LED driving power supply test method and system

Abstract

The invention provides an LED driving power supply testing method and system, and relates to the technical field of power supply testing, and the method comprises the steps: inputting a testing signal to a to-be-tested LED driving power supply, and collecting an output signal on a corresponding LED load; performing wavelet transform denoising on the output signal to obtain a denoised signal; performing feature extraction on the de-noised signal to obtain a signal feature; and inputting the signal characteristics into a trained power supply test evaluation model to obtain a test result. According to the invention, through improvement of test precision, automatic evaluation and multi-dimensional feature analysis, the test capability is enhanced, the test efficiency and cost control are promoted, and the method has significant practical value and market application potential.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply testing, and in particular to a method and system for testing an LED driving power supply. Background Art

[0002] With the global emphasis on energy conservation and environmental protection, LED lighting technology has developed rapidly. LED lamps are gradually replacing traditional lighting methods with their advantages of high efficiency, energy saving, long life and environmental protection. The widespread application of LED lamps has put forward higher requirements for their driving power supply, and the performance of the driving power supply directly affects the overall performance of LED lamps.

[0003] LED driver power supply is the core component of LED lighting system. LED needs constant current drive, and the driver power supply is responsible for providing stable current to ensure the normal operation of LED. LED driver power supply converts AC power supply into DC voltage suitable for LED to meet the working voltage requirements of LED. In addition, LED driver power supply can also provide over-current, over-voltage and short-circuit protection to prevent LED damage and extend its service life.

[0004] Traditional LED driver power supply testing methods have some limitations. Traditional testing usually relies on manual operation, which is inefficient and prone to errors. Existing testing equipment is often complex and expensive, and is not suitable for small and medium-sized enterprises. In addition, traditional technologies lack effective data analysis tools, making it difficult to fully evaluate the performance and reliability of the driver power supply. Summary of the invention

[0005] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a LED driver power supply testing method and system, which not only enhances the testing capability but also promotes testing efficiency and cost control by improving test accuracy, automated evaluation and multi-dimensional feature analysis, and has significant practical value and market application potential.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for testing an LED driving power supply, comprising:

[0008] Input a test signal to the LED driver to be tested, and collect the output signal on the corresponding LED load;

[0009] Performing wavelet transform denoising on the output signal to obtain a denoised signal;

[0010] Performing feature extraction on the denoised signal to obtain signal features;

[0011] The signal characteristics are input into a trained power supply test evaluation model to obtain a test result.

[0012] Preferably, the test signal includes: input voltage, input frequency, input power and input waveform.

[0013] Preferably, performing wavelet transform denoising on the output signal to obtain a denoised signal comprises:

[0014] Shifting the output signal to obtain a shifted signal;

[0015] Performing wavelet decomposition on the translation signal to obtain a plurality of wavelet coefficients;

[0016] Determine the filtering threshold according to the decomposition scale and length of the signal;

[0017] Constructing a denoising function according to the filtering threshold;

[0018] The denoising function is used to remove noise from the translation signal to obtain the denoised signal.

[0019] Preferably, the formula of the filtering threshold is:

[0020]

[0021] Among them, λ is the filtering threshold, j is the decomposition scale of the translation signal, and d j To decompose the wavelet coefficients with a scale of j, N represents the length of the translation signal, and median represents the median operation.

[0022] Preferably, removing the noise of the translation signal by using the denoising function to obtain the denoised signal comprises:

[0023] The corresponding wavelet coefficients are removed by using the denoising function to obtain smoothed signals corresponding to each test signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0024] Performing inverse translation on the smoothed signal to obtain respective inverse translation signals;

[0025] An average of a plurality of inverse translation signals is obtained to obtain the denoised signal.

[0026] Preferably, feature extraction is performed on the denoised signal to obtain signal features, including:

[0027] Performing normalization processing on the denoised signal to obtain a normalized signal;

[0028] Segmenting the normalized signal to obtain multiple segments of signals to be extracted;

[0029] Performing time domain extraction and frequency domain extraction on each segment of the signal to be extracted to obtain features to be processed; the features to be processed include: signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency characteristics;

[0030] Feature selection, dimension reduction and formatting are performed on the features to be processed to obtain the signal features.

