Equipment fault detection method and device, computer program product and electronic equipment

Abstract

The invention discloses an equipment fault detection method and device, a computer program product and electronic equipment. Relates to the field of artificial intelligence and big data, and the method comprises the steps: obtaining target monitoring data of a target device, the target monitoring data comprising at least one of strain data, temperature, wind speed and verticality, and the verticality being used for representing the stability of the target device; inputting the target monitoring data into a first target model to obtain monitoring features; and under the condition that the monitoring characteristics represent that the monitoring data is abnormal, the monitoring characteristics are input into a second target model to obtain a fault detection result of the target equipment, the second target model is obtained by training multiple groups of second training samples, and each group of second training samples comprises historical monitoring characteristics and a historical fault detection result. Through the method and the device, the problem of relatively low fault detection efficiency of equipment in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the fields of artificial intelligence and big data, and specifically to a method, apparatus, computer program product, and electronic device for detecting equipment faults. Background Art

[0002] With the advent of 5G (fifth-generation wireless system), the Internet of Things (IoT), and the rapid growth of IT (information technology) cloud computing applications, the number of connected business terminals (such as cameras and positioning terminals) is increasing. However, these devices face pressing challenges, including low troubleshooting efficiency, difficulty in delegating responsibilities, and high operation and maintenance costs. Some terminals are installed on signal towers. Due to the widespread distribution of these sites, many of which are remote and difficult to access, on-site troubleshooting of equipment failures is time-consuming and maintenance is inefficient. Different faults (such as those related to power, network, and equipment) are often attributed to different parties, making rapid diagnosis and fault location difficult and communication costs high. Equipment failures require manual on-site inspections, resulting in slow repair times and high maintenance costs.

[0003] In related technologies, equipment operation and maintenance monitoring relies on manually set thresholds and manual intervention. Once a fault occurs, the monitoring system fails to trigger an alarm, resulting in problems not being discovered in a timely manner. The alarm system also suffers from false positives and missed alerts, and lacks intelligent analysis, making it difficult for operators to quickly locate the cause of the fault. When faults involve multiple layers (such as hardware, operating systems, and applications), information silos easily form during the diagnostic process, making it difficult for operators to quickly and accurately locate the fault point from a vast amount of logs and indicators.

[0004] Currently, no effective solution has been proposed to the problem of low fault detection efficiency of equipment in related technologies. Summary of the Invention

[0005] The main purpose of this application is to provide a device fault detection method, apparatus, computer program product and electronic device to solve the problem of low fault detection efficiency of devices in related technologies.

[0006] To achieve the above-mentioned objectives, according to one aspect of the present application, a method for fault detection of a device is provided. The method comprises: obtaining target monitoring data of a target device, wherein the target monitoring data comprises at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device; inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples comprises historical monitoring data and historical monitoring features, and the monitoring features comprise normal data and abnormal data; when the monitoring features represent the presence of abnormalities in the monitoring data, inputting the monitoring features into a second target model to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples comprises historical monitoring features and historical fault detection results.

[0007] Optionally, the first target model is obtained by: obtaining multiple historical monitoring data collected from sensors on the target device, comparing each historical monitoring data with preset standard data to obtain a comparison result, wherein the standard data is the monitoring data of the target device when it is not in a faulty state; determining the historical monitoring characteristics of each historical monitoring data based on the comparison result, determining each historical monitoring data and its corresponding historical monitoring characteristics as a group of first training samples, and obtaining multiple groups of first training samples; training the support vector machine through the multiple groups of first training samples to obtain the first target model.

[0008] Optionally, the second target model is obtained by: obtaining a set of historical monitoring features within a preset period, and determining the predicted monitoring features at the target moment based on the set of historical monitoring features; determining the monitoring features at the target moment, calculating the difference between the monitoring features and the predicted monitoring features, and when the difference is greater than or equal to a preset threshold, determining the historical fault monitoring result of the monitoring features at the target moment as the presence of a fault; when the difference is less than a preset threshold, determining the historical fault monitoring result of the monitoring features at the target moment as the absence of a fault; determining the historical monitoring features at each moment and the historical fault detection results of the historical monitoring features as a set of second training samples to obtain multiple sets of second training samples; training the long short-term memory network through multiple sets of second training samples to obtain the second target model.

[0009] Optionally, after obtaining the target monitoring data of the target device, the method further includes: performing wavelet transform processing on the target monitoring data to obtain the data to be denoised; determining a preset noise threshold, calculating the difference between the data to be denoised and the noise threshold, and performing inverse wavelet transform processing on the difference to obtain the target monitoring data after denoising.

[0010] Optionally, after obtaining the fault detection result of the target device, the method also includes: when the fault detection result indicates that the target device has a fault, issuing an early warning message, wherein the early warning message includes abnormal data in the monitoring data; determining a preset fault maintenance strategy through the abnormal data in the early warning message, and executing the fault maintenance strategy; when the fault detection result indicates that the target device does not have a fault, continuing to execute the step of obtaining the monitoring data of the target device.

