Multi-dimensional fault diagnosis method and system for combined electrical apparatus based on electromechanical coordination monitoring

Abstract

This invention discloses a multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring, which includes the following steps: collecting electrical data, mechanical data, and environmental data of the combined electrical appliances; preprocessing the collected electrical data, mechanical data, and environmental data; extracting electrical features, mechanical features, and environmental features from the electrical data, mechanical data, and environmental data respectively, and fusing the electrical features, mechanical features, and environmental features to obtain a multi-dimensional feature vector; and inputting the multi-dimensional feature vector into a trained neural network model to output the fault category.

Description

Technical Field

[0001] This invention relates to a fault diagnosis method and system, and more particularly to a fault diagnosis method and system for combined electrical appliances. Background Technology

[0002] With the development of modern power systems towards intelligence and automation, the operating environment and workload of power equipment are becoming increasingly complex. In particular, combined electrical equipment (such as circuit breakers, disconnectors, load switches, and grounding switches) plays a crucial role in the protection, switching, and isolation functions of the power system. The reliability and safety of these devices directly affect the stability of the power system and the continuity of power supply. However, combined electrical equipment may be affected by various factors such as electrical overload, mechanical fatigue, and environmental changes during long-term operation, leading to equipment failures. In severe cases, this can even cause equipment damage, system outages, or larger-scale power accidents. Therefore, accurate fault diagnosis and condition monitoring technologies are of great significance for improving the operational safety of power equipment, extending equipment lifespan, and optimizing equipment operation and maintenance.

[0003] Traditional methods for diagnosing faults in combined electrical appliances mainly rely on single electrical monitoring methods, such as current, voltage, or temperature monitoring. While these methods can detect some basic information about faults, the multi-dimensional, complex, and concealed nature of electrical equipment faults means that a single electrical monitoring method cannot fully reflect the true operating status of the equipment, resulting in low accuracy and sensitivity in fault detection.

[0004] In recent years, with the continuous development of electromechanical collaborative monitoring technology, comprehensive data analysis combining electrical performance, mechanical performance, and environmental conditions has become an important means to improve the accuracy of fault diagnosis. Electromechanical collaborative monitoring technology achieves comprehensive perception of equipment operating status by simultaneously acquiring electrical signals (such as current, voltage, and power), mechanical signals (such as vibration, temperature, and pressure), and environmental signals (such as humidity and air temperature). By performing multi-dimensional fusion and feature extraction on these data, fault signals that traditional electrical monitoring methods cannot capture can be identified, such as fatigue damage to mechanical components and electrical faults caused by insulation aging. This multi-dimensional and comprehensive monitoring approach can effectively improve the sensitivity, accuracy, and early warning capabilities of fault diagnosis.

[0005] However, current fault diagnosis technologies for combined electrical appliances based on electromechanical collaborative monitoring still face the following challenges:

[0006] Data fusion and feature extraction are challenging: The large amount of multi-dimensional data generated by electromechanical collaborative monitoring varies greatly in terms of data type, scale, and temporal characteristics. Effectively fusing these heterogeneous data and extracting useful fault features from them is a complex challenge.

[0007] Fault mode complexity: Combined electrical appliances exhibit diverse fault modes, including electrical faults, mechanical faults, and environmental factors. Furthermore, different fault modes often interact, making it difficult to distinguish fault characteristics. Therefore, traditional fault diagnosis methods based on single signals are ill-suited to such complex fault scenarios.

[0008] Reliability and Real-Time Performance of Diagnostic Algorithms: Most existing fault diagnosis algorithms are based on traditional models or rules of thumb, which often suffer from low diagnostic accuracy and poor real-time performance when dealing with complex equipment fault characteristics. Especially in real-time operation and maintenance environments, ensuring that fault diagnosis systems can accurately and quickly respond to fault information and provide effective early warnings remains a pressing problem to be solved.

[0009] System integration and implementation challenges: Electromechanical collaborative monitoring involves the integration of multiple sensors and monitoring devices, resulting in high system complexity and implementation costs. Furthermore, different types of electrical equipment differ significantly in design and use; therefore, achieving a universal fault diagnosis model and technology across various application scenarios presents another practical challenge.

