Voltage output control method and device of charger and charger

Abstract

The present application relates to the technical field of charging control, and discloses a method, device and charger for controlling the voltage output of a charger. The method comprises: initializing the charging strategy of the charger, obtaining the initial charging strategy of each charging stage and collecting the test current signal and the test voltage signal; performing adaptive Gaussian window sparse pattern decomposition, obtaining the current signal mode and the voltage signal mode of each charging stage; performing voltage output abnormality detection, obtaining the target abnormality detection result of each charging stage; performing pattern quantity analysis, obtaining the number of current modes and the number of voltage modes of each charging stage; performing signal reconstruction, obtaining the target current signal and the target voltage signal; performing voltage output analysis and charging strategy adjustment, obtaining the target charging strategy of each charging stage, and the present application improves the voltage output control accuracy of the charger.

Description

Technical Field

[0001] The present application relates to the field of charging control technology, and in particular to a method and device for controlling voltage output of a charger, and a charger. Background Art

[0002] With the popularity of battery-powered devices such as mobile devices and electric vehicles, the requirements for charging efficiency and safety are becoming higher and higher. Traditional charger charging strategies often adopt fixed charger charging modes, such as constant current charging and constant voltage charging. Although these strategies meet the charging needs to a certain extent, due to the lack of real-time response and adjustment to the battery status, it is difficult to ensure the battery life and safety while ensuring charging efficiency. In particular, the battery's charging acceptance and safety threshold will change at different charging stages. Traditional charging strategies cannot flexibly adapt to these changes, resulting in low charging efficiency and may even damage battery health.

[0003] Existing charger charging technologies often fail to accurately capture subtle changes in the charging process when dealing with complex charging environments and changing battery states, such as the impact of temperature fluctuations, battery aging and other factors on charging strategies. This not only affects charging efficiency, but may also lead to battery performance degradation and even safety accidents. Summary of the invention

[0004] The present application provides a charger voltage output control method, device and charger, which are used to improve the accuracy of charger voltage output control.

[0005] In a first aspect, the present application provides a method for controlling a voltage output of a charger, the method comprising:

[0006] Initialize the charging strategy of the charger to obtain the initial charging strategy of each charging stage, and collect the test current signal and the test voltage signal of each charging stage;

[0007] Performing adaptive Gaussian window sparse pattern decomposition on the test current signal and the test voltage signal respectively to obtain a current signal pattern and a voltage signal pattern in each charging stage;

[0008] Perform voltage output abnormality detection in each charging stage according to the current signal mode and the voltage signal mode to obtain a target abnormality detection result in each charging stage;

[0009] According to the target abnormality detection result, the current signal mode and the voltage signal mode are respectively analyzed for mode quantity, so as to obtain the current mode quantity and the voltage mode quantity of each charging stage;

[0010] Reconstruct the test current signal and the test voltage signal according to the number of current modes and the number of voltage modes to obtain a target current signal and a target voltage signal;

[0011] The voltage output analysis and charging strategy adjustment are performed on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

[0012] In a second aspect, the present application provides a voltage output control device for a charger, the voltage output control device for the charger comprising:

[0013] The acquisition module is used to initialize the charging strategy of the charger, obtain the initial charging strategy of each charging stage, and collect the test current signal and the test voltage signal of each charging stage;

[0014] A decomposition module, used to perform adaptive Gaussian window sparse pattern decomposition on the test current signal and the test voltage signal respectively to obtain a current signal pattern and a voltage signal pattern in each charging stage;

[0015] A detection module, used to perform voltage output abnormality detection in each charging stage according to the current signal mode and the voltage signal mode, and obtain a target abnormality detection result in each charging stage;

[0016] An analysis module, configured to perform pattern quantity analysis on the current signal pattern and the voltage signal pattern respectively according to the target abnormality detection result, so as to obtain the number of current patterns and the number of voltage patterns in each charging stage;

[0017] A reconstruction module, used to reconstruct the test current signal and the test voltage signal according to the number of current modes and the number of voltage modes, respectively, to obtain a target current signal and a target voltage signal;

[0018] The adjustment module is used to perform voltage output analysis and charging strategy adjustment on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

[0019] A third aspect of the present application provides a charger voltage output control device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the charger voltage output control device executes the above-mentioned charger voltage output control method.

[0020] A fourth aspect of the present application provides a charger, which is used to execute the above-mentioned charger voltage output control method.

[0021] In the technical solution provided by the present application, by performing adaptive Gaussian window sparse mode decomposition on the test current signal and test voltage signal of each charging stage, the subtle changes of the battery state in different charging stages can be accurately captured. This adaptability ensures that the charging strategy can flexibly respond to the real-time state of the battery and changes in the external environment, thereby improving the intelligence and adaptability of the charging process. Through precise pattern quantity analysis and signal reconstruction, the charging control method can adjust the charging parameters and optimize the charging strategy in a targeted manner. This not only shortens the charging time, but also ensures the high efficiency of the charging process, while reducing energy loss and improving the overall charging efficiency. The current signal mode and the voltage signal mode are used to detect abnormal voltage output, and abnormal situations that may occur during the charging process are discovered and handled in a timely manner. This real-time monitoring and abnormal handling mechanism significantly enhances the safety of the charging process and effectively prevents overcharging, overheating, etc., which may cause battery performance degradation or safety accidents. The charging control method avoids improper charging of the battery, such as overcharging or voltage and current mutations during the charging process, through the careful management of the charging process, thereby reducing the loss and aging of the battery and extending the service life of the battery. By optimizing and adjusting the charging strategy, the stability and reliability of the charging process are ensured, and the user's charging experience is improved. Especially in application scenarios such as mobile devices and electric vehicles, it can provide faster, safer and smarter charging services, meet users' high standards for charging efficiency and safety, and thus improve the accuracy of the charger's voltage output control. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of an embodiment of a method for controlling a voltage output of a charger in an embodiment of the present application;

[0024] Figure 2 Schematic diagram of an embodiment of a voltage output control device of a charger in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The embodiments of the present application provide a method, device and charger for controlling the voltage output of a charger. The terms "first", "second", "third", "fourth", etc. (if any) 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 sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof 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 that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the charger voltage output control method of the present application, the method includes:

