47results about How to "Improve fault detection accuracy" patented technology

The invention belongs to the technical field of image processing, and discloses a transformer fault detection method based on an improved YOLOv8 model. Aiming at the problems of inaccuracy in feature extraction, low precision, poor real-time detection capability and the like existing in fault detection of a traditional transformer fault detection method, the invention establishes a transformer fault detection model based on an improved YOLOV8 model architecture, a self-calibration convolution defogging module is introduced, a generative adversarial network (GAN) and a multi-source data fusion network structure are added, a BiFPN structure is used, an ASPP module is replaced by an ASPFCSPC module, and a YOLOV8 model architecture is improved, so that the feature extraction capability of a transformer fault detection model is enhanced, the calculation complexity of the model is reduced, and the fault detection efficiency of the transformer is improved. The transformer fault detection precision, accuracy and real-time detection capability are improved, and the stability and adaptability of the model are enhanced.

The invention discloses an optical fiber line fault detection method and system, and the method comprises the steps: employing broadband light sources at the two ends of an optical fiber in an optical fiber network at the same time, and carrying out the frequency coding of a signal in the broadband light sources, and obtaining an optical signal; collecting optical fiber sensing data; performing multi-mode acquisition on the optical signal to obtain multi-mode optical signal data; preprocessing the multi-mode optical signal data and the optical fiber sensing data to obtain clear signal data; analyzing and fusing the clear signal data to obtain fused data; and inputting the fusion data into a pre-purchased fault identification model, and outputting a fault type. According to the method, through technologies of multi-modal data fusion, high-frequency signal extraction, physical model constraint, feature compression and the like, the problems that the complex fault recognition capability is insufficient and the stability of the model is weakened by environmental interference are effectively solved, and the precision, robustness and generalization capability of optical fiber line fault detection are remarkably improved.

The invention discloses a power distribution network fault detection method and device based on a traveling wave signal, and the method comprises the following steps: obtaining a fault traveling wave signal of a power distribution network, and determining the fault position of the power distribution network; performing frequency decomposition on the fault traveling wave signal based on an optimization decomposition algorithm to obtain an instantaneous frequency; performing feature extraction on the instantaneous frequency based on a pre-constructed optimization fault detection model, and giving fault features of the power distribution network; and according to the fault characteristics of the power distribution network, giving a fault type corresponding to the fault traveling wave signal. According to the invention, the specific position and category of the power distribution network fault can be accurately determined, and the accuracy and efficiency of fault detection are improved.

The invention provides a new energy equipment system maintenance decision-making method and system based on a fault coupling model, and relates to the technical field of equipment predictive maintenance, and the method comprises the steps: collecting system structure information and system fault data of a complex new energy equipment system, and carrying out the maintenance decision-making of the complex new energy equipment system according to the system structure information and the system fault data; respectively constructing a function model view, a system level fault model view and each flow fault model view; constructing a fault coupling model corresponding to the complex new energy equipment system; probability estimation is carried out on the system fault data according to a kernel density estimation method, the fault probability of each fault node in the complex new energy equipment system is obtained, and key fault nodes in all the fault nodes and key node information corresponding to the key fault nodes are obtained; and mapping the key node information into a risk level corresponding to the key node information, and obtaining an optimal maintenance decision based on a decision tree analysis method and a plurality of maintenance decisions. According to the method, multi-dimensional analysis of a complex system can be realized, and fault modes of different components and subsystems can be identified more clearly, so that the fault detection and diagnosis efficiency is improved.

The invention relates to the technical field of power system power distribution network protection, in particular to a self-adaptive current differential protection method for a power distribution network with branch loads. Firstly, current data acquisition is carried out, intelligent electric meters are installed at power distribution network nodes, and timestamps are synchronized. Current phasor estimation is carried out, and the current phasor of each node is obtained through data preprocessing, soft threshold processing, wavelet reconstruction and phasor estimation; then, differential current is calculated, and the vector difference between adjacent nodes is determined; and determining an adaptive threshold according to historical fault data and real-time power grid load. The fault is judged by comparing the differential current with a threshold value, and the fault type is recognized by adopting a mode recognition algorithm. And finally, generating a protection decision based on a result, and taking measures of cutting off branches, selecting a protection mode according to a resistor, adjusting load distribution or limiting current and the like for short circuit, grounding, overload and other faults to improve fault detection accuracy, accurately identify fault types, reasonably generate and execute the protection decision and guarantee safe and stable operation of the power distribution network.