[0010] Therefore, existing technologies have not yet provided an efficient fault diagnosis method that can fully integrate multi-dimensional data and effectively identify and diagnose complex fault modes. Summary of the Invention

[0011] One of the objectives of this invention is to provide a multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring, which can achieve high-precision, high-real-time, and low-cost fault diagnosis through the combination of signal processing, data fusion, and fault prediction algorithms.

[0012] In accordance with the aforementioned objective, this invention proposes a multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring, comprising the following steps:

[0013] Collect electrical, mechanical, and environmental data from the combined electrical equipment;

[0014] The collected electrical, mechanical, and environmental data are preprocessed.

[0015] Electrical, mechanical, and environmental features are extracted from the electrical, mechanical, and environmental data respectively, and the electrical, mechanical, and environmental features are fused to obtain a multidimensional feature vector.

[0016] Multidimensional feature vectors are input into a trained neural network model to output fault categories.

[0017] The multi-dimensional fault diagnosis method for combined electrical appliances described in this invention processes the raw data through a preprocessing step, and then performs time-domain and frequency-domain analysis through feature extraction to extract representative fault features from the raw data, such as the frequency features of current waveforms, temperature anomalies, and vibration signals, as the basis for subsequent fault diagnosis and identification.

[0018] Furthermore, due to the complexity and diversity of failure modes in combined electrical appliances, traditional single-data analysis methods are insufficient to accurately reflect the equipment's status. This invention integrates the features of extracted electrical, mechanical, and environmental data from multiple dimensions, combining data from different sources to form a multi-dimensional feature vector that comprehensively reflects the equipment's status.

[0019] This invention combines machine learning, deep learning, and pattern recognition technologies, using neural networks to analyze and process multi-dimensional data. By training fault sample data, a fault identification model is constructed to identify the fault types of combined electrical appliances (such as mechanical faults, electrical faults, and environmental faults). Furthermore, this invention can also employ multi-task learning and transfer learning methods to improve the model's adaptability to different types of equipment and operating environments.

[0020] Furthermore, in the multi-dimensional fault diagnosis method for combined electrical appliances described in this invention, the preprocessing includes at least one of noise reduction processing, missing value filling, standardization processing, and normalization processing.

[0021] Furthermore, in the multi-dimensional fault diagnosis method for combined electrical appliances described in this invention, wavelet transform algorithm is used for denoising during the denoising process.

[0022] Furthermore, in the multi-dimensional fault diagnosis method for combined electrical appliances described in this invention, weighted average and principal component analysis methods are used to fuse electrical characteristics, mechanical characteristics, and environmental characteristics.

[0023] Furthermore, in the multi-dimensional fault diagnosis method for combined electrical appliances described in this invention, the neural network model includes a support vector machine model and / or a convolutional neural network.

[0024] Another objective of this invention is to provide a multi-dimensional fault diagnosis system for combined electrical appliances based on electromechanical collaborative monitoring, which can achieve high-precision, high-real-time, and low-cost fault diagnosis through the combination of signal processing, data fusion, and fault prediction algorithms.

[0025] To achieve the above objectives, the present invention also provides a multi-dimensional fault diagnosis method system for combined electrical appliances based on electromechanical collaborative monitoring, comprising:

[0026] The data acquisition module collects electrical, mechanical, and environmental data from the combined electrical appliances.

[0027] The preprocessing module preprocesses the collected electrical, mechanical, and environmental data.

[0028] The feature extraction and fusion module extracts electrical, mechanical, and environmental features from the electrical, mechanical, and environmental data, respectively, and fuses these features to obtain a multi-dimensional feature vector.

[0029] The fault identification module, based on the input 3D feature vector, uses a trained neural network model to output the fault category.

[0030] Furthermore, in the multi-dimensional fault diagnosis system for combined electrical appliances described in this invention, the preprocessing module performs at least one of the following: noise reduction, missing value filling, standardization, and normalization.

[0031] Furthermore, in the multi-dimensional fault diagnosis system for combined electrical appliances described in this invention, when the preprocessing module performs denoising, a wavelet transform algorithm is used for denoising.