[0031] Preferably, the method for constructing the power supply test evaluation model includes:

[0032] Obtain a preset feature data set; the feature data set includes signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency feature data and corresponding annotation evaluation results;

[0033] Performing data enhancement on the feature data set to obtain an enhanced data set;

[0034] Construct an initial convolutional neural network;

[0035] The initial convolutional neural network is trained according to the feature data sets of each signal type to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, a fifth classifier and a sixth classifier;

[0036] Cascading the trained first classifier, second classifier, third classifier, fourth classifier, fifth classifier and sixth classifier to obtain a classification network;

[0037] The LSTM neural network and the fully connected layer are sequentially connected after the classification network to obtain the power supply test evaluation model.

[0038] Preferably, the loss function of the power supply test evaluation model is The formula is:

[0039]

[0040] in, is the classification loss, Among them, y i,c is the true label, is the predicted output, C is the number of classification categories, and N is the number of samples; is the regression loss, Among them, f(x j ) is the predicted value, t j is the true value of the feature, M is the number of regression features; is the sparse loss; ω kis the weight parameter of the model, P is the total number of weight parameters, α, β, and γ are the weight coefficients of classification loss, regression loss, and sparse loss, respectively.

[0041] An LED driving power supply test system, comprising:

[0042] A signal input module is used to input a test signal to the LED driving power supply to be tested and collect an output signal on the corresponding LED load;

[0043] A denoising module, used for performing wavelet transform denoising on the output signal to obtain a denoised signal;

[0044] A feature extraction module, used to extract features from the denoised signal to obtain signal features;

[0045] The test module is used to input the signal characteristics into a trained power supply test evaluation model to obtain a test result.

[0046] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0047] The present invention provides a method and system for testing an LED driving power supply, the method comprising: inputting a test signal to an LED driving power supply to be tested, and collecting an output signal on a corresponding LED load; performing wavelet transform denoising on the output signal to obtain a denoised signal; performing feature extraction on the denoised signal to obtain a signal feature; and inputting the signal feature into a trained power supply test evaluation model to obtain a test result. The present invention not only enhances the test capability, but also promotes the test efficiency and cost control by improving the test accuracy, automated evaluation and multi-dimensional feature analysis, and has significant practical value and market application potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 A flow chart of a method provided by an embodiment of the present invention;

[0050] Figure 2 A schematic diagram of the system structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] The object of the present invention is to provide a method and system for testing an LED driving power supply, which can enhance the testing capability and promote the testing efficiency and cost control.

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] Figure 1 A flow chart of a method provided by an embodiment of the present invention, such as Figure 1 As shown, the present invention provides a method for testing an LED driving power supply, comprising:

[0055] Step 100: input a test signal to the LED driving power supply to be tested, and collect an output signal on a corresponding LED load;

[0056] Step 200: performing wavelet transform denoising on the output signal to obtain a denoised signal;

[0057] Step 300: extracting features from the denoised signal to obtain signal features;

[0058] Step 400: Input the signal feature into a trained power supply test evaluation model to obtain a test result.

[0059] Preferably, the test signal includes: input voltage, input frequency, input power and input waveform.

[0060] Specifically, the input test signal of this embodiment refers to an electrical signal applied to the input terminal of the LED driving power supply, including:

[0061] Input voltage, type: alternating current (AC) or direct current (DC) voltage, depending on the specifications of the driver power supply, the input voltage is within a specific range (e.g., AC 100-240V).

[0062] The input frequency (applicable only to AC power), that is, the frequency of the AC power, is usually 50 Hz or 60 Hz. Since the change of frequency may affect the performance of the driving power supply, the input frequency needs to be considered in this embodiment.

[0063] Input power refers to the power characteristics of the input signal, usually measured in watts (W).

[0064] Input waveform. For some tests, the waveform of the input signal (such as sine wave, square wave, etc.) may be considered to evaluate the performance of the driver under different waveforms.

[0065] Furthermore, the hardware components of this embodiment include:

[0066] The input module is used to receive the input signal of the LED driver to be tested. The input module is a programmable AC power supply or DC power supply that can simulate different input conditions (such as voltage, frequency, etc.) to test the performance of the driver under various working environments. According to the test requirements, the input module provides a preset test signal to ensure that the driver is turned on under accurate working conditions.

[0067] The output module is connected to the LED load and is used to measure the output voltage and current. The output module directly connects the output of the LED driver to be tested to the LED load to simulate the actual use environment. This embodiment can change the working state of the LED through an adjustable load, thereby obtaining performance data under different load conditions.