The invention relates to the technical field of power distribution network protection, and provides a power distribution network override trip protection method based on a distributed network. S1, arranging multiple layers of intelligent distributed protection devices at different nodes of the power distribution network, wherein each layer of protection device has different fault detection and protection decision capabilities; and S2, realizing localized data processing and fault detection through the edge calculation unit of each protection device, wherein the fault detection carries out real-time fault identification based on a deep learning model. According to the arrangement scheme of the multi-layer intelligent distributed protection device, the protection device is divided into a high-voltage layer, a medium-voltage layer and a low-voltage layer according to the voltage grade and the load importance, the technical effects of step-by-step fault isolation and layered protection are achieved, and compared with the arrangement mode of a unified protection device in the prior art, the protection device has the advantages of being simple in structure and convenient to use. The problem of large-range override tripping is effectively solved, layered response of different levels of protection devices to faults is realized, the fault influence range is further controlled, and the stability of the system is improved.

The invention belongs to the technical field of industrial equipment fault detection application, and particularly discloses a multi-channel audio acquisition and intelligent fusion coal mill fault voiceprint recognition system and method, and the system comprises a multi-channel audio acquisition module, an audio preprocessing and spectrogram conversion module, a weighted fusion module, a fault judgment and decision module, and an alarm push module. The identification method comprises the following steps: S1, multi-channel audio acquisition; s2, audio preprocessing and spectrogram conversion; s3, carrying out weighted fusion; s4, fault judgment and decision making; and S5, alarm pushing. The beneficial effects of the invention are that: 1) the multi-channel audio acquisition device comprehensively acquires sound signals of key parts of the coal mill, avoids information omission, can capture unique sound features of faults of different parts, provides rich data for accurate judgment, and improves the recognition capability; and 2) the weight fusion module based on the attention mechanism dynamically adjusts the weight according to the fault importance, highlights the key part information, and improves the fault detection accuracy.

The invention provides an automobile ECU repair instrument data processing method and device, and the method comprises the steps: obtaining the ECU repair instrument data, transmitting the ECU repair instrument data to a fault detection algorithm for fault detection, obtaining a fault detection result, and enabling the fault detection algorithm to be obtained through the optimization of a preset AI algorithm in combination with a first algorithm cost and a second algorithm cost. The first algorithm cost is not less than one of a data implicit representation cost and a defect detection cost, and the data implicit representation cost represents the distinction degree of the same vehicle machine fault in the ECU repair instrument data sample library between corresponding fault representation information in the sample implicit representation; the defect detection cost represents the distinction degree of the same vehicle machine faults in the ECU repair instrument data sample library between the corresponding fault classification pre-estimation information in the sample classification pre-estimation information. According to the method, the consistency of fault diagnosis can be ensured for data processing of the same vehicle fault type, so that the fault detection accuracy of a fault detection algorithm is improved, and the algorithm cost can be saved.

The invention discloses a vehicle trajectory anomaly detection method and device based on trajectory clustering and electronic equipment, and the method comprises the steps: obtaining a target feature vector generated after each intelligent beacon carries out the data collection of a passing target vehicle, and obtaining historical feature vectors corresponding to a plurality of historical vehicles; the elements contained in each feature vector comprise the track horizontal and vertical coordinates, the speed, the time and the magnetic field intensity when the vehicle passes through the smart beacon; for each smart beacon, clustering the same class of elements in all historical feature vectors of the smart beacon to obtain a clustering center for each class of elements of the smart beacon; connecting the clustering centers of each class of elements according to the sequence of the intelligent beacons to obtain a reference trajectory; all types of elements of the target vehicle are connected according to the sequence of the intelligent beacons to obtain a to-be-detected track; and performing similarity calculation on the reference trajectory of each type of element and the trajectory to be detected to determine whether the trajectory of the target vehicle is abnormal or not. Whether the track of the vehicle is abnormal or not can be judged according to the vehicle motion information returned by the beacon.