[0011] Optionally, the method further includes: collecting all fault detection results of the target device within a preset period at intervals of a preset period; determining the number of fault detection results within the preset period that indicate faults, and obtaining the number of fault results; calculating the ratio of the number of fault results to the number of all fault detection results, and obtaining a failure rate; when the failure rate is greater than or equal to a failure rate threshold, determining whether the fault detection result matches the actual operating status of the target device; and when the fault detection result matches the actual operating status of the target device, issuing a prompt message, wherein the prompt message is used to prompt the operation and maintenance personnel that there is an abnormality in the target device.

[0012] Optionally, after determining whether the fault detection result matches the actual operating status of the target device, the method also includes: when the fault detection result does not match the actual operating status of the target device, determining the target fault detection result of the detection failure; obtaining the monitoring characteristics corresponding to the target fault detection result, and determining the monitoring characteristics and the actual operating status as a newly added second training sample; combining the newly added second training sample with multiple groups of second training samples to obtain updated multiple groups of second training samples; training the long short-term memory network based on the updated multiple groups of second training samples to obtain an updated second target model.

[0013] To achieve the above-mentioned purpose, according to another aspect of the present application, a device fault detection apparatus is provided. The apparatus comprises: an acquisition unit for acquiring target monitoring data of a target device, wherein the target monitoring data comprises at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device; a first input unit for inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples comprises historical monitoring data and historical monitoring features, and the monitoring features comprise normal data and abnormal data; a second input unit for inputting the monitoring features into a second target model when the monitoring features represent that the monitoring data are abnormal, to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples comprises historical monitoring features and historical fault detection results.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of the fault detection method of the device described in each embodiment of the present application.

[0015] Through this application, the following steps are adopted: obtaining target monitoring data of the target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed and verticality, and the verticality is used to characterize the stability of the target device; inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; when the monitoring features characterize the presence of abnormalities in the monitoring data, inputting the monitoring features into a second target model to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples includes historical monitoring features and historical fault detection results, thereby solving the problem of low fault detection efficiency of the device in the related art. By obtaining multi-dimensional monitoring data, extracting the monitoring features of the monitoring data based on the first target model, and then determining the fault detection results through the second target model and the monitoring features, the effect of improving the fault detection efficiency of the device is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0017] Figure 1 is a flowchart of a method for detecting a fault in a device according to an embodiment of the present application;

[0018] Figure 2 is a schematic diagram of a device fault monitoring system provided according to an embodiment of the present application;

[0019] Figure 3 is a schematic diagram of a device fault monitoring method provided according to an embodiment of the present application;

[0020] Figure 4 is a schematic diagram of a fault detection device for a device provided in an embodiment of the present application;

[0021] Figure 5 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0023] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] The present invention will be described below in conjunction with preferred implementation steps. Figure 1 is a flow chart of a method for detecting a fault in a device according to an embodiment of the present application. Figure 1 As shown, the method includes the following steps:

[0026] Step S101 : acquiring target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device.

[0027] In step S101, the target device may be a tower. Fault monitoring relies on extensive real-time data collection. Sensors (such as those for temperature, pressure, vibration, and sound) monitor the device and collect various operational indicators. Leveraging IoT technology, these sensors can transmit this data in real time to the cloud for processing and analysis. Various sensors installed on the tower collect sensor information, generating multi-source data, or target monitoring data. This target monitoring data may include, but is not limited to, strain, temperature, wind speed, and verticality. Sensor placement can be tailored to the tower's design and operational experience, targeting vulnerable areas and critical nodes. For example, strain sensors collect strain data, which directly impacts tower safety and stability, necessitating multi-point, targeted deployment.

[0028] The target monitoring data is transmitted to the equipment's fault detection device through the power grid communication network. Due to the lack of wired communication in some areas where the tower station is located, some wireless communications cannot be fully covered. Therefore, the power grid communication network, such as OPGW (Optical Fiber Composite Overhead Ground Wire) optical cable, is used as a signal carrier. Using OPGW optical cable for network transmission can effectively improve signal transmission efficiency and stability, avoiding the disadvantages of incomplete coverage and susceptibility to interference of wireless communication.

[0029] Step S102: input the target monitoring data into the first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data.

[0030] In step S102, the first target model can be a machine learning model using a supervised learning algorithm such as a support vector machine, decision tree, or random forest. This model is used to perform classification and regression analysis on the target monitoring data to help identify potential equipment failures. For example, the first target model is a support vector machine. The support vector machine classifies the target monitoring data, accurately distinguishing between normal and abnormal data within each type of monitoring data, determining the operating status of the equipment, and obtaining real-time status information, i.e., monitoring features, of the target equipment.

[0031] Step S103: When the monitoring feature indicates that the monitoring data is abnormal, the monitoring feature is input into the second target model to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, and each group of second training samples includes historical monitoring features and historical fault detection results.

[0032] In step S103, the second target model can be an LSTM (Long Short-Term Memory) model. LSTM is suitable for processing and predicting time series data, natural language processing tasks, and the like. When the first target model identifies monitoring features indicating an anomaly in the monitoring data, the monitoring features are input into the second target model for deeper fault detection. The second target model is trained using machine learning or deep learning methods, and its training process relies on a large amount of collected monitoring data and its corresponding fault detection results. When training the second target model, historical monitoring features are first input into the model. Through internal calculations, the model predicts the corresponding fault detection results. The predicted results are then compared with the actual historical fault detection results, a loss function is calculated, and the model parameters are adjusted through backpropagation and weight updates to minimize the gap between the predicted and actual results.