[0027] Step S101, initializing the charging strategy of the charger, obtaining the initial charging strategy of each charging stage, and collecting the test current signal and the test voltage signal of each charging stage;

[0028] It is understandable that the execution subject of the present application may be a voltage output control device of a charger, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0029] Specifically, multiple charging stages of the charger are defined, including a constant current charging stage, a constant voltage charging stage, and a trickle charging stage, and the charging stage is determined according to the relationship between the battery voltage and the preset value. When the battery voltage is lower than the first preset value, the charging process is in the constant current charging stage, which can quickly inject initial energy into the battery; as the battery voltage increases, when the battery voltage exceeds the first preset value and is lower than the second preset value, the charger enters the constant voltage charging stage, and controls the charging speed by adjusting the voltage output to protect the battery and optimize the charging efficiency; when the battery voltage exceeds the second preset value, the charging enters the trickle charging stage, and the battery is charged with a lower current for the final charge to ensure that the battery is fully charged and prolong the battery life. The voltage output control parameter analysis is performed for each charging stage, including defining the control range of the voltage and current in each stage to ensure the stable output of the voltage and current during the charging process. Through parameter analysis, an initial charging strategy is set for each charging stage, and these strategies will guide the charger on how to adjust the voltage and current at different stages to optimize the charging efficiency and protect the battery safety. Based on the initial charging strategy, the voltage output control test is carried out to verify the effectiveness of the charging strategy, and the original current signal and the original voltage signal are collected to provide data support for subsequent adjustments and optimizations. The DC component of the original current signal and the original voltage signal are removed and filtered to remove noise and unnecessary interference in the signal, ensuring the accuracy of subsequent analysis and judgment.

[0030] Step S102, performing adaptive Gaussian window sparse pattern decomposition on the test current signal and the test voltage signal respectively to obtain a current signal pattern and a voltage signal pattern in each charging stage;

[0031] Specifically, the test current signal and the test voltage signal are segmented in time series respectively. By dividing the continuous current and voltage signals into multiple sections, each section represents a specific time interval in the charging process, so as to analyze and process the signals more finely and capture the changing characteristics of the current and voltage in the charging process more accurately. The initial current signal segment and voltage signal segment are windowed by an adaptive Gaussian window function. The discontinuity of the signal edge can be effectively reduced by the smoothing characteristics of the Gaussian window function. The windowed current signal segment and voltage signal segment are subjected to discrete Fourier transform, and these signals are converted from the time domain to the frequency domain. The characteristics of the signal are analyzed from the perspective of the frequency domain, including the main frequency components and energy distribution of the signal. The sparse coding algorithm is used to decompose the patterns of multiple frequency domain current signal segments and voltage signal segments, and sparse representations are extracted from numerous frequency domain signals. These sparse representations can reveal the main components and patterns of the signal. The sparse coding algorithm can effectively compress data while retaining the key features of the signal when processing the signal. Based on the sparse representation of the current signal and the voltage signal, the current signal pattern and the voltage signal pattern of each charging stage are obtained through the reconstruction process.

[0032] Step S103, performing voltage output abnormality detection for each charging stage according to the current signal mode and the voltage signal mode, and obtaining a target abnormality detection result for each charging stage;

[0033] Specifically, the energy spectrum curve of the current signal mode and the voltage signal mode are fitted by the iterative least squares algorithm to obtain the current energy spectrum curve and the voltage energy spectrum curve. Features are separated from the energy spectrum curve to form a current feature set and a voltage feature set, and key information that can represent the current and voltage behavior characteristics is identified and extracted. The current feature set and the voltage feature set are input into the preset anomaly detection model, which includes an input layer, an attention mechanism layer, an autoencoder and an output layer. The accuracy and efficiency of anomaly detection are improved through layered processing. In the input layer of the model, the current feature set and the voltage feature set are encoded by a unique hot vector respectively and converted into a current feature encoding vector and a voltage feature encoding vector. The current feature encoding vector and the voltage feature encoding vector are input into the attention mechanism layer. The attention mechanism fuses vectors by assigning different weights to different features, and generates an attention feature encoding vector containing the importance distribution of current and voltage information. This mechanism enables the model to focus on more important features, thereby improving the accuracy of anomaly detection. The attention feature encoding vector is input into the autoencoder layer. The autoencoder is a feature learning tool that extracts deeper hidden feature encoding vectors through the hidden layer. In this process, the autoencoder learns how to reconstruct its input data and learns the compressed representation of the data in this way. The hidden feature encoding vector is processed by the ReLU function in the output layer to predict the voltage output abnormality probability and obtain the voltage output abnormality probability value for each charging stage. As a nonlinear activation function, the ReLU function helps to improve the nonlinear fitting ability of the model prediction. According to the voltage output abnormality probability value, the target abnormality detection result for each charging stage is finally output.

[0034] Step S104, performing pattern quantity analysis on the current signal pattern and the voltage signal pattern respectively according to the target abnormality detection result, and obtaining the current pattern quantity and the voltage pattern quantity of each charging stage;

[0035] Specifically, according to the target abnormality detection result, the energy distribution of the current energy spectrum curve of the current signal mode is calculated to clarify the energy distribution of the current signal at different frequencies, and the energy distribution of the voltage energy spectrum curve of the voltage signal mode is calculated to clarify the energy distribution of the voltage signal at different frequencies, and the detailed information of the current and voltage energy distribution intervals is obtained. The energy distribution interval reflects the basic characteristics of the energy change of the current and voltage signals during the entire charging process. The frequency components of the current and voltage signal modes are obtained by performing spectrum analysis on the current energy distribution interval and the voltage energy distribution interval. The energy size of each frequency component in the signal is identified to reveal the main frequency components in the current and voltage signals. The current frequency components and the voltage frequency components are clustered and analyzed respectively, and similar frequency components are classified into one category to obtain the current frequency classification results and the voltage frequency classification results. Cluster analysis identifies the main modes and change trends from a large number of frequency components. The energy contribution calculation is performed on the current frequency classification results and the voltage frequency classification results, and the energy proportion of the frequency signal in each classification result is calculated to quantify the contribution of different frequency signals to the overall signal energy. The number of current modes and the number of voltage modes in each charging stage are determined according to the energy proportion of the current frequency signal and the energy proportion of the voltage frequency signal. By analyzing the energy contribution of the frequency components, the changes in the current and voltage patterns during the charging process are identified, ensuring that the charging control strategy can be accurately adjusted to cope with various situations that may arise during the charging process.