[0068] The control module, composed of a microcontroller or FPGA, is used to control the test process. The control module is responsible for coordinating the behavior of each hardware component to ensure the accurate input of the test signal and the collection of data. This embodiment dynamically adjusts the input signal and reads the output data according to the preset test program, and controls when to start or stop the test process. And the control module of this embodiment can process feedback information from other modules to optimize the test process.

[0069] The measurement module, including voltage and current sensors, is used for data acquisition. The measurement module monitors and collects the output voltage and current data on the LED load in real time. After the data is denoised by wavelet transform, it will be used for feature extraction to ensure the accuracy and reliability of the data. This module has high precision to adapt to the sensitive characteristics of the LED drive power supply, ensuring the real-time and accuracy of data acquisition.

[0070] The communication module is used to communicate with a computer or other device to transmit test data. The communication module transmits the data collected by the measurement module to a computer or remote server via USB, Ethernet or wireless signals. This makes real-time monitoring, storage and subsequent data analysis of test data feasible, facilitating users to conduct further result analysis and report generation.

[0071] Furthermore, the test process of this embodiment includes:

[0072] 1. Test preparation: start the input module, set the parameters of the input signal, such as voltage and frequency, and then connect the LED load to the output module.

[0073] 2. Signal input and data acquisition: the control module sends a signal, and the test input module inputs the set test signal into the LED driver power supply. The measurement module monitors the output voltage and current of the LED load in real time and transmits data through the communication module.

[0074] 3. Signal processing: transfer the collected output signal to the data processing system for wavelet transform denoising. Extract feature information from the denoised signal to form a signal feature data set.

[0075] 4. Model evaluation: Input the extracted signal features into the trained power supply test evaluation model to obtain the final test results. The test results, including efficiency, power factor, harmonic distortion, etc., are displayed through the software interface, and a detailed test report is generated.

[0076] Furthermore, this embodiment can achieve comprehensive and accurate testing of the LED driver power supply through the synergistic effect of the above hardware components. The structured testing method of this embodiment can ensure the effective execution of each step, thereby improving the reliability and efficiency of the overall test.

[0077] Preferably, performing wavelet transform denoising on the output signal to obtain a denoised signal comprises:

[0078] Shifting the output signal to obtain a shifted signal;

[0079] Performing wavelet decomposition on the translation signal to obtain a plurality of wavelet coefficients;

[0080] Determine the filtering threshold according to the decomposition scale and length of the signal;

[0081] Constructing a denoising function according to the filtering threshold;

[0082] The denoising function is used to remove noise from the translation signal to obtain the denoised signal.

[0083] Preferably, the formula of the filtering threshold is:

[0084]

[0085] Among them, λ is the filtering threshold, j is the decomposition scale of the translation signal, and d j To decompose the wavelet coefficients with a scale of j, N represents the length of the translation signal, and median represents the median operation.

[0086] Specifically, in the wavelet threshold denoising process, the quality of threshold selection directly affects the denoising effect. Since the wavelet coefficient amplitude of noise has the characteristic of decreasing with the increase of decomposition scale, this embodiment introduces the decomposition scale to adaptively adjust the filtering threshold.

[0087] Preferably, removing the noise of the translation signal by using the denoising function to obtain the denoised signal comprises:

[0088] The corresponding wavelet coefficients are removed by using the denoising function to obtain smoothed signals corresponding to each test signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0089] Performing inverse translation on the smoothed signal to obtain respective inverse translation signals;

[0090] An average of a plurality of inverse translation signals is obtained to obtain the denoised signal.

[0091] Furthermore, commonly used threshold functions include hard threshold function and soft threshold function. Due to the discontinuity of the contraction function, the hard threshold function will produce some artificial noise points in the restored signal, causing the signal to oscillate; the soft threshold function shrinks the wavelet coefficients greater than the threshold. Although the processed wavelet coefficients have good continuity, some useful high-frequency information will be lost. Based on this, this embodiment proposes the above denoising function to address the shortcomings of the hard threshold function and the soft threshold function.

[0092] Furthermore, this embodiment uses wavelet coefficients to measure the attenuation deviation, so that the attenuation of wavelet coefficients with larger absolute values ​​decreases as their absolute values ​​increase, thereby effectively avoiding the loss of high-frequency information and improving the signal-to-noise ratio of the signal.

[0093] Preferably, feature extraction is performed on the denoised signal to obtain signal features, including:

[0094] Performing normalization processing on the denoised signal to obtain a normalized signal;

[0095] Segmenting the normalized signal to obtain multiple segments of signals to be extracted;

[0096] Performing time domain extraction and frequency domain extraction on each segment of the signal to be extracted to obtain features to be processed; the features to be processed include: signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency characteristics;

[0097] Feature selection, dimension reduction and formatting are performed on the features to be processed to obtain the signal features.

