Grid fault rapid diagnosis method based on artificial intelligence technology

Abstract

The invention discloses a rapid power grid fault diagnosis method based on an artificial intelligence technology. The rapid power grid fault diagnosis method comprises the following steps of multi-modal data acquisition, edge calculation preprocessing, quantum feature coding, dynamic knowledge graph construction, hybrid intelligent diagnosis and combined intelligent analysis. According to the rapid power grid fault diagnosis method based on the artificial intelligence technology, multi-modal data fusion, quantum computing acceleration and a dynamic knowledge graph are used, and equipment states, power parameters and environment information are combined, so that rapid diagnosis and recovery are carried out on power grid faults by effectively combining data; and meanwhile, the reliability of power supply and the efficiency of operation and maintenance are remarkably improved, and experiments prove that accurate judgment and maintenance can be effectively and quickly performed on the artificial intelligence technology by performing accurate analysis on the acquired information by using big data; therefore, the problem of loss caused by the fact that a power grid is not timely maintained due to the traditional technology is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid fault diagnosis, and in particular to a method for rapid power grid fault diagnosis based on artificial intelligence technology. Background Art

[0002] The innovation in rapid power grid fault diagnosis technology stems from the deep integration of cross-domain methods. Early signal processing technology was limited to sampling frequency (only 0.1MHz), resulting in a capture rate of high-frequency transient components of less than 35%, which seriously restricted the accuracy of fault moment calibration. Traditional matrix algorithms did not consider nonlinear interference such as circuit breaker action delay and transformer saturation when constructing power grid correlation models, resulting in an error rate of over 12%.

[0003] When used in the existing technology, the existing power grid fault diagnosis has the problem of data dimension limitation, the model calculation efficiency is insufficient, and the existing knowledge thrust is relatively weak, which makes it impossible to deeply and quickly diagnose the faults in the power grid. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for rapid diagnosis of power grid faults based on artificial intelligence technology to solve the problem of data dimension limitations in the existing power grid fault diagnosis in the above-mentioned background technology. At the same time, the model calculation efficiency is insufficient, and the existing knowledge thrust is relatively weak, which makes it impossible to deeply and quickly diagnose the faults in the power grid.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for rapid diagnosis of power grid faults based on artificial intelligence technology, comprising the following steps: multimodal data acquisition, edge computing preprocessing, quantum feature encoding, dynamic knowledge graph construction, hybrid intelligent diagnosis and combined intelligent analysis; The multimodal data collection includes collecting internal and external data of the power system through distributed fiber optic sensors, drone inspection systems, and Beidou positioning equipment; The edge computing preprocessing includes denoising, feature extraction and lightweight model reasoning of data at the edge node; The quantum feature coding includes mapping the fault signal into a quantum state and designing a quantum gate operation coding feature; The dynamic knowledge graph construction includes integrating multi-source data to build a "equipment-fault-cause" knowledge network; The hybrid intelligent diagnosis and control combines the Transformer-GNN model with reinforcement learning to achieve fault classification, location and recovery strategy optimization; The combined intelligent analysis transmits the specific diagnostic data directly to the Internet, performs artificial intelligence analysis through DeepSeek, and provides maintenance strategies based on actual conditions.

[0006] Preferably, the multimodal data acquisition obtains three types of data: power parameters, equipment status, and environmental information through devices such as vibration sensors, infrared thermal imagers, and drones.

[0007] Preferably, the edge computing preprocessing performs noise reduction, feature extraction and anomaly screening on the original data at the edge node, compressing the data transmission volume by more than 70%, using Huawei Atlas 500 3000 model devices, and supporting local inference of Huawei Ascend 910C.

[0008] Preferably, the quantum feature encoding reduces computational complexity through quantum gate decomposition technology.

[0009] Preferably, the dynamic knowledge graph constructs real-time data updates, supports explainable reasoning of fault causes, and combines with the rule engine to trace the fault causal chain to generate explainable diagnostic conclusions.