[0036] Step S105, reconstructing the test current signal and the test voltage signal according to the number of current modes and the number of voltage modes, respectively, to obtain a target current signal and a target voltage signal;

[0037] Specifically, the number of current modes and the test current signal are input into a preset first signal reconstruction model, and the model structure includes multiple convolution pooling layers, a fully connected layer, a long short-term memory (LSTM) layer, and a first signal merging layer. The number of current modes and the test current signal are convolutionally operated through multiple first convolution pooling layers to extract local features of the signal, and the feature dimension is reduced through pooling operations to improve the robustness of the feature. The current convolution feature is input into the fully connected layer for feature dimension reduction processing to simplify the data structure, making the feature set more compact while retaining important information. The current dimension reduction feature is input into the LSTM layer for time series fusion. The LSTM layer is suitable for processing time series data and can effectively capture the dynamic changes of the signal in the time dimension. All time series current signals are integrated through the first signal merging layer for signal reconstruction to obtain the target current signal. The number of voltage modes and the test voltage signal are input into a preset second signal reconstruction model, and the second signal reconstruction model includes: multiple second convolution pooling layers, multiple second fully connected layers, multiple second LSTM layers, and a second signal merging layer. The convolutional pooling layer is used to extract and reduce the dimensions of the voltage pattern number and the test voltage signal, and then the fully connected layer is used to further compress the feature space. The LSTM layer is then used to process the timing information of the features. Finally, the signal merging layer is used to integrate the results of each processing step to reconstruct the target voltage signal.

[0038] Step S106: perform voltage output analysis and charging strategy adjustment on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

[0039] Specifically, the target current signal and the target voltage signal are analyzed by the digital-to-analog converter, and the current and voltage signals reconstructed by the signal processing technology are converted into actual voltage output, so as to analyze the actual working state of the charger and obtain any voltage output deviation that may exist in each charging stage, so as to provide a basis for adjusting the control parameters. Based on the adjustment control parameters obtained by the voltage output analysis, the voltage output of the charger is precisely controlled. This includes real-time adjustment of the voltage and current, and feedback parameter collection of the charger. These feedback parameter data sets contain the real-time voltage, current and other important parameters during the charging process. By analyzing the feedback parameter data set, the strategy adjustment gradient of each charging stage is calculated. By analyzing the feedback data, the direction and amplitude of the charging strategy adjustment are determined to ensure that each adjustment is made in the direction of improving charging efficiency and safety. The strategy adjustment gradient is the key to determining the details of the charging strategy adjustment, and guides the specific operations of the charging strategy adjustment, including adjusting the charging voltage, the current and the switching of the charging mode. The charging strategy is adjusted for the initial charging strategy of each charging stage according to the strategy adjustment gradient to obtain the target charging strategy of each charging stage.

[0040] In the embodiment of the present application, by performing adaptive Gaussian window sparse mode decomposition on the test current signal and the test voltage signal of each charging stage, the subtle changes in the battery state at different charging stages can be accurately captured. This adaptability ensures that the charging strategy can flexibly respond to the real-time state of the battery and changes in the external environment, thereby improving the intelligence and adaptability of the charging process. Through precise pattern quantity analysis and signal reconstruction, the charging control method can adjust the charging parameters and optimize the charging strategy in a targeted manner. This not only shortens the charging time, but also ensures the high efficiency of the charging process, while reducing energy loss and improving the overall charging efficiency. The current signal mode and the voltage signal mode are used to detect voltage output anomalies, and abnormal situations that may occur during the charging process are promptly discovered and handled. This real-time monitoring and abnormality handling mechanism significantly enhances the safety of the charging process and effectively prevents situations such as overcharging and overheating that may cause battery performance degradation or safety accidents. The charging control method avoids improper charging of the battery, such as overcharging or voltage and current mutations during the charging process, through the careful management of the charging process, thereby reducing the loss and aging of the battery and extending the service life of the battery. By optimizing and adjusting the charging strategy, the stability and reliability of the charging process are ensured, and the user's charging experience is improved. Especially in application scenarios such as mobile devices and electric vehicles, it can provide faster, safer and smarter charging services, meet users' high standards for charging efficiency and safety, and thus improve the accuracy of the charger's voltage output control.

[0041] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0042] (1) defining multiple charging stages of the charger, the multiple charging stages including: a constant current charging stage, a constant voltage charging stage, and a trickle charging stage, wherein when the battery voltage is less than a first preset value, the charging stage is determined to be a constant current charging stage, when the first preset value is less than the battery voltage and less than a second preset value, the charging stage is determined to be a constant voltage charging stage, when the second preset value is less than the battery voltage, the charging stage is determined to be a trickle charging stage, and the first preset value is less than the second preset value;

[0043] (2) Analyze the voltage output control parameters for multiple charging stages respectively to obtain the voltage output control parameter range for each charging stage;

[0044] (3) Initializing the charging strategies for the multiple charging stages of the charger according to the voltage output control parameter range to obtain the initial charging strategy for each charging stage;

[0045] (4) Based on the initial charging strategy of each charging stage, the charger is tested for voltage output control, and the original current signal and original voltage signal of each charging stage are collected;

[0046] (5) The original current signal and the original voltage signal are respectively subjected to DC component removal and filtering processing to obtain the original current signal and the original voltage signal of each charging stage.