[0033] After the second target model is trained, the monitoring features are fed into it. Based on the patterns and rules learned during training, the model analyzes the input features to determine whether the target device has experienced a fault and the specific type of fault. If the output fault detection results indicate a device failure, the early warning system will immediately activate and issue a warning signal, allowing for timely maintenance or other necessary measures.

[0034] The fault detection method of the equipment provided in the embodiment of the present application obtains target monitoring data of the target equipment, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed and verticality, and the verticality is used to characterize the stability of the target equipment; the target monitoring data is input into the first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; when the monitoring features characterize the presence of abnormalities in the monitoring data, the monitoring features are input into the second target model to obtain the fault detection results of the target equipment, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples includes historical monitoring features and historical fault detection results, thereby solving the problem of low fault detection efficiency of equipment in related technologies. By obtaining multi-dimensional monitoring data, extracting the monitoring features of the monitoring data based on the first target model, and then determining the fault detection results through the second target model and the monitoring features, the effect of improving the fault detection efficiency of the equipment is achieved.

[0035] In order to identify whether there is abnormal data in the monitoring data, it is necessary to train a first target model. Optionally, in the fault detection method of the device provided in the embodiment of the present application, the first target model is obtained in the following manner: obtaining multiple historical monitoring data collected from the sensor on the target device, comparing each historical monitoring data with preset standard data to obtain a comparison result, wherein the standard data is the monitoring data of the target device when there is no fault; determining the historical monitoring features of each historical monitoring data based on the comparison result, determining each historical monitoring data and its corresponding historical monitoring features as a group of first training samples, and obtaining multiple groups of first training samples; training the support vector machine through the multiple groups of first training samples to obtain the first target model.

[0036] In some examples, a large amount of historical monitoring data is first collected from sensors on the target device (such as strain sensors, temperature sensors, wind speed sensors, etc.). The historical monitoring data includes data from both normal operation and fault conditions of the device, so that the model can learn features that distinguish normal data from abnormal data. Each piece of historical monitoring data is compared with preset standard data. Standard data is the normal operating data of the target device when no fault conditions occur. This can be statistical data collected during the initial installation of the device or after a long period of stable operation to represent the "healthy" state of the device. Based on the comparison results, historical monitoring features are extracted for each piece of historical monitoring data. If a piece of historical monitoring data deviates significantly from the standard data, for example, a strain value outside the normal range, an abnormally high temperature, or an abnormal wind speed, these deviations will be used as historical monitoring features for subsequent model training.

[0037] Each piece of historical monitoring data and its corresponding extracted historical monitoring features are combined into a set of first training samples. If the features of a piece of historical monitoring data differ significantly from the standard data, this set of samples may be labeled as abnormal or faulty; if there are no significant differences, they may be labeled as normal. In this way, a sample library containing both normal and abnormal device states is constructed. These first training samples (containing normal and abnormal monitoring data and their features) are used to train the support vector machine model. During training, the support vector machine model searches for an optimal hyperplane (which may be a hypersurface in high-dimensional space) that maximizes the separation between the two types of samples (normal and abnormal) while ensuring the maximum margin distance to improve the model's generalization ability. During training, the support vector machine parameters (such as the kernel function, regularization parameter C, tolerance, etc.) are adjusted to optimize model performance.

[0038] This embodiment trains a first target model to initially determine whether the monitoring data is abnormal. If the monitoring data deviates from the standard data range, the model will identify abnormal patterns in the data based on the features and classification boundaries learned during training. If the monitoring features indicate an anomaly, the data is further input into the second target model for more detailed fault detection and location. This two-layer model design enables rapid detection and intelligent early warning of tower equipment faults. The first target model serves as an initial screening, while the second target model performs more precise fault diagnosis, effectively improving fault response speed and maintenance efficiency.

[0039] In order to accurately detect equipment failures, after obtaining the monitoring features, a second target model is required to identify the fault detection results. Optionally, in the fault detection method for the equipment provided in the embodiment of the present application, the second target model is obtained in the following manner: obtaining a set of historical monitoring features within a preset period, and determining the predicted monitoring features at the target moment based on the set of historical monitoring features; determining the monitoring features at the target moment, calculating the difference between the monitoring features and the predicted monitoring features, and when the difference is greater than or equal to a preset threshold, determining the historical fault monitoring result of the monitoring features at the target moment as the presence of a fault; when the difference is less than the preset threshold, determining the historical fault monitoring result of the monitoring features at the target moment as the absence of a fault; determining the historical monitoring features at each moment and the historical fault detection results of the historical monitoring features as a set of second training samples to obtain multiple groups of second training samples; training the long short-term memory network through multiple groups of second training samples to obtain the second target model.

