Charging pile fault identification method and device based on electric power fingerprint, and medium

Abstract

The invention discloses a charging pile fault identification method and device based on an electric power fingerprint and a medium, and the method comprises the steps: collecting electric power signal data during the operation of a charging pile, carrying out the preprocessing of the electric power signal data, and carrying out the multi-dimensional feature extraction of the preprocessed electric power signal data; carrying out dimension reduction processing on the multi-dimensional features through a principal component analysis technology, and generating a real-time electric power fingerprint vector used for representing the operation state of the charging pile; obtaining a fault power fingerprint vector corresponding to the historical fault type of the charging pile and a standard power fingerprint vector of the charging pile in a normal state, so as to construct a corresponding fault feature library; and calculating cosine similarity between the real-time electric power fingerprint vector and each electric power fingerprint vector in the fault feature library, and inputting the cosine similarity into a pre-trained deep learning model to determine a fault type corresponding to the real-time electric power fingerprint vector based on the cosine similarity and a model output result.

Description

Technical Field

[0001] The present application relates to the field of fault identification technology, and in particular to a charging pile fault identification method, device and medium based on power fingerprint. Background Art

[0002] With the rapid adoption of new energy vehicles, charging piles, as critical components of electric vehicle energy supply, have a significant impact on grid security and user experience. Current mainstream charging pile fault detection technologies rely on manual inspections and fixed threshold detection mechanisms, which are insufficient to meet the demands of real-time monitoring and precise diagnosis of large-scale charging infrastructure. Traditional approaches typically employ hardware sensor arrays deployed at key nodes within charging piles, detecting faults by collecting a single physical quantity. However, such solutions require integrating a large number of sensors within each device, significantly increasing hardware costs. Furthermore, sensors are susceptible to electromagnetic interference, aging, and other factors, resulting in high false alarm rates. Furthermore, the complex wiring of multiple sensors makes ongoing maintenance challenging, and compatibility with diverse charging pile models is difficult.

[0003] Furthermore, while threshold discrimination technology based on a single electrical parameter can simplify deployment, its reliance on time-domain statistics presents inherent flaws. By ignoring frequency-domain harmonic components and transient waveform mutations, such methods are severely inadequate for identifying multi-factor coupled faults. Furthermore, fixed threshold settings cannot adapt to aging equipment or variations in operating parameters across charging stations, resulting in persistently low rates of missed detection and false positives.

[0004] Existing technologies also have problems with real-time performance and insufficient intelligence. Most solutions have long data upload cycles and are unable to capture millisecond-level transient faults, which are often the main cause of charging safety accidents. In addition, fault diagnosis relies on manual experience to update the rule base and lacks the ability to adaptively learn unknown fault modes, resulting in a significantly increased risk of failure when the system faces new power electronic devices or complex fault scenarios. Summary of the Invention The embodiments of the present application provide a charging pile fault identification method, device and medium based on power fingerprint to solve the above technical problems.

[0005] On the one hand, an embodiment of the present application provides a charging pile fault identification method based on power fingerprint, including: Collecting power signal data when the charging pile is running, preprocessing the power signal data, and extracting multidimensional features from the preprocessed power signal data; the multidimensional features include time domain features, frequency domain features, and transient features; Performing dimensionality reduction processing on the multi-dimensional features using principal component analysis technology to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile; Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in normal state to build the corresponding fault feature library; Calculate the cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library, and input the cosine similarity into a pre-trained deep learning model to determine the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output result.

[0006] In one implementation of the present application, multi-dimensional feature extraction is performed on the pre-processed power signal data, specifically including: Extracting time domain features from the preprocessed power signal data; the time domain features include voltage RMS, current RMS, active power, and reactive power, and the voltage RMS and current RMS are calculated using root mean square values; Performing a fast Fourier transform on the preprocessed power signal data to calculate the amplitudes of the fundamental wave and harmonic components, and calculating the total harmonic distortion rate based on the amplitudes of the fundamental wave and harmonic components; The wavelet transform method is used to decompose the preprocessed power signal data to obtain the waveform mutation characteristics of the transient process.

[0007] In one implementation of the present application, the multidimensional features are reduced in dimension by principal component analysis technology to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile, specifically including: Standardizing the multidimensional feature matrix corresponding to the multidimensional features to eliminate dimensional differences; Calculating the covariance matrix of the standardized multidimensional feature matrix and extracting the eigenvectors of the covariance matrix; The extracted feature vectors are sorted according to the size of the eigenvalues, and a preset number of principal components are selected in turn according to the sorting to achieve dimensionality reduction of the features, so as to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile after dimensionality reduction.