[0032] Furthermore, in the multi-dimensional fault diagnosis system for combined electrical appliances described in this invention, the feature extraction and fusion module uses weighted average and principal component analysis methods to fuse electrical features, mechanical features, and environmental features.

[0033] Furthermore, in the multi-dimensional fault diagnosis system for combined electrical appliances described in this invention, the neural network model includes a support vector machine model and / or a convolutional neural network.

[0034] The multi-dimensional fault diagnosis method and system for combined electrical appliances described in this invention have the following advantages and beneficial effects:

[0035] This invention enables multi-dimensional monitoring and data fusion, overcoming the limitations of traditional single-monitoring technologies by integrating electrical, mechanical, and environmental data. Multi-dimensional data fusion provides more comprehensive and accurate equipment operating status information, significantly improving the accuracy of fault diagnosis and early warning capabilities.

[0036] This invention employs intelligent algorithms based on machine learning and deep learning, enabling continuous optimization of the fault diagnosis model based on historical and real-time data, and possessing self-learning and adaptive capabilities. This system is not only applicable to a single type of combined electrical equipment, but can also be adjusted and optimized according to the specific needs of different power systems, demonstrating strong versatility.

[0037] This invention, by combining multiple fault diagnosis algorithms, can accurately identify fault types, reducing the problem of large diagnostic errors in traditional methods. Especially in complex fault scenarios, it can significantly improve the accuracy and response speed of fault diagnosis.

[0038] This invention can quickly respond to equipment fault signals during real-time monitoring and process them in real time through intelligent algorithms, providing maintenance personnel with immediate fault warnings. Rapid diagnosis and location of faults after they occur can greatly reduce downtime and improve the reliability of the power system, thus possessing both real-time performance and high efficiency.

[0039] This invention, through an advanced fault prediction model, can not only diagnose faults but also predict the remaining lifespan and potential fault risks of equipment, providing decision support for intelligent operation and maintenance and resource optimization of equipment.

[0040] This invention is applicable to fields such as smart grids, distribution networks, industrial automation, and power transmission and transformation equipment. It is particularly suitable for the operation monitoring of combined electrical appliances (such as circuit breakers, disconnectors, grounding switches, load switches, etc.) in high-voltage, ultra-high-voltage, and medium- and low-voltage power grids. It can combine electrical performance (such as current, voltage, power, etc.), mechanical performance (such as vibration, temperature, pressure, etc.), environmental conditions (such as humidity, air temperature, etc.) and fault mechanism modeling to achieve multi-dimensional monitoring and fault diagnosis of combined electrical appliances in complex working environments. Attached Figure Description

[0041] Figure 1 The flowchart illustrates the steps of one embodiment of the multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring according to the present invention.

[0042] Figure 2 The diagram shows the architecture of the multi-dimensional fault diagnosis system for combined electrical appliances based on electromechanical collaborative monitoring according to the present invention in one embodiment. Detailed Implementation

[0043] The following will further explain and illustrate the multi-dimensional fault diagnosis method and system for combined electrical appliances based on electromechanical collaborative monitoring according to the present invention, with reference to the accompanying drawings and specific embodiments. However, this explanation and illustration do not constitute an undue limitation on the technical solution of the present invention.

[0044] Figure 1 The flowchart illustrates the steps of one embodiment of the multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring according to the present invention.

[0045] like Figure 1 As shown, in one embodiment, the multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring may include the following steps:

[0046] Step 100: Collect electrical, mechanical, and environmental data of the combined electrical equipment.

[0047] In some specific implementations, the electrical data comes from data sensed by electrical sensors. In some more specific implementations, the electrical sensors may include at least one of a current sensor, a voltage sensor, a power sensor, and a frequency sensor to monitor various electrical parameters of the combined electrical equipment in real time.

[0048] In some specific embodiments, the mechanical data comes from data sensed by mechanical sensors. In some more specific embodiments, the mechanical sensors may include at least one of vibration sensors, pressure sensors, and displacement sensors, for monitoring the condition of the mechanical components of the combined electrical appliance, such as wear, thermal expansion, and vibration.