[0040] In some examples, a set of historical monitoring features over a preset period is acquired from sensors on a target device. This data set can include monitoring values for strain, temperature, wind speed, verticality, and other parameters at multiple time points, covering the device's operation under both normal and faulty conditions. Based on this set of historical monitoring features, statistical or machine learning methods are used to predict the monitoring features at the next time point (i.e., the target moment). A sliding window technique can be used to input historical data into a learning model to predict the data at the target moment. After obtaining the actual monitoring features at the target moment, the difference between the actual monitoring features and the predicted features is calculated. If the difference is greater than or equal to a preset threshold, the actual monitoring features deviate significantly from the predicted values, potentially indicating a device fault. If the difference is less than the preset threshold, the device is considered to be operating normally and not faulty. The preset threshold can be determined through statistical analysis or expert experience, taking into account the noise and variation range of the historical data.

[0041] The historical monitoring features and their corresponding fault detection results (fault presence or absence) at each time point are used as a set of second training samples. Using this constructed second training sample set, an LSTM model is trained. LSTMs are capable of effectively processing and predicting time series data, capturing long-term dependencies within the sequence. LSTM model training aims to minimize the difference between predicted and actual fault detection results, i.e., the model's loss function. During training, model parameters (such as the number of network layers, nodes, and learning rate) are adjusted to optimize model performance. Regularization is also required to prevent overfitting. After training, the performance of the LSTM model is evaluated through cross-validation or an independent validation set to ensure that it can accurately predict faults even on unseen data.

[0042] This embodiment trains a second target model to perform real-time monitoring and fault detection on tower equipment. When real-time monitoring features are fed into the second target model, it can predict the equipment's operating status and determine whether a fault exists based on its learned patterns and regularities. This second target model enables intelligent fault detection and early warning for tower equipment, improving the efficiency and accuracy of equipment maintenance, reducing downtime caused by faults, and ensuring stable operation of the communication network.

[0043] In order to improve the accuracy of monitoring feature extraction, the monitoring data can be denoised before extracting the monitoring features. Optionally, in the fault detection method of the device provided in the embodiment of the present application, after obtaining the target monitoring data of the target device, the method also includes: performing wavelet transform processing on the target monitoring data to obtain data to be denoised; determining a preset noise threshold, calculating the difference between the data to be denoised and the noise threshold, and performing wavelet inverse transform processing on the difference to obtain the target monitoring data after denoising.

[0044] In some cases, raw monitoring data contains noise, missing values, and other issues, necessitating preprocessing through signal processing and data cleaning techniques. Feature engineering techniques can be used to extract features that aid fault diagnosis, such as time series features and frequency domain features, from large amounts of data. Wavelet transforms are a signal processing tool used to reduce the noise of non-stationary signals (i.e., signals whose characteristics vary over time). Using wavelet transforms to process target monitoring data can effectively remove high-frequency noise, retain useful information, and improve the accuracy of subsequent fault detection.

[0045] The wavelet basis function determines the properties of the wavelet transform, and a wavelet basis that matches the characteristics of the monitoring data is selected. The target monitoring data is input into the wavelet transform to obtain a series of wavelet coefficients. The wavelet coefficients represent the characteristics of the signal at different scales and times. The high-frequency part of the wavelet coefficients represents noise or detail information, and the low-frequency part represents the basic trend or skeleton of the signal. Based on the statistical characteristics of the wavelet coefficients (such as standard deviation, median absolute deviation, etc.), a noise threshold is set to distinguish between noise components and useful components in the signal. For each wavelet coefficient, if its absolute value is less than the noise threshold, it is set to zero (hard threshold processing) or reduced according to certain rules (soft threshold processing). The wavelet coefficients that have undergone threshold processing are input into the inverse wavelet transform to reconstruct the denoised signal.

[0046] This embodiment transforms raw noisy monitoring data into noise-reduced target monitoring data, improving signal clarity and the performance of subsequent fault detection algorithms. Wavelet transform noise reduction accurately identifies and removes high-frequency noise, preserving the signal's fundamental form and key features, providing a more reliable and pure data foundation for fault prediction.

[0047] After determining that a fault exists in the target device, it is necessary to issue an early warning message and execute a preset fault maintenance strategy. Optionally, in the fault detection method for the device provided in the embodiment of the present application, after obtaining the fault detection result of the target device, the method further includes: when the fault detection result indicates that a fault exists in the target device, issuing an early warning message, wherein the early warning message includes abnormal data in the monitoring data; determining a preset fault maintenance strategy based on the abnormal data in the early warning message, and executing the fault maintenance strategy; when the fault detection result indicates that there is no fault in the target device, continuing to execute the step of obtaining the monitoring data of the target device.

[0048] In some examples, the first target model continuously analyzes the monitoring data of the target device to identify possible abnormal data points. The second target model conducts in-depth analysis of the abnormal data points to determine whether the device is actually faulty and generates a fault detection result. When the fault detection result indicates that the target device is faulty, an early warning message is automatically generated. The early warning message contains key abnormal data, which may be the direct cause of the fault or an indicator that strongly suggests an abnormal device status. The early warning message can list in detail the specific type, value, and measurement time of the abnormal data, the possible location of the fault, the fault type, and preliminary suggestions. For example, if the fault detection result indicates that the stress of the tower steel exceeds the limit or the overall stability is insufficient, the early warning message includes the steel with excessive stress and proposes a maintenance strategy for the stress limit.