[0008] In one implementation of the present application, the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in the normal state are obtained to construct the corresponding fault feature library, specifically including: Simulating multiple preset fault scenarios through a simulation model and generating simulated power signal data corresponding to the multiple preset fault scenarios; Collect fault case data from the actual operation of the charging pile and extract the corresponding historical power fingerprint vectors from the fault case data; A fault feature library is constructed, and the simulated power signal data and the historical power fingerprint vector are stored in the fault feature library.

[0009] In one implementation of the present application, calculating the cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library specifically includes: Calculating the similarity between the real-time power fingerprint vector and the normal-state power fingerprint vector, and comparing the similarity with a preset normal threshold; If the similarity is lower than the preset normal threshold, the real-time power fingerprint vector is marked as abnormal, and similarity calculations are performed on the abnormal real-time power fingerprint vector and the power fingerprint vectors of various fault types in the fault feature library in sequence.

[0010] In one implementation of the present application, determining the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output result specifically includes: When the cosine similarity between the real-time power fingerprint vector and the power fingerprint vector in the fault feature library exceeds a first preset threshold, determining that the real-time power fingerprint vector belongs to the fault type corresponding to the power fingerprint vector in the fault feature library; Determining a maximum output probability and a corresponding fault type among multiple output probabilities of the deep learning model, and when the maximum output probability exceeds a second preset threshold, determining that the real-time power fingerprint vector belongs to the fault type corresponding to the maximum output probability; When the cosine similarity between the real-time power fingerprint vector and the power fingerprint vector in the fault feature library does not reach the first preset threshold, and the maximum output probability of the deep learning model does not exceed the second preset threshold, the cosine similarity and the output probability are weightedly multiplied, and it is determined that the real-time power fingerprint vector belongs to the fault type corresponding to the maximum value of the product.

[0011] In one implementation of the present application, after inputting the cosine similarity into a pre-trained deep learning model, the method further includes: Inputting multiple cosine similarities and the fault type label corresponding to each cosine similarity into the deep learning model; The model parameters of the deep learning model are optimized by the gradient descent algorithm, and the weight matrix and bias vector are updated within a preset period.

[0012] In one implementation of the present application, collecting power signal data during operation of the charging pile and preprocessing the power signal data specifically includes: Using non-invasive sensors, the power signal data of the charging pile is collected during operation; the power signal data includes voltage, current, power factor, and harmonic components; The power signal data is denoised using a median filter, and the denoised power signal data is normalized using maximum normalization.

[0013] On the other hand, an embodiment of the present application further provides a charging pile fault identification device based on power fingerprint, the device comprising: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned charging pile fault identification method based on power fingerprint.

[0014] On the other hand, an embodiment of the present application further provides a non-volatile computer storage medium storing computer-executable instructions. When the computer-executable instructions are executed, a charging pile fault identification method based on power fingerprint as described above is implemented.

[0015] The embodiments of the present application provide a charging pile fault identification method, device, and medium based on power fingerprint, which have at least the following beneficial effects: By directly collecting power signals such as voltage and current through non-invasive sensors, there is no need to modify the charging pile hardware or deploy additional sensor arrays, which greatly reduces the equipment modification cost and deployment complexity, and avoids the false alarm problem caused by electromagnetic interference or aging of traditional sensors; combining the time domain effective value, frequency domain harmonic component and transient waveform mutation characteristics, a power fingerprint covering the full-dimensional operating status of the charging pile is constructed, which solves the problem of missed detection caused by the single feature of existing technologies and significantly improves the sensitivity of fault detection; principal component analysis is used to reduce the dimensionality of multi-dimensional features, while retaining key information, reducing data redundancy and improving real-time computing efficiency; through the dual judgment logic of cosine similarity matching and deep learning model, it can not only quickly respond to high-similarity faults, but also realize accurate classification of complex scenarios based on probability distribution, solving the problems of poor adaptability of traditional threshold methods and delayed updates of manual rule bases. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A schematic diagram of a flow chart of a charging pile fault identification method based on power fingerprint provided in an embodiment of the present application; Figure 2 A schematic diagram of the internal structure of a charging pile fault identification device based on power fingerprint provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0019] Figure 1 A flow chart of a charging pile fault identification method based on power fingerprint provided in an embodiment of the present application.

[0020] The analysis method involved in the embodiments of the present application can be implemented by a terminal device or a server, and the present application does not impose any special restrictions on this. For ease of understanding and description, the following embodiments are described in detail using a server as an example.