[0047] Specifically, multiple charging stages of the charger are defined. They include a constant current charging stage, a constant voltage charging stage, and a trickle charging stage, each corresponding to different requirements and characteristics in the battery charging process. In the constant current charging stage, the charger charges the battery with a fixed current. This stage is suitable for when the battery voltage is low, and can quickly increase the battery power to a certain level. When the battery voltage reaches a first preset value, the charging enters the constant voltage charging stage. In this stage, the charger adjusts the output to keep the voltage constant, thereby protecting the battery and avoiding overcharging. When the battery voltage exceeds the second preset value, it enters the trickle charging stage, and continues to charge with a smaller current to ensure that the battery is fully charged without overcharging. The voltage output control parameters of the charging stage are analyzed. The voltage and current requirements of each charging stage are evaluated to determine the optimal voltage output control parameter range for each stage. For example, the constant current charging stage may require a higher current output but a lower voltage output, while the constant voltage charging stage requires a stable voltage output, and the current output will automatically adjust with the battery state. Based on the analysis of the voltage output control parameters, the charging strategy of the multiple charging stages of the charger is initialized. An initial charging scheme is set for each charging stage, and the scheme is based on the voltage output control parameter range. The initial charging strategy will guide the charger on how to adjust its output at each stage to charge the battery most efficiently and safely. Based on the initial charging strategy for each charging stage, the charger is tested for voltage output control, and the raw current signal and raw voltage signal for each charging stage are collected to evaluate the actual effect of the initial charging strategy and identify possible problems or optimization opportunities. For example, in the constant current charging stage, if the actual data shows that the current fluctuates greatly, the charging strategy needs to be adjusted to stabilize the current output. The raw current signal and raw voltage signal are subjected to DC component removal and filtering to remove noise and irrelevant signal components from the collected data. For example, by using a low-pass filter, high-frequency noise introduced by power supply fluctuations or other electromagnetic interference is removed to obtain smoother current and voltage signal curves.

[0048] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0049] (1) performing time series segmentation on the test current signal and the test voltage signal respectively to obtain a plurality of initial current signal segments and a plurality of initial voltage signal segments;

[0050] (2) performing windowing processing on the multiple initial current signal segments and the multiple initial voltage signal segments respectively by using an adaptive Gaussian window function to obtain multiple windowed current signal segments and multiple windowed voltage signal segments;

[0051] (3) performing discrete Fourier transform on the multiple windowed current signal segments and the multiple windowed voltage signal segments respectively to obtain multiple frequency domain current signal segments and multiple frequency domain voltage signal segments;

[0052] (4) performing pattern decomposition on multiple frequency-domain current signal segments by using a sparse coding algorithm to obtain a sparse representation of the current signal, and performing pattern decomposition on multiple frequency-domain voltage signal segments by using a sparse coding algorithm to obtain a sparse representation of the voltage signal;

[0053] (5) The current signal pattern of each charging stage is reconstructed based on the sparse representation of the current signal, and the voltage signal pattern of each charging stage is reconstructed based on the sparse representation of the voltage signal.

[0054] Specifically, the test current signal and the test voltage signal are segmented into time series respectively, and the continuous signal is decomposed into multiple independent segments so as to analyze each segment more carefully. The initial current signal segment and the initial voltage signal segment are windowed by an adaptive Gaussian window function. The Gaussian window is a smooth window function that can reduce the discontinuity of the signal edge and make the signal smoother, thereby optimizing the spectral characteristics of the signal so that the subsequent frequency domain analysis can more accurately identify the important frequency components in the signal. The signal segment after windowing is discrete Fourier transformed to convert it from the time domain to the frequency domain. Discrete Fourier transform is a commonly used signal processing technology that can reveal the characteristics of the signal in the frequency domain. By performing Fourier transform on each windowed current signal segment and windowed voltage signal segment, the corresponding frequency domain current signal segment and frequency domain voltage signal segment are obtained. The frequency domain signal segment is pattern decomposed using a sparse coding algorithm. Sparse coding is a signal processing technology based on optimization that finds a sparse representation of the signal so that the signal can be represented by as few active components as possible. After applying the sparse coding algorithm to the frequency domain current signal segments and the frequency domain voltage signal segments, sparse representations of the current signal and the voltage signal are obtained, which represent the most important patterns in the signal. The current signal pattern and the voltage signal pattern of each charging stage are reconstructed based on the sparse representation of the current signal and the voltage signal. The sparse representation obtained by sparse coding is converted back to the time domain signal in order to understand the specific change pattern of the current and voltage in each charging stage. The reconstruction process considers the main frequency components identified in the sparse coding stage, combines their relative importance, and restores the pattern that best represents the change of the original signal.

[0055] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0056] (1) Using the iterative least squares algorithm, the energy spectrum curves of the current signal mode and the voltage signal mode are fitted to obtain the current energy spectrum curve and the voltage energy spectrum curve;

[0057] (2) Extracting features from the current energy spectrum curve and the voltage energy spectrum curve respectively to obtain a current feature set and a voltage feature set;

[0058] (3) Inputting the current feature set and the voltage feature set into a preset anomaly detection model, which includes: an input layer, an attention mechanism layer, an autoencoder, and an output layer;

[0059] (4) Performing one-hot vector encoding on the current feature set and the voltage feature set through the input layer to obtain a current feature encoding vector and a voltage feature encoding vector;

[0060] (5) Input the current feature encoding vector and the voltage feature encoding vector into the attention mechanism layer for vector fusion to obtain the attention feature encoding vector;

[0061] (6) Input the attention feature encoding vector into the autoencoder to extract hidden features and obtain the hidden feature encoding vector;

[0062] (7) The voltage output anomaly probability is predicted for the hidden feature encoding vector through the ReLU function in the output layer to obtain the voltage output anomaly probability value for each charging stage, and the target anomaly detection result for each charging stage is output based on the voltage output anomaly probability value.

[0063] Specifically, the energy spectrum curve of the current signal mode and the voltage signal mode is fitted by the iterative least squares algorithm. The actual energy spectrum distribution of the current and voltage signals is gradually approached by the iterative least squares algorithm to obtain the current energy spectrum curve and the voltage energy spectrum curve, effectively deal with the nonlinear problem of the signal, and gradually reduce the fitting error by iterative calculation, and finally obtain the energy spectrum curve matching the actual signal. Features are extracted from the energy spectrum curve, and information that is helpful for subsequent analysis, such as peak value, average energy, energy distribution range, etc., is identified to obtain the current feature set and the voltage feature set. The feature set is input into the preset anomaly detection model. The model includes an input layer, an attention mechanism layer, an autoencoder and an output layer. In the input layer, the feature set of current and voltage is converted into a unique hot vector encoding to effectively represent the classified data and provide the model with an easy-to-process feature form. The automatic evaluation of the importance of different features is introduced into the model through the attention mechanism layer, so that the model can pay more attention to those features that are more important for anomaly detection and improve the detection accuracy. The attention feature encoding vector is input into the autoencoder for hidden feature extraction. The autoencoder is an unsupervised learning neural network that captures key information in the data by learning the compressed representation of the input data. The ReLU function in the output layer predicts the voltage output abnormality probability of the hidden feature encoding vector, and obtains the voltage output abnormality probability value of each charging stage. As an activation function, the ReLU function can increase the nonlinearity of the model. Based on the abnormal probability value, the voltage output abnormality detection result of each charging stage is finally obtained.