[0049] Based on the abnormal data in the early warning information, operations and maintenance personnel can compare it with a pre-set database of fault maintenance strategies to find a maintenance strategy that matches the current fault type. This can include emergency shutdown, remote restart, issuing a repair work order, or deploying backup equipment. After determining the appropriate maintenance strategy, it is immediately executed to prevent further deterioration of the fault or to restore normal operation as quickly as possible. If the fault detection results indicate that the target device is not faulty, the next step is to obtain monitoring data from the target device to maintain real-time monitoring of the device's operating status.

[0050] This embodiment quickly responds to equipment failures by issuing early warning information for fault detection results. Through continuous monitoring and generation of early warning information, it achieves comprehensive understanding of the equipment operating status and preventive maintenance, effectively improving operation and maintenance efficiency and the stability of the tower network.

[0051] The fault detection results are regularly matched with the actual operating status of the target device to ensure the accuracy of the fault detection results. Optionally, in the fault detection method for the device provided in the embodiment of the present application, the method also includes: collecting all fault detection results of the target device within the preset period at intervals of a preset period; determining the number of fault detection results within the preset period that are faults, and obtaining the number of fault results; calculating the ratio of the number of fault results to the number of all fault detection results, and obtaining the failure rate; when the failure rate is greater than or equal to the failure rate threshold, determining whether the fault detection result matches the actual operating status of the target device; when the fault detection result matches the actual operating status of the target device, issuing a prompt message, wherein the prompt message is used to prompt the operation and maintenance personnel that there is an abnormality in the target device.

[0052] In some examples, a preset period is set, such as every 24 hours, every week, or every month, for summarizing and analyzing fault detection results. Within each preset period, all fault detection results are automatically extracted from the monitoring data, including detection records of normal and abnormal states. The number of fault detection results with faults within the preset period is counted, that is, the number of fault results. The ratio of the number of fault results to the number of all fault detection results within the period is calculated to obtain the failure rate. A failure rate threshold is set based on historical data and equipment operation conditions. The calculated failure rate is compared with the preset failure rate threshold. If the failure rate is greater than or equal to the failure rate threshold, the next state matching process is entered.

[0053] If the failure rate exceeds the threshold, the system further compares the fault detection results with the actual operating status of the target device to confirm whether they match the actual device status. If the high frequency of fault detection results corresponds to actual abnormal conditions of the device, such as sustained increases in structural strain or abnormal temperature fluctuations, the fault detection results are confirmed to be accurate. If the fault detection results match the actual operating status, a prompt message is generated to notify maintenance personnel that the device is abnormal and requires inspection or repair. The prompt message can include detailed information such as the device number, abnormal time period, specific fault type (such as structural instability, abnormal temperature control, etc.), and recommended maintenance strategies, allowing maintenance personnel to quickly locate the problem and take action.

[0054] This embodiment reduces invalid alarms and avoids interference with operation and maintenance personnel through such a periodic evaluation and status matching process, ensuring that accurate prompt information is issued in a timely manner when the equipment really needs attention, thereby effectively improving the operation and maintenance efficiency and safety of the equipment.

[0055] If the fault detection result does not match the actual operating status of the target device, the second target model needs to be updated in a timely manner. Optionally, in the fault detection method for the device provided in the embodiment of the present application, after determining whether the fault detection result matches the actual operating status of the target device, the method also includes: when the fault detection result does not match the actual operating status of the target device, determining the target fault detection result of the detection error; obtaining the monitoring features corresponding to the target fault detection result, and determining the monitoring features and the actual operating status as a newly added second training sample; combining the newly added second training sample with multiple groups of second training samples to obtain updated multiple groups of second training samples; training the long short-term memory network based on the updated multiple groups of second training samples to obtain an updated second target model.

[0056] In some cases, after receiving an early warning message, the operation and maintenance personnel or system need to confirm the actual operating status of the target device through on-site inspection or other verification methods to determine whether the fault detection results are accurate. If the fault detection result indicates that a fault exists, but the device is actually operating normally, or if the fault detection result indicates that the device is operating normally, but the device is actually faulty, it means that a detection error has occurred. For each instance of a detection error, the difference between the fault detection result and the actual operating status is recorded in detail, including information such as monitoring time, monitoring features, and device status. The corresponding monitoring features are extracted from the instance of the detection error, and these monitoring features are associated with the actual operating status (faulty / no fault) to form a new second training sample.

[0057] The newly added second training samples are combined with the existing sets of second training samples to generate updated sets of second training samples, ensuring that the training dataset reflects the latest equipment operating status and false alarms. Based on these updated sets of second training samples, the long-short-term memory network is retrained to adjust model parameters and optimize the model's ability to identify and predict monitoring features. After training is complete, the updated model replaces the original second target model, ensuring the system can more accurately detect and predict equipment failures.

[0058] This embodiment continuously adapts to the operating characteristics of the equipment and environmental changes through self-learning and self-correction, improves the accuracy of fault prediction, and provides more reliable and effective fault warnings for operation and maintenance personnel, thereby reducing equipment maintenance costs and improving overall operation and maintenance efficiency.