[0021] It should be noted that the server can be a single device or a system composed of multiple devices, that is, a distributed server, and this application does not make any specific restrictions on this.

[0022] like Figure 1 As shown, the embodiment of the present application provides a charging pile fault identification method based on power fingerprint, including: Step 101: Collect power signal data when the charging pile is running, pre-process the power signal data, and perform multi-dimensional feature extraction on the pre-processed power signal data.

[0023] It should be noted that the multidimensional features in the embodiment of the present application include time domain features, frequency domain features and transient features.

[0024] In this embodiment, power signal data is collected using non-invasive sensors. It is understood that non-invasive sensors, such as Hall current sensors or voltage couplers, are deployed on the surface of the charging pile's external power supply cables. These sensors, such as Hall current sensors or voltage couplers, do not require access to the charging pile's internal circuitry, thus avoiding modifications to existing hardware structures. For example, the sensors collect current signals through electromagnetic induction and voltage signals through capacitive coupling, ensuring the continuity and safety of signal acquisition.

[0025] Preprocessing of collected power signal data involves denoising and normalization. Denoising utilizes a median filter, whose core logic involves calculating the median value of the signal sequence using a sliding window. This effectively suppresses impulse noise and high-frequency interference while preserving the integrity of the fundamental and harmonic components. Normalization utilizes maximum normalization, mapping characteristic values of different dimensions, such as voltage and current, to the [0, 1] range.

[0026] When performing multi-dimensional feature extraction on pre-processed power signal data, it specifically involves the extraction of time domain, frequency domain, and transient features. For example, for time domain feature extraction, the root mean square (RMS) values of voltage and current are calculated to reflect the effective energy of the signal; active power, which is the product integral of the in-phase components of voltage and current, and reactive power, which is the product integral of the out-of-phase components, are extracted to evaluate the load status of the charging pile. It is understandable that when the charging gun has poor contact, the effective current value may suddenly drop below the rated value, at which time the active power drops significantly.

[0027] To extract frequency domain features, a fast Fourier transform (FFT) is performed on the voltage and current signals to analyze the amplitude proportions of the fundamental wave (50 Hz) and each harmonic (such as the 3rd, 5th, and 7th). Total harmonic distortion (THD) is calculated as the ratio of the square root of the sum of the squared harmonic amplitudes to the fundamental wave amplitude and is used to quantify power quality. For example, a rectifier module failure may cause an abnormally high amplitude of the 3rd harmonic.

[0028] The Discrete Wavelet Transform (DWT) is used to perform multi-level decomposition of the signal to extract the waveform mutation characteristics of the transient process. Specifically, the high-frequency detail components (such as the D5 level) of the wavelet decomposition can be used to capture the voltage sag at the closing moment or the current spike caused by the contact arc.

[0029] Step 102: Perform dimensionality reduction processing on the multi-dimensional features by using principal component analysis technology to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile.

[0030] In this embodiment, principal component analysis (PCA) is used to eliminate redundant information between multidimensional features and reduce data dimensionality. The multidimensional feature matrix is normalized so that each feature has a mean of 0 and a variance of 1 to eliminate dimensional differences. For example, if the time domain effective value ranges from [0, 100] and the harmonic distortion rate ranges from [0, 10], normalization ensures that their contributions to the principal component are balanced.

[0031] Calculate the covariance matrix of the standardized feature matrix. The covariance matrix reflects the linear correlation between different features. For example, the voltage RMS value may have a positive correlation with the harmonic distortion rate, while the transient high-frequency energy may have a weak correlation with other features.

[0032] Eigenvalue decomposition is used to obtain the eigenvectors and corresponding eigenvalues of the covariance matrix. A larger eigenvalue indicates a greater data variance in the direction of the corresponding eigenvector, meaning it contains more information. Specifically, the eigenvalues are sorted from largest to smallest, and the first several principal components (for example, the first 10 principal components) are selected as the power fingerprint vector after dimensionality reduction.

[0033] The original feature matrix is projected onto the selected principal component direction to generate a low-dimensional power fingerprint vector. As can be seen, the power fingerprint vector not only retains the key information of the original data but also significantly reduces the complexity of subsequent matching and calculation. For example, the 20-dimensional original features of a charging station can be compressed into a 10-dimensional fingerprint vector after PCA dimensionality reduction.

[0034] Step 103: Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in a normal state to construct a corresponding fault feature library.