The invention relates to the technical field of fault detection, in particular to a steel belt fault detection method for spindle conveying, and the method comprises the steps: collecting echo signals received after an ultrasonic probe transmits ultrasonic waves to a steel belt in an operation process; obtaining each return band; calculating the echo time difference of each echo band; acquiring a peak-to-peak delay coefficient of each echo band; calculating an echo consistency coefficient of each echo band; obtaining all first quartered return bands; acquiring a steel belt echo attenuation value of the echo signal; after wavelet decomposition is carried out on the echo signal, a defect noise estimated value of each wavelet coefficient is obtained; and further obtaining a self-adaptive wavelet threshold value of each wavelet coefficient, and carrying out fault detection on the spindle transportation steel belt. According to the invention, the fault detection precision of the ultrasonic transmission mode is improved.

The invention relates to the technical field of electric power operation and maintenance, and particularly discloses a middle and low voltage distribution network fault online detection method based on cross wavelet coherence analysis, which comprises the following steps: carrying out cross wavelet coherence analysis on zero-sequence current of all lines, finding out a common high-energy frequency band, and selecting the line current with the highest energy as a detection object; monitoring the selected current sequence by using a sliding window method, and dividing the current sequence into small time periods for detection; analyzing the current waveform in each time window through a fractal theory, calculating the fractal dimension of the current waveform, and constructing a multi-scale fractal dimension feature vector; and using a Dirichlet process Gaussian mixture model to perform clustering analysis on the feature vectors so as to accurately identify faults. According to the medium and low voltage power distribution network fault online detection method based on cross wavelet coherence analysis, the optimal frequency band / scale can be determined in a self-adaptive mode, rapid fault detection is achieved in a short time window, and therefore the efficiency and accuracy of fault detection are remarkably improved.

The invention relates to a heterogeneous elevator fault detection method and system based on a deep reinforcement learning algorithm, and the method comprises the steps: obtaining elevator operation data, extracting a corresponding operation characteristic variable, carrying out the data format conversion of the operation characteristic variable through combining the operation time of an elevator, obtaining the operation characteristic data of a unified data format, and carrying out the fault detection of the elevator. Under the same elevator running time, the variable incidence relation between the related running characteristic data is analyzed, a fault prediction model of the heterogeneous elevator is constructed, fault characteristic variables and corresponding fault running states when the elevator is abnormal are obtained, the fault characteristic variables and the corresponding fault running states are input into the fault prediction model, and the fault prediction model is used for predicting the abnormal elevator running state. The method comprises the following steps: acquiring an elevator fault prediction result, carrying out fault judgment processing on the elevator fault prediction result, and carrying out fault tracing processing on the elevator according to the fault judgment result to obtain fault detection data of the heterogeneous elevator. The method has the effect of improving the fault detection accuracy of the heterogeneous elevators.

The invention relates to an electric energy meter fault detection method and device based on an electric energy meter fault detection model, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring sample broadband impedance spectroscopy data associated with a sample electric energy meter; performing feature extraction processing on the sample broadband impedance spectrum data to obtain a feature vector corresponding to the sample broadband impedance spectrum data; dividing the feature vectors to obtain a training data set and a verification data set; according to the training data set and the verification data set, training the electric energy meter fault detection model to obtain a trained electric energy meter fault detection model; and obtaining to-be-analyzed broadband impedance spectrum data associated with the to-be-detected electric energy meter, inputting the to-be-analyzed broadband impedance spectrum data to the trained electric energy meter fault detection model, obtaining a target working state of the to-be-detected electric energy meter, and taking the target working state as a fault detection result of the to-be-detected electric energy meter. By adopting the method, the fault detection accuracy of the electric energy meter can be improved.

The invention discloses an analog circuit fault detection method and device based on CNN-LSTM, readable storage equipment and a program product, and relates to the technical field of deep learning for circuit fault diagnosis. The problems that an existing fault diagnosis method based on deep learning is difficult to effectively deal with complex characteristics of high-dimensional data and cannot fully capture key characteristic information related to faults, so that the diagnosis effect is poor, and the comprehensiveness and reliability of diagnosis results are poor are solved. By constructing the CNN-LSTM fault detection model, spatial features and time sequence features in circuit fault signals can be processed at the same time, and accurate detection and classification of faults can be realized without depending on an accurate mathematical model. The method is mainly used for fault diagnosis of the analog circuit.