[0049] In some specific implementations, the environmental data comes from data sensed by environmental sensors. In some more specific implementations, the environmental sensors may include at least one of a humidity sensor, a temperature sensor, and a barometric pressure sensor, used to monitor the temperature, humidity, and barometric pressure of the environment in which the combined electrical appliances are located.

[0050] Step 200: Preprocess the collected electrical, mechanical and environmental data.

[0051] In some specific implementations, preprocessing may include at least one of denoising, missing value imputation, standardization, and normalization.

[0052] In some specific implementations, wavelet transform algorithms can be used for denoising.

[0053] In addition, in some specific implementations, noise reduction processing can utilize low-pass filtering and Kalman filtering methods to remove noise from sensor signals.

[0054] In some specific implementations, missing value filling can be achieved by using interpolation methods (such as linear interpolation or spline interpolation) to fill missing values ​​caused by sensor failure or unstable data transmission.

[0055] In some specific implementations, standardizing and normalizing the data output from different sensors is beneficial for subsequent feature extraction and data fusion.

[0056] Step 300: Extract the electrical, mechanical, and environmental features from the electrical, mechanical, and environmental data respectively, and fuse the electrical, mechanical, and environmental features to obtain a multidimensional feature vector.

[0057] In some specific implementations, when extracting electrical features, frequency domain features can be extracted by Fourier transform based on current and voltage waveform analysis to identify abnormal fluctuations in electrical parameters, which can be used to identify typical electrical faults such as overload and short circuit.

[0058] In some specific implementations, vibration analysis (such as time-domain and frequency-domain analysis) and temperature change analysis can be used to identify mechanical component faults, such as wear and overheating, when extracting mechanical features. In some more specific implementations, wavelet transform and empirical mode decomposition (EMD) methods can be used to extract features from time-domain signals.

[0059] In some specific implementations, after multi-dimensional feature extraction, weighted averaging and principal component analysis (PCA) methods can be used to fuse electrical, mechanical, and environmental features. This step integrates multi-source data into a unified multi-dimensional feature space, improving the accuracy and robustness of fault diagnosis.

[0060] In some more specific implementations, the extracted features may include skewness S k Steepness Ku, discharge factor Q, cross-correlation coefficient CC, and phase asymmetry ψ.

[0061] Among them, skewness S k This is used to describe the shape differences in PRPD graphs, indicating the degree of skewness. If the skewness is 0, the graph is symmetrical; if the skewness is greater than 0, the graph is positively skewed, with the skew direction to the left of the arithmetic mean; if the skewness is less than 0, the graph is negatively skewed, with the skew direction to the right of the arithmetic mean. The specific calculation formula is:

[0062]

[0063] In the formula, N represents the number of phase windows in the PRPD pattern within half a power frequency cycle; This represents the phase of the i-th phase window in the PRPD diagram; Indicates the phase width; parameters μ, p i And σ respectively represent the... Let y be the mean, probability density, and variance of partial discharge defects occurring within the i-th phase window of the PRPD spectrum when y is a variable. i The vertical axis represents the two-dimensional graph.

[0064] Steepness K u Kurtosis describes the degree of ridges in the PRPD map shape compared to a normal distribution. A kurtosis of 0 indicates that the map shape distribution is consistent with a normal distribution; a positive kurtosis indicates that the map outline is sharper and steeper than a normal distribution; and a negative kurtosis indicates that the map outline is flatter than a normal distribution. The formula for calculating kurtosis is:

[0065]

[0066] The discharge factor Q reflects the difference in discharge quantity in the PRPD spectrum during the positive and negative power frequency half-cycles. The calculation formula is as follows:

[0067]

[0068] In the formula, q s + and q s － n represents the total discharge amount during the positive and negative half-cycles of the phase, respectively. s + and n s － These represent the total number of discharges during the positive and negative half-cycles of the phase, respectively.

[0069] The cross-correlation coefficient (CC) describes the similarity of the PRPD (Pressure Point Diagram) profiles across the positive and negative power frequency half-cycles. A CC closer to 0 indicates a greater difference in the profiles across the positive and negative half-cycles; a CC closer to 1 indicates greater similarity. The formula for calculating CC is:

[0070]

[0071] In the formula, q i + and q i － These represent the average discharge quantities of the positive and negative half-cycles within the i-th phase window of the PRPD spectrum, respectively.