[0035] In this embodiment, the fault signature library is constructed by integrating simulation data with actual fault cases. Circuit simulation software is used to simulate various preset fault scenarios, such as contactor failure, line aging, overvoltage / undervoltage, and so on, to generate corresponding power signal data. For example, in the contactor fault model, the power loss caused by contact oxidation is simulated by increasing the contact resistance value, and the power fingerprint vector under the fault state is extracted.

[0036] Obtain historical charging pile fault records from the operation and maintenance platform and extract power signal data before and after the fault occurs. For example, if a charging pile's reactive power slowly decreases due to capacitor aging, the fingerprint of this gradual change can serve as a reference template for aging faults.

[0037] The simulated data is merged with the actual data and stored in a sparse storage database. As you can see, sparse storage uses a key-value pair structure, with the fault type as the key and the fingerprint vector as the value. This supports fast retrieval and dynamic expansion. When a new fault type is added, only the entry needs to be appended, without rebuilding the entire database, improving the system's scalability.

[0038] Step 104: Calculate the cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library, and input the cosine similarity into a pre-trained deep learning model to determine the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output result.

[0039] In this embodiment, fault diagnosis is achieved through the synergy of cosine similarity matching and deep learning models. The similarity between the real-time power fingerprint vector and each vector in the fault signature library is calculated using the cosine similarity formula, which measures the directional consistency between the two in multidimensional space.

[0040] The similarity results are input into a pre-trained 3-layer fully connected network. It is understood that the network consists of an input layer (the dimension is related to the number of fault types), a hidden layer (e.g., 64 nodes), and an output layer (Softmax probability distribution).

[0041] In this embodiment, during each update cycle, new similarity data and fault type classification results are input into the deep learning model, and the loss function and gradient are recalculated. The Adam optimization algorithm is then used to update the model parameters based on the new gradients, allowing the model to adapt to changes in the charging pile's operating status and emerging fault modes.

[0042] In this embodiment, when the similarity between the real-time fingerprint and a certain type of fault fingerprint exceeds a first threshold, it is directly determined to be that type of fault. For example, the fingerprint characteristics of a contactor fault are clear, and the similarity may remain above the threshold.

[0043] If the maximum probability output by the model exceeds the second threshold, the corresponding fault type is adopted. For example, if the model output probability of line aging fault reaches 0.9, it can be directly judged.

[0044] If both the similarity and model probability fall below the threshold, the product of the two is used as a comprehensive score, and the fault type corresponding to the maximum value is selected. For example, if a transient fault has a similarity of 0.8 and a model probability of 0.7, the weighted product is 0.56. If this is the highest value, it is determined to be this type of fault.

[0045] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, this application embodiment also provides a charging pile fault identification device based on power fingerprint, the structure of which is as follows: Figure 2 shown.

[0046] Figure 2 This is a schematic diagram of the internal structure of a charging pile fault identification device based on power fingerprint provided in an embodiment of the present application. Figure 2 As shown, the equipment includes: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Collect power signal data from charging piles during operation, pre-process the power signal data, and extract multi-dimensional features from the pre-processed power signal data; the multi-dimensional features include time domain features, frequency domain features, and transient features; The principal component analysis technique is used to reduce the dimensionality of multi-dimensional features and generate a real-time power fingerprint vector to characterize the operating status of the charging pile. Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in normal state to build the corresponding fault feature library; The cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library is calculated, and the cosine similarity is input into a pre-trained deep learning model to determine the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output results.

[0047] The present application also provides a non-volatile computer storage medium storing computer-executable instructions. When the computer-executable instructions are executed, they can: Collect power signal data from charging piles during operation, pre-process the power signal data, and extract multi-dimensional features from the pre-processed power signal data; the multi-dimensional features include time domain features, frequency domain features, and transient features; The principal component analysis technique is used to reduce the dimensionality of multi-dimensional features and generate a real-time power fingerprint vector to characterize the operating status of the charging pile. Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in normal state to build the corresponding fault feature library; The cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library is calculated, and the cosine similarity is input into a pre-trained deep learning model to determine the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output results.

[0048] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device and medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0049] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0050] 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.

[0051] 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.

[0052] 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.

[0053] 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.

[0054] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0055] Memory may include non-permanent storage 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. Memory is an example of a computer-readable medium.

[0056] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The 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 memory (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 technology, 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 (transitory media), such as modulated data signals and carrier waves.

[0057] 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.

