect local overheat faults, the ultraviolet light sensor is used for detecting local discharge ultraviolet radiation on the surface of an insulator in an MMC valve hall, and the optical fiber Bragg grating sensor is arranged at a key position of the submodule or a bus and used for measuring temperature change so as to realize distributed temperature monitoring.
5. The fault detection system based on acousto-optic and electric fusion in the wind power MMC system according to claim 1 is characterized in that the electric acquisition unit comprises a voltage sensor, a current sensor and an ultrahigh frequency partial discharge detection antenna, wherein the voltage sensor and the current sensor are connected to an MMC sub-module and used for acquiring capacitance voltage and bridge arm current signals of the sub-module in real time, and the ultrahigh frequency partial discharge detection antenna is installed in a converter valve hall and used for capturing ultrahigh frequency electromagnetic pulse signals generated by partial discharge.
6. The fault detection method based on acousto-optic-electric fusion in the wind power MMC system is characterized by comprising the following steps of: the acoustic monitoring unit collects acoustic emission signals and ultrasonic signals of the MMC converter valve internal power module; The optical monitoring unit collects an infrared temperature field and an insulation discharge ultraviolet radiation signal of the power module; the electric acquisition unit acquires voltage, current and partial discharge electromagnetic signals of the power module; The AI algorithm processing module performs fusion analysis on the multi-mode data from the acoustic monitoring unit, the optical monitoring unit and the electrical acquisition unit to identify faults, and performs control on the MMC system according to the diagnosis result of the AI algorithm processing module.
7. The fault detection method based on acousto-optic fusion in a wind power MMC system according to claim 6, wherein the method specifically comprises: The method comprises the steps of collecting original information from three dimensions of electricity, acoustics and optics through an acoustic monitoring unit, an optical monitoring unit and an electric collecting unit, enabling an electric signal to reflect voltage, current and electromagnetic discharge changes of a power module, recording mechanical vibration and acoustic emission events through an acoustic signal, monitoring temperature distribution and insulation states through an infrared sensor, an ultraviolet sensor and an optical fiber sensor, aligning time synchronization with data in a second step, enabling all mode data to correspond under the same time standard through a unified clock and timestamp mechanism, then entering a mode preprocessing stage, conducting filtering, denoising and standardization on the original signal, enabling the electric signal to adopt band-pass or notch filtering interference, enabling the acoustic signal to be subjected to envelope extraction to strengthen weak characteristics, enabling the optical signal to protrude out of a local discharge area through radiation calibration and image enhancement, enabling processed data to be converted into quantifiable characteristic vectors in a characteristic extraction stage, enabling the electric signal to extract energy, frequency and waveform mutation indexes, enabling characteristics of different modes to be unified to be mapped into a shared embedding space after the characteristics of different modes are normalized and aligned, enabling all modes to enter the same dimension mode preprocessing stage to enter a meaning mode preprocessing stage, enabling the same as the electrical fault type to be automatically fused with the current fault type, enabling the fault type to be automatically judged through an AI (analog fault type) to be obtained through an automatic fault analysis system, and finally judging the fault type of the fault type to be integrated, and the fault type of the fault type is integrated, and the fault type is automatically detected through an AI is integrated, and the fault type is detected by the fault type of the fault type is detected, the diagnosis results are fed back to the control system in the closed loop and self-learning stage, fault isolation, derating operation or alarm are realized, and new data are used for updating the AI model, so that the algorithm is continuously optimized and self-adaptive.
8. The fault detection method based on acousto-optic fusion in a wind power MMC system according to claim 7, wherein the AI algorithm processing module counts each partial discharge pulse into a phase distribution map according to the alternating current phase and the discharge signal amplitude of each partial discharge pulse, so as to form a partial discharge phase resolution spectrum, wherein the partial discharge phase resolution spectrum is used as one of characteristic inputs of acoustic and electric signal fusion and is used for identifying insulation discharge faults.
9. The fault detection method based on acousto-optic fusion in a wind power MMC system according to claim 7, wherein when the AI algorithm processing module receives a trend of continuous decrease of the health score of the power module, the AI algorithm processing module enters a preventive fault-tolerant control mode, automatically adjusts the operation parameters of the module to reduce the thermal stress or the electric stress of the module, and dispatches operation staff to overhaul, thereby preventing further deterioration of the fault.