[0072] The phase asymmetry ψ represents the difference between the starting phases of the positive and negative half-cycles of discharge in the statistical spectrum, and its calculation formula is as follows:

[0073]

[0074] In the formula, and These represent the initial discharge phases of the PRPD spectrum in the positive and negative half-power frequency cycles, respectively.

[0075] Step 400: Input the multidimensional feature vector into the trained neural network model to output the fault category.

[0076] In some specific implementations, the neural network module includes Support Vector Machine (SVM), Random Forest (RF), Convolutional Neural Network (CNN), and Long Short-Term Memory Network (LSTM). Trained using sample data sources calibrated for fault types, it can accurately identify patterns of different faults, including determining the severity and type of the fault.

[0077] In some embodiments, the method described in this invention further includes fault prediction, for example, it can use regression analysis and time series analysis to predict the remaining life of the combined electrical equipment. By combining historical fault data of the combined electrical equipment and environmental changes, a fault prediction model is established, and the occurrence of potential faults can be predicted in a timely manner through real-time monitoring of the equipment's operating status.

[0078] In another embodiment of the present invention, a multi-dimensional fault diagnosis method system for combined electrical appliances based on electromechanical collaborative monitoring is also provided.

[0079] Figure 2 The diagram shows the architecture of the multi-dimensional fault diagnosis system for combined electrical appliances based on electromechanical collaborative monitoring according to the present invention in one embodiment.

[0080] like Figure 2 As shown, in some embodiments, the multi-dimensional fault diagnosis system for the combined electrical appliances may include:

[0081] The data acquisition module 502 collects electrical, mechanical, and environmental data from the combined electrical appliances.

[0082] The preprocessing module 504 preprocesses the collected electrical data, mechanical data, and environmental data.

[0083] The feature extraction and fusion module 506 extracts electrical features, mechanical features, and environmental features from the electrical data, mechanical data, and environmental data, respectively, and fuses the electrical features, mechanical features, and environmental features to obtain a multi-dimensional feature vector;

[0084] The fault identification module 508, based on the input 3D feature vector, uses a trained neural network model to output the fault category.

[0085] In some more specific implementations, the preprocessing module 504 performs at least one of the following: noise reduction, missing value filling, standardization, and normalization.

[0086] In some more specific implementations, when the preprocessing module 504 performs denoising, it uses a wavelet transform algorithm for denoising.

[0087] In some more specific implementations, the feature extraction and fusion module 506 uses weighted average and principal component analysis methods to fuse electrical features, mechanical features and environmental features.

[0088] In some more specific implementations, the neural network model includes a support vector machine model and / or a convolutional neural network.

[0089] Furthermore, in some embodiments, the system of the present invention may also include a positioning module, which, once a fault is diagnosed, accurately identifies the location of the fault source through data positioning. For example, in some more specific embodiments, the faulty equipment part is located by combining spatially distributed multi-point sensor data with information on signal strength and frequency response.

[0090] Furthermore, in some embodiments, the system of the present invention also includes an early warning module, which is configured to issue an early warning signal when the system detects that the combined electrical appliances are about to malfunction or have a potential for malfunction, notifying maintenance personnel to conduct inspection and handling. The early warning information may include the type of malfunction, its location, and the possible scope of impact.

[0091] In the system described in this invention, each module can be implemented in any suitable manner, for example, in the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320.

[0092] Furthermore, those skilled in the art will recognize that, besides implementing the controller using purely computer-readable program code, the control module can achieve the same functionality by logically programming the method steps, making it available in the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a control module can be considered a hardware component, and the devices within it used to implement various functions can also be considered structures within that hardware component. Alternatively, the devices used to implement various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0093] Therefore, the multi-dimensional fault diagnosis method and system for combined electrical appliances based on electromechanical collaborative monitoring described in this invention comprehensively collects multi-source data from electrical, mechanical, and environmental sources, and uses data fusion and intelligent algorithms to monitor the operating status of combined electrical appliances in real time, identify equipment fault types, fault locations, and development trends. It can accurately and timely detect multi-dimensional faults in combined electrical appliances, improve the accuracy, real-time performance, and intelligence level of fault diagnosis, and thus effectively improve the safety, reliability, and economy of the power system.