The invention discloses a photovoltaic panel fault detection method and device based on multi-branch feature interaction, equipment and a medium. The method comprises the following steps: acquiring a visible light-infrared image pair of a photovoltaic panel to be detected; through a preset target multi-branch fault detection model, image features are extracted from the visible light-infrared image pair, iterative feature interaction is carried out, and a plurality of input feature maps are generated; performing fault detection by adopting all the input feature maps through the target multi-branch fault detection model, and judging whether the photovoltaic panel to be detected has a fault or not; if yes, the fault position is marked through a target multi-branch fault detection model, and the fault category is output; and if not, continuously monitoring whether a new visible light-infrared image pair is received or not. Therefore, through the multi-branch structure of the fault detection model, feature interaction is carried out from different image dimensions to adapt to the condition of diversified arrangement scenes of the photovoltaic panel, and the fault detection precision of the photovoltaic panel is effectively improved.

The invention discloses an AI-based power distribution cabinet fault area detection method and system, and relates to the field of power distribution cabinet detection. The method is applied to a power distribution cabinet monitoring system, and comprises the following steps: acquiring multi-modal monitoring data of a target power distribution cabinet, and generating an operation condition vector of the target power distribution cabinet; inputting the operation condition vector into a preset typical condition vector library for condition decomposition to obtain a plurality of sub-operation condition vectors; inputting the plurality of sub-operation condition vectors into a preset power distribution cabinet fault detection model to obtain fault information corresponding to the plurality of sub-operation condition vectors; inputting the plurality of pieces of fault information into a preset fault reasoning knowledge graph to obtain a plurality of fault paths; screening from the plurality of fault paths to obtain a plurality of target fault paths, and identifying starting nodes in the plurality of target fault paths; and starting nodes in the multiple target fault paths are output, and fault early warning is generated, so that maintenance personnel can be reminded of timely maintenance. The problem of low fault detection precision of a current power distribution cabinet is solved.

The invention discloses a manufacturing equipment fault diagnosis system based on the Internet of Things, which comprises a data acquisition module, a node management module and a fault diagnosis module, and is characterized in that the data acquisition module is used for collecting operation state data, environment state data and manufacturing product data of equipment; the resource management module is used for detecting load conditions of processing nodes and allocating memory resources according to the load conditions, the fault diagnosis module is used for detecting equipment faults and analyzing equipment fault causes, and the data acquisition module, the resource management module and the fault diagnosis module are electrically connected with one another. The data acquisition module comprises an information input module and a data collection module, the information input module is used for inputting manufactured product data and operation state data of manufactured product equipment into the system, and the data collection module is used for collecting fault features and fault causes of the equipment. The method has the characteristics of high precision and high efficiency.

The invention discloses a vibration-based wind turbine generator blade anomaly detection method and system, and relates to the technical field of wind power. The wind turbine generator blade anomaly detection method based on vibration comprises the following steps: obtaining a sensitivity evaluation index; key feature parameter evaluation indexes are obtained; and judging whether the current vibration characteristics are abnormal or not. According to the method, the time domain analysis and the frequency domain analysis are performed on the preprocessed real-time vibration data to obtain the sensitivity evaluation index, and then whether key feature parameter extraction is executed or not is judged based on the obtained sensitivity evaluation index; and finally, comparing the obtained key characteristic parameter evaluation indexes with the vibration data reference model to judge whether the current vibration characteristics of the specified wind turbine generator blade are abnormal or not, so that the effect of improving the fault detection accuracy of the wind turbine generator blade in the vibration process is achieved. The problem of low fault detection accuracy in the vibration process of the wind turbine generator blade in the prior art is solved.

The invention provides a high-accuracy phased array radar subarray power supply unit health management method, which comprises hard fault detection and soft fault detection, and realizes power supply unit hard fault detection based on voltage, current and temperature detection circuits in order to detect hard faults such as overvoltage, overcurrent, short circuit and over-temperature of a subarray power supply unit; in order to detect a soft fault of a sub-array power supply unit caused by the change of basic parameters of a component, a core parameter waveform threshold of the component is determined based on Gaussian process regression taking a square exponential covariance function, a Matern covariance function and a periodic covariance function as kernel functions, and the threshold is dynamically adjusted according to the noisy condition of the waveform. And forming a soft fault safety threshold region, detecting a core parameter waveform of the basic component in real time, and comparing the core parameter waveform with the soft fault safety threshold region to realize soft fault detection. Through combination of soft and hard faults, the hard faults can be rapidly detected, and meanwhile, the failure problem of the device can be accurately predicted through a soft fault detection technology, so that intervention is performed in advance.