10. The fault detection method based on acousto-optic fusion in the wind power MMC system according to claim 7, wherein the AI algorithm processing module comprises a sudden fault detection module and an intelligent feature analysis module, and the multi-mode data from the acoustic monitoring unit, the optical monitoring unit and the electric acquisition unit are subjected to fusion analysis to identify faults, and are output in two paths, wherein one path directly enters the sudden fault detection module and is used for identifying short-time high-frequency anomalies, and the other path enters the intelligent feature analysis module and is used as input of deep feature extraction.

Description

MMC fault detection method and system based on acousto-optic-electric fusion in wind power generation Technical Field The invention belongs to the field of power electronic equipment fault monitoring, and particularly relates to an MMC fault detection method and system based on acousto-optic-electric fusion in wind power generation. Background The rapid development of wind power places higher demands on long-range large-scale clean power transmission. Because wind farms are usually far away from inland load centers, the transmission distance can reach hundreds to thousands of kilometers, and traditional alternating current transmission is high in loss and low in efficiency due to reactive power compensation and stability limitation. The flexible direct current transmission technology becomes a main scheme of long-distance wind power grid connection by virtue of the advantages of low loss, large capacity, friendly weak current network and the like. The Modular Multilevel Converter (MMC) is widely applied to wind power transmission engineering as a core topology of a third generation voltage source type converter technology. The MMC overcomes the problem of voltage withstand of serial connection of devices through cascading of a large number of sub-modules, and realizes hundreds of megawatts to gigawatts engineering application in projects such as +/-500 kV north direct current power grids and +/-525 kV north sea wind power transmission. However, with the rise of wind farm scale and voltage class, thousands of IGBT power devices integrated in MMC valve halls face severe marine environments and long-term operational stresses, and their operational reliability is challenged unprecedented. How to find and locate the device-level faults in time has great significance for guaranteeing continuous and stable operation of the wind power flexible-direct system. The current state monitoring technology still has difficulty in fully meeting the requirements, and mainly has the following problems and disadvantages: The early fault sensitivity is lacking, and the traditional monitoring means mainly rely on an electric quantity sensor to collect signals such as voltage, current and the like of a submodule, so that serious faults can be found only on a millisecond time scale. For early degradation of devices such as bonding wire loosening and welding spot aging, and hidden defects such as partial discharge in a package, the pure electric quantity monitoring lacks sufficient spatial resolution and sensitivity, and is often difficult to capture initial symptoms of faults in time. The monitoring information is single, the functions of the existing monitoring system are relatively single, most of the existing monitoring system focuses on single-dimensional monitoring of electric signals or temperature signals, and the existing monitoring system cannot comprehensively sense the abnormality of multiple physical quantities such as acoustics, optics and the like. The single approach leads to insufficient criteria for complex faults, and misjudgment or missed detection easily occurs. The intelligent analysis and fault-tolerant decision-making are lacking, namely, the current monitoring system generally uploads the perception data to a remote control center, and the perception data is analyzed by manual or simple threshold judgment, so that the decision-making is difficult to make in time. The lack of an advanced intelligent algorithm for fusion processing of multi-source heterogeneous data can not fully mine the association between different fault characteristics, so that the uncertainty of fault early warning is high. Disclosure of Invention In order to solve at least part of the problems, the invention provides an MMC fault detection method and system based on acousto-optic-electric fusion in wind power generation. In order to achieve the above purpose, the present invention adopts the following technical scheme: The fault detection system based on acousto-optic-electric fusion in the wind power MMC system comprises an acoustic monitoring unit, an optical monitoring unit, an electric acquisition unit and an AI algorithm processing module; The acoustic monitoring unit is used for collecting acoustic emission signals and ultrasonic signals of the MMC converter valve internal power module; the optical monitoring unit is used for collecting an infrared temperature field and an insulation discharge ultraviolet radiation signal of the power module; the electric acquisition unit is used for acquiring voltage, current and partial discharge electromagnetic signals of the power module; The AI algorithm processing module is used for carrying out fusion analysis on the multi-mode data from the acoustic monitoring unit, the optical monitoring unit and the electrical acquisition unit to identify faults, and controlling the MMC system according to the diagnosis result of the AI algorithm processing module. The AI algorithm processing module compr