[0064] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0065] (1) According to the target abnormality detection result, the energy distribution of the current energy spectrum curve of the current signal mode is calculated to obtain the current energy distribution range, and the energy distribution of the voltage energy spectrum curve of the voltage signal mode is calculated to obtain the voltage energy distribution range;

[0066] (2) Performing spectrum analysis on the current energy distribution interval to obtain the current frequency component of the current signal mode, and performing spectrum analysis on the voltage energy distribution interval to obtain the voltage frequency component of the voltage signal mode;

[0067] (3) Perform cluster analysis on the current frequency components to obtain the current frequency classification results, and perform cluster analysis on the voltage frequency components to obtain the voltage frequency classification results;

[0068] (4) Calculate the energy contribution of the current frequency classification results to obtain the current frequency signal ratio, and calculate the energy contribution of the voltage frequency classification results to obtain the voltage frequency signal ratio;

[0069] (5) The number of current modes in each charging stage is determined according to the proportion of the current frequency signal, and the number of voltage modes in each charging stage is determined according to the proportion of the voltage frequency signal.

[0070] Specifically, the energy distribution of the current energy spectrum curve of the current signal mode and the voltage signal mode is calculated according to the target abnormality detection result. The energy distribution of the current and voltage signals in the entire frequency domain is analyzed. For example, the current energy spectrum curve is obtained by Fourier transforming the current signal, which describes the energy size of the current signal at different frequencies. Similarly, the voltage signal is analyzed to obtain the voltage energy spectrum curve. The frequency components of the current and voltage signals are obtained by performing spectrum analysis on the energy distribution interval. The main frequency components in the current and voltage signals are determined by identifying the peaks in the energy spectrum curve. For example, if there are obvious energy peaks at 50Hz and 150Hz on the current energy spectrum curve, it means that the current signal has significant energy concentration at these two frequencies, reflecting the periodic changes or interference that may exist in the charging process. The frequency components of the current and voltage are clustered and analyzed, and the frequency components are grouped according to their similarities to obtain the frequency classification results of the current and voltage. For example, using the K-means clustering algorithm, it is found that the frequency components of the current signal can be divided into three categories: low frequency, medium frequency and high frequency, and the frequency components in each category have similar energy characteristics. The energy contribution of the current and voltage frequency classification results is calculated, the proportion of each category of frequency components in the total energy is analyzed, and the degree of influence of different frequency categories on the overall signal is quantified. Based on the energy proportion of the current and voltage frequency signals, the number of current and voltage modes in each charging stage is determined. According to the energy contribution of the frequency components, the change pattern of current and voltage during the charging process is evaluated. For example, if it is found that the low-frequency mode of the current accounts for the vast majority in the constant current charging stage, and the medium-frequency and high-frequency modes begin to appear in the constant voltage stage, it reflects the dynamics of the current and voltage changes during the charging process, and the charging strategy can be adjusted to cope with these changes.

[0071] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0072] (1) inputting the current pattern number and the test current signal into a preset first signal reconstruction model, the first signal reconstruction model comprising: a plurality of first convolutional pooling layers, a plurality of first fully connected layers, a plurality of first LSTM layers and a first signal merging layer;

[0073] (2) performing convolution feature operations on the number of current modes and the test current signal through multiple first convolution pooling layers to obtain multiple current convolution features, performing feature dimension reduction processing on the multiple current convolution features through multiple first fully connected layers to obtain multiple current dimension reduction features, performing current time series fusion on the multiple current dimension reduction features through multiple first LSTM layers to obtain multiple time series current signals, and performing signal reconstruction on the multiple time series current signals through the first signal merging layer to obtain a target current signal;

[0074] (3) inputting the voltage pattern number and the test voltage signal into a preset second signal reconstruction model, where the second signal reconstruction model includes: a plurality of second convolutional pooling layers, a plurality of second fully connected layers, a plurality of second LSTM layers, and a second signal merging layer;

[0075] (4) Convolutional feature operations are performed on the number of voltage patterns and the test voltage signal through multiple second convolutional pooling layers to obtain multiple voltage convolutional features, and feature dimension reduction is performed on the multiple voltage convolutional features through multiple second fully connected layers to obtain multiple voltage dimension reduction features. Voltage time series fusion is performed on the multiple voltage dimension reduction features through multiple second LSTM layers to obtain multiple time series voltage signals, and signal reconstruction is performed on the multiple time series voltage signals through the second signal merging layer to obtain the target voltage signal.

[0076] Specifically, the number of current patterns and the test current signal are input into a preset first signal reconstruction model. The model includes a convolution pooling layer, a fully connected layer, a long short-term memory (LSTM) layer, and a signal merging layer. The current signal is convoluted feature operation through multiple convolution pooling layers, and specific patterns in the signal, such as periodic changes or sudden jumps, are identified by applying multiple filters. The convolution operation captures the local dependency of the signal, while the pooling operation reduces the spatial dimension of the feature, which helps to reduce the amount of calculation and extract more abstract feature representations. The current convolution feature is input into the fully connected layer for feature dimension reduction processing, and the output of the previous layer is converted into a more compact form, which helps to reduce the complexity of the model and improve the computational efficiency. The current feature after dimension reduction processing is input into the LSTM layer. The LSTM layer is specially designed to process time series data and can learn long-term dependencies. Through the processing of the LSTM layer, the model can capture the dynamic characteristics of the current signal changing over time and generate a current signal with a time series structure. Multiple time series current signals are integrated through the first signal merging layer to reconstruct the target current signal. All processing results are summarized to ensure that the reconstructed current signal is as close to the real charging current signal as possible. Similarly, the number of voltage patterns and the test voltage signal are input into the second signal reconstruction model, and after convolution pooling, feature dimension reduction, time series fusion and signal merging, the target voltage signal is finally obtained. The reconstruction process takes into account the unique properties of the voltage signal and the voltage change characteristics during the charging process to ensure the accuracy and reliability of the reconstructed signal.