[0059] According to another embodiment of the present application, a device fault monitoring system is provided. Figure 2 Schematic diagram of a device fault monitoring system according to an embodiment of the present application. Figure 2 As shown, the system includes: a data acquisition device, a data communication device, a data analysis device, an early warning device and a data storage device.

[0060] Specifically, the data acquisition device includes multiple data acquisition modules for collecting multi-source data from the tower. This multi-source data includes, but is not limited to, strain information, temperature information, wind speed information, and verticality information. Strain and verticality information are used to determine the tower's overall stability, while temperature and wind speed information are used to determine the tower's real-time status. A data communication device receives this multi-source data from the tower, converts it into a new format, and transmits it to the tower station. This data communication device transmits this multi-source data via wireless communication.

[0061] The data analysis device is used to analyze and calculate the multi-source data information transmitted to the tower station. Based on data preprocessing, feature extraction and classification technologies, the collected signals are denoised by using wavelet transform, retaining valid data and eliminating high-frequency noise. Feature extraction and classification are performed through support vector machines to assess whether there are faults or structural abnormalities in the tower equipment. Specifically, the classification boundaries of the feature data are calculated to obtain the real-time status information of the tower. The data analysis device first denoises the collected signals through wavelet transform. After the signal undergoes wavelet transform and removes high-frequency noise, the key valid information is retained. The specific formula is:

[0062]

[0063] in, is the wavelet transform of the original signal, threshold is the set noise threshold, and the inverse wavelet transform ( ) to obtain a denoised signal. The denoising process can significantly improve the accuracy of subsequent data analysis. A support vector machine is used to extract and classify features from the processed data to assess whether the tower equipment is faulty. The support vector machine algorithm formula is as follows:

[0064]

[0065] Among them, K( , x) is a kernel function (such as a Gaussian kernel function), is the weight of the support vector, is the class label, and b is the deviation term. Through the support vector machine, the system can accurately distinguish normal and abnormal data and determine the operating status of the equipment.

[0066] The early warning device is used to detect abnormalities in the real-time status information of the tower. It can effectively capture long-term dependencies in time series and identify potential fault signals in real time. The fault warning is triggered by the abnormal detection results of LSTM; the early warning device is connected to the data analysis module, and the data analysis module transmits the obtained abnormal data signal to the early warning device. When an abnormal signal is detected, such as strain exceeding the limit or wind speed being too high, the system will perform anomaly detection through the long short-term memory network. LSTM can process time series data and capture the long-term dependencies therein. On this basis, the system can detect possible abnormal signals in real time. The loss function formula of LSTM is as follows:

[0067]

[0068] in, is the actual value, is the predicted value, and T is the length of the time series. If the model's prediction error exceeds a preset threshold, an early warning mechanism is triggered. The LSTM network helps the system identify patterns and trends in time series data, effectively detecting abnormal signals before a failure occurs. The early warning device issues a timely warning, preventing the problem from worsening.

[0069] The data storage device is used to store the multi-source data information transmitted to the tower station and the data information analyzed by the data analysis device.

[0070] This embodiment provides an equipment fault monitoring system that uses algorithms such as support vector machines (SVM) and long short-term memory networks (LSTM) to achieve real-time monitoring, fault detection, and intelligent early warning of tower station equipment, thereby improving monitoring efficiency and accuracy.

[0071] According to another embodiment of the present application, a device fault monitoring method applied to a device fault monitoring system is also provided. Figure 3 Schematic diagram of a device fault monitoring method according to an embodiment of the present application. Figure 3 As shown, the method includes:

[0072] The data acquisition device is used to collect multi-source data information of the tower, including strain information, temperature information, wind speed information and verticality information; the multi-source data information collected by the data acquisition device is transmitted to the tower station through the data communication device; the received multi-source data information is analyzed and calculated by the data analysis device, and compared with normal data, and a signal is output to determine whether the data is abnormal; the early warning device is used to determine whether the received data is abnormal, and if the data is an abnormal signal, an early warning message is issued; the multi-source data information collected by the data acquisition device and the data of the data analysis device are stored through the data storage device.

[0073] This embodiment uses the equipment fault monitoring method to realize real-time monitoring of the operating status of the tower station and its overall stability. When an abnormality occurs in the real-time monitoring data information, an early warning message can be issued in time through the data early warning device, and the tower fault area and the cause of the fault can be provided in time, so that maintenance and other work can be carried out quickly, reducing the difficulty of inspection, maintenance and emergency repairs.

[0074] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0075] The present application also provides a device fault detection apparatus. It should be noted that the device fault detection apparatus of the present application embodiment can be used to execute the device fault detection method provided in the present application embodiment. The following describes the device fault detection apparatus provided in the present application embodiment.

[0076] Figure 4 Schematic diagram of a fault detection device for a device according to an embodiment of the present application. Figure 4 As shown, the device includes:

[0077] An acquisition unit 401 is configured to acquire target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device;

[0078] A first input unit 402 is configured to input target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple sets of first training samples, each set of first training samples including historical monitoring data and historical monitoring features, wherein the monitoring features include normal data and abnormal data;

[0079] The second input unit 403 is used to input the monitoring features into the second target model when the monitoring features represent an abnormality in the monitoring data, so as to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, and each group of second training samples includes historical monitoring features and historical fault detection results.