[0010] Preferably, the hybrid intelligent diagnosis automatically triggers equipment control, work order generation or user warning according to the fault confidence and impact range.

[0011] Preferably, the power parameters include current, voltage, frequency, and harmonic distortion rate; the equipment status includes vibration acceleration, infrared temperature, and dissolved gas concentration in oil; and the environmental information includes wind speed, precipitation, user complaint text, and drone inspection images.

[0012] Preferably, the edge computing preprocessing includes wavelet packet decomposition and energy feature extraction of vibration signals, target detection and temperature rise rate calculation of infrared images, and natural language processing and keyword extraction of text data.

[0013] Compared with the prior art, the present invention has the following beneficial effects: First, the present invention uses multimodal data fusion, quantum computing acceleration and dynamic knowledge graphs, and combines equipment status, power parameters and environmental information to effectively combine data to quickly diagnose and recover power grid faults, while significantly improving power supply reliability and operation and maintenance efficiency. Experiments have shown that by using big data to accurately analyze the information obtained, artificial intelligence technology can be effectively and quickly judged and maintained, thereby avoiding the problem of losses caused by power grid faults not being maintained in time due to traditional technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of the rapid fault diagnosis process of the present invention; Figure 2 This is a schematic diagram of the fusion formula of the present invention; Figure 3 Schematic diagram of the loss formula of the present invention. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0016] See also Figure 1-2 ,A fast diagnosis method for power grid faults based on artificial intelligence technology, includes the following steps: multi-modal data acquisition, edge computing preprocessing, quantum feature encoding, dynamic knowledge graph construction, hybrid intelligent diagnosis, and combined intelligent analysis; The multimodal data collection includes collecting internal and external data of the power system through distributed fiber optic sensors, drone inspection systems, and Beidou positioning equipment; The edge computing preprocessing includes denoising, feature extraction and lightweight model reasoning of data at the edge node; The quantum feature coding includes mapping the fault signal into a quantum state and designing a quantum gate operation coding feature; The dynamic knowledge graph construction includes integrating multi-source data to build a "equipment-fault-cause" knowledge network; The hybrid intelligent diagnosis and control combines the Transformer-GNN model with reinforcement learning to achieve fault classification, location and recovery strategy optimization; The combined intelligent analysis transmits the specific diagnostic data directly to the Internet, performs artificial intelligence analysis through DeepSeek, and provides maintenance strategies based on actual conditions.

[0017] Furthermore, the multimodal data acquisition obtains three types of data: power parameters, equipment status, and environmental information through vibration sensors, infrared thermal imagers, drones and other equipment.

[0018] Furthermore, the edge computing preprocessing performs noise reduction, feature extraction, and anomaly screening on the raw data at the edge node, compressing the data transmission volume by more than 70%. It uses Huawei Atlas 500 3000 model devices and supports local inference of Huawei Ascend 910C.

[0019] Furthermore, the quantum feature encoding reduces computational complexity through quantum gate decomposition technology.

[0020] Furthermore, the dynamic knowledge graph constructs real-time data updates, supports explainable reasoning of fault causes, and combines with the rule engine to trace the fault causal chain to generate explainable diagnostic conclusions.

[0021] Furthermore, the hybrid intelligent diagnosis automatically triggers equipment control, work order generation or user warning according to the fault confidence and impact range.

[0022] Furthermore, the power parameters include current, voltage, frequency, and harmonic distortion rate; the equipment status includes vibration acceleration, infrared temperature, and dissolved gas concentration in oil; and the environmental information includes wind speed, precipitation, user complaint text, and drone inspection images.

[0023] Furthermore, the edge computing preprocessing includes wavelet packet decomposition and energy feature extraction of vibration signals, target detection and temperature rise rate calculation of infrared images, and natural language processing and keyword extraction of text data.