[0077] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0078] (1) Performing voltage output analysis on the target current signal and the target voltage signal through a digital-to-analog converter to obtain voltage output adjustment control parameters for each charging stage;

[0079] (2) adjusting the control parameters according to the voltage output to control the voltage output of the charger, and collecting the feedback parameters of the charger to obtain a feedback parameter data set for each charging stage;

[0080] (3) The strategy adjustment gradient of each charging stage is calculated according to the feedback parameter data set, and the initial charging strategy of each charging stage is adjusted according to the strategy adjustment gradient to obtain the target charging strategy of each charging stage.

[0081] Specifically, the target current signal and the target voltage signal are analyzed by a digital-to-analog converter to convert the digital signal into an analog signal, and the analog signals are analyzed to determine the optimal voltage output setting for each charging stage. By analyzing the target signal, a set of voltage output adjustment control parameters are obtained, which guide the charger on how to adjust its output to meet the charging requirements of different stages. The voltage output of the charger is controlled according to the voltage output adjustment control parameters. This includes adjusting the voltage and current output to meet the requirements of the target signal, and also includes real-time monitoring of the performance of the charger to ensure that the charging process is carried out according to the predetermined strategy. The feedback parameter collection of the charger involves collecting various runtime data, such as actual voltage and current output, charging time, and charger status information. These feedback parameter data sets are the basis for evaluating the actual performance of the charger and adjusting the charging strategy, so that the charging process can be optimized based on real-time data. Based on the collected feedback parameter data sets, the strategy adjustment gradient for each charging stage is calculated. By analyzing the deviation between the feedback data and the target signal, an algorithm is used to calculate the direction and magnitude of the adjustment required to minimize the deviation and optimize the charging performance. The strategy adjustment gradient reflects the specific adjustment measures required to transition from the current state to the target state, guides the detailed adjustment of the charging strategy, and ensures that the charging process can adapt to changes in the battery state and the influence of external conditions. Based on the strategy adjustment gradient, the initial charging strategy of each charging stage is adjusted accordingly. This includes modifying the size of the charging current and voltage, adjusting the shape of the charging curve, or changing the switching point of the charging stage. Through dynamic adjustment, the target charging strategy for each charging stage is determined, and a more accurate and efficient charging process can be achieved.

[0082] The above describes the voltage output control method of the charger in the embodiment of the present application. The following describes the voltage output control device of the charger in the embodiment of the present application. Figure 2 In one embodiment of the present application, a voltage output control device of a charger includes:

[0083] The acquisition module 201 is used to initialize the charging strategy of the charger, obtain the initial charging strategy of each charging stage, and acquire the test current signal and the test voltage signal of each charging stage;

[0084] A decomposition module 202, used to perform adaptive Gaussian window sparse pattern decomposition on the test current signal and the test voltage signal respectively to obtain a current signal pattern and a voltage signal pattern in each charging stage;

[0085] A detection module 203 is used to perform voltage output abnormality detection for each charging stage according to the current signal mode and the voltage signal mode to obtain a target abnormality detection result for each charging stage;

[0086] An analysis module 204 is used to perform pattern quantity analysis on the current signal pattern and the voltage signal pattern respectively according to the target abnormality detection result, so as to obtain the number of current patterns and the number of voltage patterns in each charging stage;

[0087] A reconstruction module 205, configured to reconstruct the test current signal and the test voltage signal according to the number of current modes and the number of voltage modes, respectively, to obtain a target current signal and a target voltage signal;

[0088] The adjustment module 206 is used to perform voltage output analysis and charging strategy adjustment on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

[0089] Through the cooperation of the above components, the test current signal and the test voltage signal of each charging stage are adaptively decomposed by Gaussian window sparse mode, which can accurately capture the subtle changes of battery state in different charging stages. This adaptability ensures that the charging strategy can flexibly respond to the real-time state of the battery and the changes in the external environment, thereby improving the intelligence and adaptability of the charging process. Through precise pattern quantity analysis and signal reconstruction, the charging control method can adjust the charging parameters and optimize the charging strategy in a targeted manner. This not only shortens the charging time, but also ensures the high efficiency of the charging process, while reducing energy loss and improving the overall charging efficiency. The current signal mode and the voltage signal mode are used to detect the abnormality of voltage output, and the abnormal conditions that may occur during the charging process are timely discovered and handled. This real-time monitoring and abnormality handling mechanism significantly enhances the safety of the charging process and effectively prevents overcharging, overheating, and other situations that may cause battery performance degradation or safety accidents. The charging control method avoids improper charging of the battery, such as overcharging or voltage and current mutations during the charging process, through the careful management of the charging process, thereby reducing the loss and aging of the battery and extending the service life of the battery. By optimizing and adjusting the charging strategy, the stability and reliability of the charging process are ensured, and the user's charging experience is improved. Especially in application scenarios such as mobile devices and electric vehicles, it can provide faster, safer and smarter charging services, meet users' high standards for charging efficiency and safety, and thus improve the accuracy of the charger's voltage output control.

[0090] The present application also provides a charger voltage output control device, the charger voltage output control device includes a memory and a processor, the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the charger voltage output control method in the above-mentioned embodiments.

[0091] The present application also provides a charger, which is used to execute the above-mentioned charger voltage output control method.