The application discloses a power transformer fault sound print detection method based on an extreme learning machine, first, audio signals of transformer faults are collected for framing, and preprocessed through fast Fourier transform; then, sound print characteristic parameters of the audio signals are extracted through an improved Mel filter, a fault sound print detection model based on the extreme learning machine is constructed, and model training is conducted; finally, a test set is established, real-time test samples are input into the trained fault sound print detection model, test identification is conducted, sound print identification results of transformer faults are obtained, and real-time fault sound print early warning is realized. The improved Mel filter can improve the spectral resolution of transformer sound print feature extraction, reduce the system complexity, the fault sound print detection model is constructed through the extreme learning machine algorithm, the transformer fault detection accuracy is effectively improved, and the method is easy to implement in engineering.

The invention belongs to the technical field of airborne embedded computers, and provides a processor starting method based on a dual-redundancy Flash storage architecture, and the method comprises the following steps: configuring a main Flash and a standby Flash which are mutually backed up for a processor, and storing the same bootstrap program for the main Flash and the standby Flash; in the starting process of the processor, sensing the starting state of the processor in real time, and judging whether the bootstrap program of the main Flash is abnormal or not according to the starting state; if the SPI chip selection signal is abnormal, dynamically switching the SPI chip selection signal of the processor to the standby Flash by the FPGA, triggering the processor to reset and restarting from the standby Flash; and after the processor is successfully started from the standby Flash, the bootstrap program of the main Flash is automatically repaired. According to the method, an NVRAM is not needed, all SPI interface processors are adapted, the false alarm rate is lower than 0.1%, the system recovery time is smaller than or equal to 2 seconds, and the starting reliability of the airborne system in the severe environment is remarkably improved.

This invention provides a fault detection method and system based on low-rank decomposition and common trend decoupling. The method first acquires historical process data under normal operating conditions of an industrial process as training data. Then, it decomposes the training data into clean data representing the main structural information and sparse noise data representing noise interference using low-rank decomposition. Next, it constructs a common trend learning model based on the clean data, reconstructs the non-stationary common trend, and uses the residual between the clean data and the non-stationary common trend as a stationary feature. Decoupling is achieved by applying stationarity constraints. Further, a fault detection model is obtained through joint optimization. In the detection stage, the process data to be detected is input into the model, and monitoring statistics based on the reconstruction error of the stationary feature and the common trend are constructed respectively. The result of the control limit comparison determines whether a fault has occurred, thereby improving the robustness of fault detection in non-stationary industrial processes.

The application relates to the technical field of data processing, and discloses a sulfur hexafluoride gas data monitoring management system. The system comprises the following units: a collecting unit that collects SF6 gas density values and temperature values, generates a monitoring original data set in combination with a device identification code and a time stamp; a screening unit that calculates a density-temperature rho-T correlation coefficient and screens reliable monitoring data; a prediction unit that inputs the reliable data into an SF6-NN neural network to predict density values and temperature compensation amounts; a division unit that stores data in a graded manner according to density ranges; and an analysis unit that calculates density stability indexes and device health degree scores and generates SF6 management data. The application solves the technical problems of incomplete data collection coverage, difficult guarantee of data reliability, lack of prediction capability and low data management efficiency in a traditional SF6 gas monitoring system, and improves the data quality, prediction accuracy and management efficiency of SF6 device monitoring.

This invention, entitled "A Fault Detection Method for Ship Gearboxes Based on Multi-Scale Convolutional Neural Networks and Bidirectional Attention Mechanisms," belongs to the field of intelligent operation and maintenance technology for ship gearboxes. The technical problem it aims to solve is the insufficient assessment accuracy in complex maritime environments. The key technical solutions involve adaptively extracting multi-band features of vibration signals using multi-scale CNNs, enhancing the focusing ability of key fault features through attention mechanisms, and combining this with bidirectional LSTM for deep mining of temporal dependencies.

The invention discloses a fault detection method and device for a distributed system and electronic equipment, and relates to the field of distributed technologies or other related technologies, and the method comprises the steps: carrying out the preprocessing of the original operation data of the distributed system, and obtaining a target data set; dividing the L nodes based on the calling relation data in the target data set to obtain M connected components; determining a target health score of each connected component based on the performance index data and the calling relation data in the target data set, the target health score being used for representing an operation state of a node in each connected component; and performing fault detection operation on the distributed system based on the target health score of each connected component. According to the invention, the technical problem of low accuracy of fault detection of the distributed system based on the prior art is solved.