[0094] This invention, through retrospective analysis of historical operating data and continuous monitoring of real-time data, can predict the occurrence time and development trend of equipment failures. Combined with equipment remaining life prediction technology, it can identify potential failure risks in advance and generate fault warning signals. This fault prediction system can provide a scientific basis for power system operation and maintenance decisions, helping maintenance personnel to take timely measures and reduce downtime and maintenance costs.

[0095] This invention can be combined with a spatially distributed sensor network to achieve precise location of faults in combined electrical appliances. Through data fusion and feature matching, it identifies the location and type of the fault and provides accurate fault source information. Accurate fault location helps accelerate the troubleshooting process, shorten downtime, and reduce maintenance costs. Simultaneously, the fault analysis module of this invention can further analyze the causes of the fault, providing a basis for equipment repair and optimization.

[0096] It should be noted that the scope of protection of the prior art in this invention is not limited to the embodiments given in this application. All prior art that does not contradict the solution of this invention, including but not limited to prior patent documents, prior publications, prior public uses, etc., can be included in the scope of protection of this invention.

[0097] Furthermore, the combination of the technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0098] It should also be noted that the embodiments listed above are merely specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments, and similar changes or modifications made thereto are those that can be directly derived or easily conceived by those skilled in the art from the content disclosed in the present invention, and should all fall within the protection scope of the present invention.

Claims

1. A multi-dimensional fault diagnosis method for combined electrical appliances based on electromechanical collaborative monitoring, characterized in that, Including the following steps: Collect electrical, mechanical, and environmental data from the combined electrical equipment; The collected electrical, mechanical, and environmental data are preprocessed. Electrical, mechanical, and environmental features are extracted from the electrical, mechanical, and environmental data respectively, and the electrical, mechanical, and environmental features are fused to obtain a multidimensional feature vector. Multidimensional feature vectors are input into a trained neural network model to output fault categories.

2. The multi-dimensional fault diagnosis method for combined electrical appliances as described in claim 1, characterized in that, The preprocessing includes at least one of the following: denoising, missing value imputation, standardization, and normalization.

3. The multi-dimensional fault diagnosis method for combined electrical appliances as described in claim 2, characterized in that, When performing denoising, wavelet transform algorithm is used.

4. The multi-dimensional fault diagnosis method for combined electrical appliances as described in claim 1, characterized in that, We used weighted average and principal component analysis to fuse electrical, mechanical and environmental characteristics.

5. The multi-dimensional fault diagnosis method for combined electrical appliances as described in claim 1, characterized in that, The neural network model includes a support vector machine model and / or a convolutional neural network.

6. A multi-dimensional fault diagnosis system for combined electrical appliances based on electromechanical collaborative monitoring, characterized in that, include: The data acquisition module collects electrical, mechanical, and environmental data from the combined electrical appliances. The preprocessing module preprocesses the collected electrical, mechanical, and environmental data. The feature extraction and fusion module extracts electrical, mechanical, and environmental features from the electrical, mechanical, and environmental data, respectively, and fuses these features to obtain a multi-dimensional feature vector. The fault identification module, based on the input 3D feature vector, uses a trained neural network model to output the fault category.

7. The multi-dimensional fault diagnosis system for combined electrical appliances as described in claim 6, characterized in that, The preprocessing module performs at least one of the following: noise reduction, missing value filling, standardization, and normalization.

8. The multi-dimensional fault diagnosis system for combined electrical appliances as described in claim 7, characterized in that, When the preprocessing module performs denoising, it uses a wavelet transform algorithm.

9. The multi-dimensional fault diagnosis system for combined electrical appliances as described in claim 6, characterized in that, The feature extraction and fusion module uses weighted average and principal component analysis methods to fuse electrical, mechanical and environmental features.

10. The multi-dimensional fault diagnosis system for combined electrical appliances as described in claim 6, characterized in that, The neural network model includes a support vector machine model and / or a convolutional neural network.

Images