[0092] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0093] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0094] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for controlling voltage output of a charger, characterized in that: The voltage output control method of the charger includes: Initialize the charging strategy of the charger to obtain the initial charging strategy of each charging stage, and collect the test current signal and the test voltage signal of each charging stage; The test current signal and the test voltage signal are respectively subjected to adaptive Gaussian window sparse pattern decomposition to obtain the current signal pattern and the voltage signal pattern of each charging stage; specifically comprising: performing time series segmentation on the test current signal and the test voltage signal to obtain multiple initial current signal segments and multiple initial voltage signal segments; performing windowing processing on the multiple initial current signal segments and the multiple initial voltage signal segments by an adaptive Gaussian window function to obtain multiple windowed current signal segments and multiple windowed voltage signal segments; performing discrete Fourier transform on the multiple windowed current signal segments and the multiple windowed voltage signal segments to obtain multiple frequency domain current signal segments and multiple frequency domain voltage signal segments; performing pattern decomposition on the multiple frequency domain current signal segments by a sparse coding algorithm to obtain a sparse representation of the current signal, and performing pattern decomposition on the multiple frequency domain voltage signal segments by the sparse coding algorithm to obtain a sparse representation of the voltage signal; reconstructing the current signal pattern of each charging stage according to the sparse representation of the current signal, and reconstructing the voltage signal pattern of each charging stage according to the sparse representation of the voltage signal; According to the current signal pattern and the voltage signal pattern, voltage output anomaly detection is performed in each charging stage to obtain a target anomaly detection result in each charging stage; specifically, the following steps are performed: energy spectrum curve fitting is performed on the current signal pattern and the voltage signal pattern respectively through an iterative least squares algorithm to obtain a current energy spectrum curve and a voltage energy spectrum curve; feature extraction is performed on the current energy spectrum curve and the voltage energy spectrum curve respectively to obtain a current feature set and a voltage feature set; the current feature set and the voltage feature set are input into a preset anomaly detection model, and the anomaly detection model includes: an input layer, an attention mechanism layer, an autoencoder and an output layer; the input layer is used to respectively Performing one-hot vector encoding on the current feature set and the voltage feature set to obtain a current feature encoding vector and a voltage feature encoding vector; inputting the current feature encoding vector and the voltage feature encoding vector into the attention mechanism layer for vector fusion to obtain an attention feature encoding vector; inputting the attention feature encoding vector into the autoencoder for hidden feature extraction to obtain a hidden feature encoding vector; performing voltage output abnormality probability prediction on the hidden feature encoding vector through the ReLU function in the output layer to obtain a voltage output abnormality probability value for each charging stage, and outputting a target abnormality detection result for each charging stage according to the voltage output abnormality probability value; According to the target abnormality detection result, the current signal mode and the voltage signal mode are respectively analyzed for mode quantity to obtain the current mode quantity and the voltage mode quantity of each charging stage; specifically including: according to the target abnormality detection result, energy distribution calculation is performed on the current energy spectrum curve of the current signal mode to obtain the current energy distribution interval, and energy distribution calculation is performed on the voltage energy spectrum curve of the voltage signal mode to obtain the voltage energy distribution interval; spectrum analysis is performed on the current energy distribution interval to obtain the current frequency component of the current signal mode, and spectrum analysis is performed on the voltage energy distribution interval to obtain the voltage frequency component of the voltage signal mode; cluster analysis is performed on the current frequency component to obtain the current frequency classification result, and cluster analysis is performed on the voltage frequency component to obtain the voltage frequency classification result; energy contribution calculation is performed on the current frequency classification result to obtain the current frequency signal proportion, and energy contribution calculation is performed on the voltage frequency classification result to obtain the voltage frequency signal proportion; the number of current modes in each charging stage is determined according to the current frequency signal proportion, and the number of voltage modes in each charging stage is determined according to the voltage frequency signal proportion; According to the number of current modes and the number of voltage modes, the test current signal and the test voltage signal are respectively reconstructed to obtain the target current signal and the target voltage signal; specifically comprising: inputting the number of current modes and the test current signal into a preset first signal reconstruction model, the first signal reconstruction model comprising: a plurality of first convolutional pooling layers, a plurality of first fully connected layers, a plurality of first LSTM layers and a first signal merging layer; performing convolution feature operations on the number of current modes and the test current signal respectively through the plurality of first convolutional pooling layers to obtain a plurality of current convolution features, and performing feature dimensionality reduction processing on the plurality of current convolution features respectively through the plurality of first fully connected layers to obtain a plurality of current dimensionality reduction features, performing current time series fusion on the plurality of current dimensionality reduction features through the plurality of first LSTM layers to obtain a plurality of time series current signals, and The first signal merging layer performs signal reconstruction on the multiple time-series current signals to obtain a target current signal; the voltage mode number and the test voltage signal are input into a preset second signal reconstruction model, the second signal reconstruction model includes: multiple second convolution pooling layers, multiple second fully connected layers, multiple second LSTM layers and a second signal merging layer; convolution feature operations are performed on the voltage mode number and the test voltage signal through the multiple second convolution pooling layers to obtain multiple voltage convolution features, and feature dimension reduction processing is performed on the multiple voltage convolution features through the multiple second fully connected layers to obtain multiple voltage dimension reduction features, and voltage time series fusion is performed on the multiple voltage dimension reduction features through the multiple second LSTM layers to obtain multiple time series voltage signals, and signal reconstruction is performed on the multiple time series voltage signals through the second signal merging layer to obtain a target voltage signal; The voltage output analysis and charging strategy adjustment are performed on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

2. The voltage output control method of a charger according to claim 1, characterized in that: The step of initializing the charging strategy of the charger, obtaining the initial charging strategy of each charging stage, and collecting the test current signal and the test voltage signal of each charging stage includes: Define multiple charging stages of the charger, the multiple charging stages include: a constant current charging stage, a constant voltage charging stage and a trickle charging stage, wherein when the battery voltage is less than a first preset value, the charging stage is determined to be a constant current charging stage, when the first preset value is less than the battery voltage and less than a second preset value, the charging stage is determined to be a constant voltage charging stage, when the second preset value is less than the battery voltage, the charging stage is determined to be a trickle charging stage, and the first preset value is less than the second preset value; Performing voltage output control parameter analysis on the multiple charging stages respectively to obtain a voltage output control parameter range for each charging stage; Initializing charging strategies for multiple charging stages of the charger according to the voltage output control parameter range to obtain an initial charging strategy for each charging stage; Based on the initial charging strategy of each charging stage, a voltage output control test is performed on the charger, and an original current signal and an original voltage signal of each charging stage are collected; The original current signal and the original voltage signal are respectively subjected to direct current component removal and filtering processing to obtain an original current signal and an original voltage signal in each charging stage.

3. The voltage output control method of a charger according to claim 1, characterized in that: The performing voltage output analysis and charging strategy adjustment on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage includes: Performing voltage output analysis on the target current signal and the target voltage signal through a digital-to-analog converter to obtain voltage output adjustment control parameters for each charging stage; Controlling the voltage output of the charger according to the voltage output adjustment control parameter, and collecting feedback parameters of the charger to obtain a feedback parameter data set for each charging stage; The strategy adjustment gradient of each charging stage is calculated according to the feedback parameter data set, and the charging strategy is adjusted for the initial charging strategy of each charging stage according to the strategy adjustment gradient to obtain the target charging strategy of each charging stage.