[0080] The fault detection device of the equipment provided in the embodiment of the present application obtains target monitoring data of the target equipment through an acquisition unit 401, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed and verticality, and the verticality is used to characterize the stability of the target equipment; a first input unit 402 inputs the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; a second input unit 403 inputs the monitoring features into a second target model when the monitoring features characterize the presence of abnormalities in the monitoring data, to obtain a fault detection result of the target equipment, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples includes historical monitoring features and historical fault detection results, thereby solving the problem of low fault detection efficiency of equipment in the related art. By acquiring multi-dimensional monitoring data, the monitoring features of the monitoring data are extracted based on the first target model, and then the fault detection results are determined by the second target model and the monitoring features, thereby achieving the effect of improving the fault detection efficiency of the equipment.

[0081] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: a comparison unit, used to obtain multiple historical monitoring data collected from sensors on the target device, and compare each historical monitoring data with preset standard data to obtain a comparison result, wherein the standard data is the monitoring data of the target device when there is no fault; a first determination unit, used to determine the historical monitoring characteristics of each historical monitoring data based on the comparison result, and determine each historical monitoring data and its corresponding historical monitoring characteristics as a group of first training samples to obtain multiple groups of first training samples; a first training unit, used to train a support vector machine through multiple groups of first training samples to obtain a first target model.

[0082] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: a second determination unit, used to obtain a set of historical monitoring features within a preset period, and determine the predicted monitoring features at the target moment based on the set of historical monitoring features; a third determination unit, used to determine the monitoring features at the target moment, calculate the difference between the monitoring features and the predicted monitoring features, and determine that the historical fault monitoring result of the monitoring features at the target moment is that a fault exists when the difference is greater than or equal to a preset threshold; a fourth determination unit, used to determine that the historical fault monitoring result of the monitoring features at the target moment is that no fault exists when the difference is less than a preset threshold; a fifth determination unit, used to determine the historical monitoring features at each moment and the historical fault detection results of the historical monitoring features as a set of second training samples to obtain multiple groups of second training samples; a second training unit, used to train the long short-term memory network through multiple groups of second training samples to obtain a second target model.

[0083] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: a first processing unit, used to perform wavelet transform processing on the target monitoring data to obtain data to be denoised; a second processing unit, used to determine a preset noise threshold, calculate the difference between the data to be denoised and the noise threshold, and perform wavelet inverse transform processing on the difference to obtain the target monitoring data after denoising.

[0084] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: an early warning unit, which is used to issue an early warning message when the fault detection result indicates that the target device has a fault, wherein the early warning message includes abnormal data in the monitoring data; a first execution unit, which is used to determine a preset fault maintenance strategy based on the abnormal data in the early warning message and execute the fault maintenance strategy; and a second execution unit, which is used to continue to execute the step of obtaining the monitoring data of the target device when the fault detection result indicates that the target device does not have a fault.

[0085] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: a collection unit, which is used to collect all fault detection results of the target device within a preset period every preset period; a sixth determination unit, which is used to determine the number of fault detection results within the preset period that are faults, and obtain the number of fault results; a calculation unit, which is used to calculate the ratio of the number of fault results to the number of all fault detection results, and obtain the failure rate; a judgment unit, which is used to judge whether the fault detection result matches the actual operating status of the target device when the failure rate is greater than or equal to the failure rate threshold; and a prompt unit, which is used to issue a prompt message when the fault detection result matches the actual operating status of the target device, wherein the prompt message is used to prompt the operation and maintenance personnel that there is an abnormality in the target device.

[0086] Optionally, in the fault detection device of the equipment provided in the embodiment of the present application, the device also includes: a seventh determination unit, used to determine the target fault detection result of the detection failure when the fault detection result does not match the actual operating status of the target device; an eighth determination unit, used to obtain the monitoring characteristics corresponding to the target fault detection result, and determine the monitoring characteristics and the actual operating status as newly added second training samples; a combination unit, used to combine the newly added second training samples with multiple groups of second training samples to obtain updated multiple groups of second training samples; a third training unit, used to train the long short-term memory network based on the updated multiple groups of second training samples to obtain an updated second target model.

[0087] The fault detection device of the equipment includes a processor and a memory. The acquisition unit 401, the first input unit 402 and the second input unit 403 are all stored in the memory as program units, and the processor executes the program units stored in the memory to realize corresponding functions.

[0088] The processor contains a kernel, which retrieves the corresponding program unit from the memory. You can set one or more kernels, and adjust the kernel parameters to improve the device's fault detection efficiency.

[0089] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0090] An embodiment of the present invention provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a method for detecting a fault of a device is implemented.

[0091] An embodiment of the present invention provides a processor, which is used to run a program, wherein the program executes a device fault detection method when running.