[0024] Implementation method 1: Multimodal data fusion and spatiotemporal alignment are used to synchronously collect electrical quantities (voltage / current waveforms, sampling rate ≥ 10kHz), equipment temperature (accuracy ±0.5°C), and meteorological data (wind speed, lightning location). Multi-node data is synchronized using the GPS clock, with a time error of ≤ 0.1ms. A power grid topology correlation matrix is established to derive the fusion formula.

[0025] Implementation 2: A hybrid deep learning model is constructed, utilizing spatial feature extraction and a graph convolutional network (GCN) to model device topology. Node features include electrical parameters and temperature gradients, and edge weights are calculated based on line impedance. Time series features are then extracted, and a bidirectional LSTM is used to process the fault waveform propagation path. The sliding window length is set to 10 power frequency cycles (200ms). Status judgment is performed by setting overloaded device nodes, specifically when the temperature is greater than 80°C or the current exceeds the limit by 30%, to derive a loss function. FocalLoss addresses sample imbalance (normal:faulty = 100:1). TopoConsistencyLoss constrains the spatial consistency of device nodes.

[0026] Implementation method 3: Through a collaborative fault judgment and prediction system, a real-time diagnosis module is set up to output the probability distribution of fault types (Softmax classification) and the heat map of the top three suspicious nodes. A prediction and warning module is set up to predict the equipment status for the next hour based on the LSTM-Transformer model. The input is: historical 72-hour electrical quantity and temperature trends; the output is: overload risk probability and recommended maintenance time window.

[0027] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for rapid diagnosis of power grid faults based on artificial intelligence technology, characterized in that: The following steps are involved: Multimodal data acquisition, edge computing preprocessing, quantum feature encoding, dynamic knowledge graph construction, hybrid intelligent diagnosis and combined intelligent analysis; The multimodal data collection includes collecting internal and external data of the power system through distributed fiber optic sensors, drone inspection systems, and Beidou positioning equipment; The edge computing preprocessing includes denoising, feature extraction and lightweight model reasoning of data at the edge node; The quantum feature coding includes mapping the fault signal into a quantum state and designing a quantum gate operation coding feature; The construction of the dynamic knowledge graph includes integrating multi-source data to build a "equipment-fault-cause" knowledge network; The hybrid intelligent diagnosis and control combines the Transformer-GNN model with reinforcement learning to achieve fault classification, location and recovery strategy optimization; The combined intelligent analysis transmits the specific diagnostic data directly to the Internet, performs artificial intelligence analysis through DeepSeek, and provides maintenance strategies based on actual conditions.

2. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1, characterized in that: The multimodal data acquisition acquires three types of data: power parameters, equipment status, and environmental information through vibration sensors, infrared thermal imagers, drones and other equipment.

3. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1, characterized in that: The edge computing preprocessing performs noise reduction, feature extraction, and anomaly screening on the raw data at the edge node, compressing the data transmission volume by more than 70%. It uses Huawei Atlas 500 3000 model devices and supports local inference of Huawei Ascend 910C.

4. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1, characterized in that: The quantum feature encoding reduces computational complexity through quantum gate decomposition technology.

5. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1, characterized in that: The dynamic knowledge graph builds real-time data updates, supports explainable reasoning of fault causes, and combines with the rule engine to trace the fault causal chain to generate explainable diagnostic conclusions.

6. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1, characterized in that: The hybrid intelligent diagnosis automatically triggers equipment control, work order generation or user warning according to the fault confidence and impact scope.

7. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 2, characterized in that: The power parameters include current, voltage, frequency, and harmonic distortion rate; the equipment status includes vibration acceleration, infrared temperature, and dissolved gas concentration in oil; and the environmental information includes wind speed, precipitation, user complaint text, and drone inspection images.

8. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 3 is characterized by: The edge computing preprocessing includes wavelet packet decomposition and energy feature extraction of vibration signals, target detection and temperature rise rate calculation of infrared images, and natural language processing and keyword extraction of text data.

Images