[0058] 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 charging pile fault identification method based on power fingerprint, characterized in that: The method comprises: Collecting power signal data when the charging pile is running, preprocessing the power signal data, and extracting multidimensional features from the preprocessed power signal data; the multidimensional features include time domain features, frequency domain features, and transient features; Performing dimensionality reduction processing on the multi-dimensional features using principal component analysis technology to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile; Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in normal state to build the corresponding fault feature library; Calculate the cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library, and input the cosine similarity into a pre-trained deep learning model to determine the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output result.

2. A charging pile fault identification method based on power fingerprint according to claim 1, characterized in that: Perform multi-dimensional feature extraction on the pre-processed power signal data, including: Extracting time domain features from the preprocessed power signal data; the time domain features include voltage RMS, current RMS, active power, and reactive power, and the voltage RMS and current RMS are calculated using root mean square values; Performing a fast Fourier transform on the preprocessed power signal data to calculate the amplitudes of the fundamental wave and harmonic components, and calculating the total harmonic distortion rate based on the amplitudes of the fundamental wave and harmonic components; The wavelet transform method is used to decompose the preprocessed power signal data to obtain the waveform mutation characteristics of the transient process.

3. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: The multi-dimensional features are reduced in dimension using principal component analysis technology to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile, specifically including: Standardizing the multidimensional feature matrix corresponding to the multidimensional features to eliminate dimensional differences; Calculating the covariance matrix of the standardized multidimensional feature matrix and extracting the eigenvectors of the covariance matrix; The extracted feature vectors are sorted according to the size of the eigenvalues, and a preset number of principal components are selected in turn according to the sorting to achieve dimensionality reduction of the features, so as to generate a real-time power fingerprint vector for characterizing the operating status of the charging pile after dimensionality reduction.

4. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: Obtain the fault power fingerprint vector corresponding to the historical fault type of the charging pile and the standard power fingerprint vector of the charging pile in normal state to build the corresponding fault feature library, specifically including: Simulating multiple preset fault scenarios through a simulation model and generating simulated power signal data corresponding to the multiple preset fault scenarios; Collect fault case data from the actual operation of the charging pile and extract the corresponding historical power fingerprint vectors from the fault case data; A fault feature library is constructed, and the simulated power signal data and the historical power fingerprint vector are stored in the fault feature library.

5. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: Calculating the cosine similarity between the real-time power fingerprint vector and each power fingerprint vector in the fault feature library specifically includes: Calculating the similarity between the real-time power fingerprint vector and the normal-state power fingerprint vector, and comparing the similarity with a preset normal threshold; If the similarity is lower than the preset normal threshold, the real-time power fingerprint vector is marked as abnormal, and similarity calculations are performed on the abnormal real-time power fingerprint vector and the power fingerprint vectors of various fault types in the fault feature library in sequence.

6. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: Determining the fault type corresponding to the real-time power fingerprint vector based on the cosine similarity and the model output result specifically includes: When the cosine similarity between the real-time power fingerprint vector and the power fingerprint vector in the fault feature library exceeds a first preset threshold, determining that the real-time power fingerprint vector belongs to the fault type corresponding to the power fingerprint vector in the fault feature library; Determining a maximum output probability and a corresponding fault type among multiple output probabilities of the deep learning model, and when the maximum output probability exceeds a second preset threshold, determining that the real-time power fingerprint vector belongs to the fault type corresponding to the maximum output probability; When the cosine similarity between the real-time power fingerprint vector and the power fingerprint vector in the fault feature library does not reach the first preset threshold, and the maximum output probability of the deep learning model does not exceed the second preset threshold, the cosine similarity and the output probability are weightedly multiplied, and it is determined that the real-time power fingerprint vector belongs to the fault type corresponding to the maximum value of the product.

7. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: After inputting the cosine similarity into a pre-trained deep learning model, the method further includes: Inputting multiple cosine similarities and the fault type label corresponding to each cosine similarity into the deep learning model; The model parameters of the deep learning model are optimized by the gradient descent algorithm, and the weight matrix and bias vector are updated within a preset period.

8. The method for identifying charging pile faults based on power fingerprint according to claim 1, characterized in that: Collecting power signal data when the charging pile is running and preprocessing the power signal data, specifically including: Using non-invasive sensors, the power signal data of the charging pile is collected during operation; the power signal data includes voltage, current, power factor, and harmonic components; The power signal data is denoised using a median filter, and the denoised power signal data is normalized using maximum normalization.

9. A charging pile fault identification device based on power fingerprint, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a charging pile fault identification method based on power fingerprint as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: When the computer-executable instructions are executed, a charging pile fault identification method based on power fingerprint as described in any one of claims 1 to 8 is implemented.

Images