This application relates to the field of fault detection technology, providing a fault detection method, electronic device, computer-readable storage medium, and program product based on multi-source time-series signals. The method can be applied to fault detection of elevator drum brakes, including: acquiring multi-source time-series signals of the elevator drum brake within a single operating cycle; constructing a graph structure based on the multi-source time-series signals; obtaining an energy difference coefficient based on the multi-source time-series signals and the graph structure, whereby the energy difference coefficient characterizes the degree of energy difference between electrical energy consumption and mechanical energy generation between adjacent operating events; inputting the graph structure and the energy difference coefficient into a deep learning model to obtain a graph embedding vector; and obtaining a fault probability value based on the graph embedding vector, whereby the fault probability value is used to determine the fault status of the elevator drum brake. This application can improve the monitoring sensitivity and detection accuracy of early local faults in elevator drum brakes.

The invention relates to the technical field of power data processing, and provides a multi-scene intelligent thermal imaging abnormal state comprehensive management system which integrates artificial intelligence and thermal imaging technologies and is used for monitoring temperature distribution of power equipment in real time and early warning abnormal states. The system comprises a data acquisition module, an equipment monitoring module, a fault classification module, a fault equipment determination module and a fault inspection unit. The device state is predicted by training the random forest model, the infrared detector is used for capturing the infrared radiation on the surface of the device, the fault device is rapidly positioned and detected, the system can also identify the fault sensor and continuously monitor the fault device, the stable operation of the power system is improved, and the power supply reliability is improved. The method effectively solves the problems that the number of devices in the new energy power plant is large, and the fault is difficult to locate, and improves the safety of the devices.

Embodiments of the present application provide a kind of coupling fault detection method, equipment and readable storage medium, electronic equipment obtains the first vibration signal of gear box, the second vibration signal of generator, the speed signal of generator and the power signal of generator in current period, determine multiple order rotation frequency according to speed signal, and then determine the vibration amplitude of coupling in each order rotation frequency in multiple order rotation frequency, to obtain multiple vibration amplitudes.In addition, electronic equipment also determines the target working condition interval of coupling, and then carries out fault detection according to target working condition interval and multiple vibration amplitudes.By using this kind of scheme, the vibration signal of the two ends of coupling is collected, and the vibration amplitude of coupling in each order rotation frequency is extracted, whether the coupling is faulty is detected based on vibration amplitude in real time, precise detection and reliable early warning of coupling fault are realized, the damage of wind turbine generator unit caused by the expansion of fault and the risk of ESH are avoided, and the purpose of guaranteeing the safe operation of wind turbine generator unit is realized.

The present application relates to the field of data processing technology, and specifically to a power supply control method for tunnel lighting and a multi-pulse rectifier transformer, the method comprising: obtaining temperature data, grounding current data and vibration data of the multi-pulse rectifier transformer at each acquisition moment; determining the temperature rise abnormality of the multi-pulse rectifier transformer; determining the correlation mutation coefficient between temperature and vibration; determining the current standard of the grounding current based on the discrete degree and distribution range of the grounding current data; determining the abnormal factor of the correlation between the grounding current and the temperature based on the relationship between the changing trend of the grounding current and the temperature, combined with the changing trend of the temperature; determining the abnormal index of the multi-pulse rectifier transformer, combining the temperature data, grounding current data and vibration data of the multi-pulse rectifier transformer under normal conditions, and judging the fault condition of the multi-pulse rectifier transformer. The present application aims to improve the fault detection accuracy of the multi-pulse rectifier transformer.

The application discloses a hard disk fault detection method, device and equipment and a medium, and relates to the field of hard disk detection.The method comprises the following steps: acquiring real-time vibration data and real-time noise data of a hard disk in operation; performing feature extraction and fusion on the real-time vibration data and the real-time noise data to obtain fused features; performing feature extraction on the real-time vibration data by using target statistical features to obtain statistical feature values; performing multi-layer wavelet analysis on the real-time noise data, and performing statistical analysis on the obtained wavelet coefficients to obtain a skewness value; splicing the fused features, the statistical feature values and the skewness value, and inputting the spliced features into a pre-trained long short-term memory network for multi-classification fault detection to determine the working state of the hard disk.The application can detect the hard disk fault at any time, realize early warning of the hard disk fault, ensure the real-time performance and accuracy of the hard disk fault detection, and ensure the data safety of a storage server with the hard disk as a main component.