4. A voltage output control device for a charger, characterized in that: The voltage output control device of the charger comprises: The acquisition module is used to initialize the charging strategy of the charger, obtain the initial charging strategy of each charging stage, and collect the test current signal and the test voltage signal of each charging stage; A decomposition module, used to perform adaptive Gaussian window sparse pattern decomposition on the test current signal and the test voltage signal respectively, to obtain the current signal pattern and the voltage signal pattern of each charging stage; specifically comprising: performing time series segmentation on the test current signal and the test voltage signal respectively, to obtain multiple initial current signal segments and multiple initial voltage signal segments; performing windowing processing on the multiple initial current signal segments and the multiple initial voltage signal segments respectively through an adaptive Gaussian window function, to obtain multiple windowed current signal segments and multiple windowed voltage signal segments; performing discrete Fourier transform on the multiple windowed current signal segments and the multiple windowed voltage signal segments respectively, to obtain multiple frequency domain current signal segments and multiple frequency domain voltage signal segments; performing pattern decomposition on the multiple frequency domain current signal segments through a sparse coding algorithm, to obtain a sparse representation of the current signal, and performing pattern decomposition on the multiple frequency domain voltage signal segments through the sparse coding algorithm, to obtain a sparse representation of the voltage signal; reconstructing the current signal pattern of each charging stage according to the sparse representation of the current signal, and reconstructing the voltage signal pattern of each charging stage according to the sparse representation of the voltage signal; The detection module is used to perform voltage output anomaly detection for each charging stage according to the current signal mode and the voltage signal mode to obtain a target anomaly detection result for each charging stage; specifically comprising: performing energy spectrum curve fitting on the current signal mode and the voltage signal mode respectively through an iterative least squares algorithm to obtain a current energy spectrum curve and a voltage energy spectrum curve; performing feature extraction on the current energy spectrum curve and the voltage energy spectrum curve respectively to obtain a current feature set and a voltage feature set; inputting the current feature set and the voltage feature set into a preset anomaly detection model, the anomaly detection model comprising: an input layer, an attention mechanism layer, an autoencoder and an output layer; The input layer performs one-hot vector encoding on the current feature set and the voltage feature set respectively to obtain a current feature encoding vector and a voltage feature encoding vector; the current feature encoding vector and the voltage feature encoding vector are input into the attention mechanism layer for vector fusion to obtain an attention feature encoding vector; the attention feature encoding vector is input into the autoencoder for hidden feature extraction to obtain a hidden feature encoding vector; the hidden feature encoding vector is predicted for voltage output abnormality probability through the ReLU function in the output layer to obtain a voltage output abnormality probability value for each charging stage, and the target abnormality detection result for each charging stage is output according to the voltage output abnormality probability value; An analysis module, used to perform mode quantity analysis on the current signal mode and the voltage signal mode respectively according to the target abnormality detection result to obtain the number of current modes and the number of voltage modes in each charging stage; specifically including: performing energy distribution calculation on the current energy spectrum curve of the current signal mode according to the target abnormality detection result to obtain the current energy distribution interval, and performing energy distribution calculation on the voltage energy spectrum curve of the voltage signal mode to obtain the voltage energy distribution interval; performing spectrum analysis on the current energy distribution interval to obtain the current frequency component of the current signal mode, and performing spectrum analysis on the voltage energy distribution interval to obtain the voltage frequency component of the voltage signal mode; performing cluster analysis on the current frequency component to obtain the current frequency classification result, and performing cluster analysis on the voltage frequency component to obtain the voltage frequency classification result; performing energy contribution calculation on the current frequency classification result to obtain the current frequency signal proportion, and performing energy contribution calculation on the voltage frequency classification result to obtain the voltage frequency signal proportion; determining the number of current modes in each charging stage according to the current frequency signal proportion, and determining the number of voltage modes in each charging stage according to the voltage frequency signal proportion; A reconstruction module is used to reconstruct the test current signal and the test voltage signal according to the number of current modes and the number of voltage modes, respectively, to obtain a target current signal and a target voltage signal; specifically comprising: inputting the number of current modes and the test current signal into a preset first signal reconstruction model, the first signal reconstruction model comprising: a plurality of first convolutional pooling layers, a plurality of first fully connected layers, a plurality of first LSTM layers and a first signal merging layer; performing convolution feature operations on the number of current modes and the test current signal respectively through the plurality of first convolutional pooling layers to obtain a plurality of current convolution features, and performing feature dimensionality reduction processing on the plurality of current convolution features respectively through the plurality of first fully connected layers to obtain a plurality of current dimensionality reduction features, performing current time series fusion on the plurality of current dimensionality reduction features through the plurality of first LSTM layers to obtain a plurality of time series current signals, and The first signal merging layer performs signal reconstruction on the multiple time-series current signals to obtain a target current signal; the voltage mode quantity and the test voltage signal are input into a preset second signal reconstruction model, the second signal reconstruction model includes: multiple second convolution pooling layers, multiple second fully connected layers, multiple second LSTM layers and a second signal merging layer; the multiple second convolution pooling layers respectively perform convolution feature operations on the voltage mode quantity and the test voltage signal to obtain multiple voltage convolution features, and the multiple second fully connected layers respectively perform feature dimension reduction processing on the multiple voltage convolution features to obtain multiple voltage dimension reduction features, the multiple second LSTM layers perform voltage time series fusion on the multiple voltage dimension reduction features to obtain multiple time series voltage signals, and the second signal merging layer performs signal reconstruction on the multiple time series voltage signals to obtain a target voltage signal; The adjustment module is used to perform voltage output analysis and charging strategy adjustment on the initial charging strategy of each charging stage according to the target current signal and the target voltage signal to obtain the target charging strategy of each charging stage.

5. A voltage output control device for a charger, characterized in that: The voltage output control device of the charger comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instruction in the memory to enable the voltage output control device of the charger to execute the voltage output control method of the charger according to any one of claims 1 to 3.

6. A charger, characterized in that: The charger is used to implement the voltage output control method of the charger according to any one of claims 1 to 3.

Images