[0092] Figure 5 Schematic diagram of an electronic device according to an embodiment of the present application. Figure 5 As shown, electronic device 501 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: obtaining target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed, and verticality, wherein verticality is used to characterize the stability of the target device; inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; when the monitoring features indicate that the monitoring data is abnormal, inputting the monitoring features into a second target model to obtain a fault detection result for the target device, wherein the second target model is trained by multiple groups of second training samples, each group of second training samples includes historical monitoring features and historical fault detection results. The device in this article can be a server, PC, PAD, mobile phone, etc.

[0093] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialized program having the following method steps: obtaining target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed and verticality, and the verticality is used to characterize the stability of the target device; inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; when the monitoring features characterize the existence of abnormalities in the monitoring data, inputting the monitoring features into a second target model to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, and each group of second training samples includes historical monitoring features and historical fault detection results.

[0094] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0098] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0099] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0100] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0101] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for detecting a fault of an equipment, characterized in that: include: Acquiring target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device; Inputting the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; When the monitoring feature indicates that the monitoring data is abnormal, the monitoring feature is input into a second target model to obtain a fault detection result of the target device, wherein the second target model is trained by multiple groups of second training samples, and each group of second training samples includes historical monitoring features and historical fault detection results.

2. The method according to claim 1, characterized in that The first target model is obtained by: Acquire multiple historical monitoring data collected from sensors on the target device, and compare each historical monitoring data with preset standard data to obtain a comparison result, wherein the standard data is monitoring data of the target device when no fault occurs; Determine a historical monitoring feature of each historical monitoring data based on the comparison result, determine each historical monitoring data and its corresponding historical monitoring feature as a group of first training samples, and obtain multiple groups of first training samples; The support vector machine is trained using the multiple groups of first training samples to obtain the first target model.

3. The method according to claim 1, characterized in that The second target model is obtained by: Obtaining a set of historical monitoring features within a preset period, and determining predicted monitoring features at a target moment based on the set of historical monitoring features; determining a monitoring feature at the target time, calculating a difference between the monitoring feature and the predicted monitoring feature, and determining that a historical fault monitoring result of the monitoring feature at the target time is a fault if the difference is greater than or equal to a preset threshold; When the difference is less than the preset threshold, determining that the historical fault monitoring result of the monitoring feature at the target moment is that no fault exists; Determining the historical monitoring features at each moment and the historical fault detection results of the historical monitoring features as a set of second training samples, thereby obtaining multiple sets of second training samples; The long short-term memory network is trained using the multiple groups of second training samples to obtain the second target model.

4. The method according to claim 1, wherein After acquiring the target monitoring data of the target device, the method further includes: Performing wavelet transform processing on the target monitoring data to obtain data to be denoised; A preset noise threshold is determined, a difference between the data to be denoised and the noise threshold is calculated, and an inverse wavelet transform is performed on the difference to obtain the target monitoring data after denoising.

5. The method according to claim 1, wherein After obtaining the fault detection result of the target device, the method further includes: When the fault detection result indicates that the target device has a fault, issuing a warning message, wherein the warning message includes abnormal data in the monitoring data; Determining a preset fault maintenance strategy based on the abnormal data in the warning information, and executing the fault maintenance strategy; When the fault detection result indicates that the target device does not have a fault, the step of obtaining monitoring data of the target device is continued.

6. The method according to claim 1, characterized in that The method further comprises: Collect all fault detection results of the target device within the preset period at every preset period; Determine the number of fault detection results indicating that a fault exists within the preset period, and obtain the number of fault results; Calculating the ratio of the number of the fault results to the number of all fault detection results to obtain a fault rate; If the failure rate is greater than or equal to a failure rate threshold, determining whether the failure detection result matches the actual operating state of the target device; In the case where the fault detection result matches the actual operating status of the target device, a prompt message is issued, wherein the prompt message is used to prompt the operation and maintenance personnel that the target device has an abnormality.

7. The method according to claim 6, characterized in that After determining whether the fault detection result matches the actual operating state of the target device, the method further includes: In the case where the fault detection result does not match the actual operating state of the target device, determining a target fault detection result of a detection error; Obtaining a monitoring feature corresponding to the target fault detection result, and determining the monitoring feature and the actual operating state as a newly added second training sample; Combining the newly added second training samples with the multiple groups of second training samples to obtain multiple updated groups of second training samples; A long short-term memory network is trained based on the updated multiple groups of second training samples to obtain an updated second target model.

8. A device for detecting a fault of an equipment, characterized in that: include: an acquisition unit, configured to acquire target monitoring data of a target device, wherein the target monitoring data includes at least one of the following: strain data, temperature, wind speed, and verticality, wherein the verticality is used to characterize the stability of the target device; a first input unit, configured to input the target monitoring data into a first target model to obtain monitoring features, wherein the first target model is trained by multiple groups of first training samples, each group of first training samples includes historical monitoring data and historical monitoring features, and the monitoring features include normal data and abnormal data; The second input unit is used to input the monitoring feature into the second target model to obtain the fault detection result of the target device when the monitoring feature represents that the monitoring data is abnormal, wherein the second target model is trained by multiple groups of second training samples, and each group of second training samples includes historical monitoring features and historical fault detection results.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the fault detection method of the device according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: It includes one or more processors and a memory, the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the fault detection method of the device according to any one of claims 1 to 7.

Images