[0098] Specifically, feature extraction of denoised signals includes multiple steps. First, the denoised signal is normalized to eliminate the influence of different signal amplitudes, usually using minimum-maximum normalization or Z-score normalization. Next, the normalized signal is divided into multiple small segments, and the window size and step size are defined using sliding window technology so that each segment of the signal can be locally analyzed. Then, time domain and frequency domain feature extraction are performed on each segment of the signal respectively. In time domain feature extraction, the mean, variance, kurtosis and skewness of the signal are calculated to reflect the basic statistical characteristics of the signal; in frequency domain feature extraction, the frequency components of the signal are analyzed using fast Fourier transform (FFT), the energy distribution of each frequency component is calculated, and the main frequency features are identified. After feature extraction is completed, feature selection and dimensionality reduction are performed, and methods such as correlation analysis or principal component analysis (PCA) are used to screen out the most representative features and reduce the feature dimensions. Finally, the processed features are organized into feature vectors in a unified format so that they can be input into subsequent evaluation models for analysis and prediction.

[0099] Preferably, the method for constructing the power supply test evaluation model includes:

[0100] Obtain a preset feature data set; the feature data set includes signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency feature data and corresponding annotation evaluation results;

[0101] Performing data enhancement on the feature data set to obtain an enhanced data set;

[0102] Construct an initial convolutional neural network;

[0103] The initial convolutional neural network is trained according to the feature data sets of each signal type to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, a fifth classifier and a sixth classifier;

[0104] Cascading the trained first classifier, second classifier, third classifier, fourth classifier, fifth classifier and sixth classifier to obtain a classification network;

[0105] The LSTM neural network and the fully connected layer are sequentially connected after the classification network to obtain the power supply test evaluation model.

[0106] Specifically, the construction process of the power supply test evaluation model first starts with obtaining a preset feature data set, which contains important features such as signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency characteristics, as well as corresponding annotated evaluation results. These labels are used to supervise model training. Next, data enhancement techniques such as rotation, translation, scaling and adding noise are applied to expand the feature data set to improve the generalization ability of the model and obtain an enhanced data set. After that, an initial convolutional neural network (CNN) is constructed, and the appropriate number of network layers and filter sizes are designed to adapt to the dimensions and characteristics of the feature data. According to the different signal types, each signal feature data set is input into the initial CNN for training, and multiple trained classifiers are obtained, including the first to sixth classifiers, corresponding to different types of signals. These classifiers are cascaded to form a comprehensive classification network to make full use of the pattern information learned by each classifier. On this basis, a long short-term memory (LSTM) neural network and a fully connected layer are added in sequence after the classification network to model the time series features and strengthen the network's ability to capture dynamic changes, and finally a complete power supply test evaluation model is obtained, which is used to accurately evaluate and classify the performance of different LED driver power supplies. The model construction process of this embodiment emphasizes the effective combination of feature extraction and learning, model training and optimization, aiming to achieve intelligent evaluation of power supply performance.

[0107] Furthermore, the results output by the model of this embodiment include key information such as the efficiency, power factor, total harmonic distortion (THD), current waveform distortion, stability of output voltage and current, transient response characteristics, etc. of the power supply. In addition, the model will also classify the power supply according to the input signal characteristics and provide classification confidence values ​​to reflect the adaptability to different signal types. These results support users in fault diagnosis, performance optimization and reliability analysis, making the use of power supplies safer and more reliable. At the same time, the evaluation results can also provide feedback for product improvement and design, helping engineers to formulate corresponding technical solutions and optimize the design and application of LED driver power supplies.

[0108] Preferably, the loss function of the power supply test evaluation model is The formula is:

[0109]

[0110] in, is the classification loss, Among them, y i,c is the true label, is the predicted output, C is the number of classification categories, and N is the number of samples; is the regression loss, Among them, f(xj ) is the predicted value, t j is the true value of the feature, M is the number of regression features; is the sparse loss; ω k is the weight parameter of the model, P is the total number of weight parameters, α, β, and γ are the weight coefficients of classification loss, regression loss, and sparse loss, respectively.

[0111] Optionally, the loss function of the model of this embodiment is designed as a combined loss, integrating classification error and regression error, to ensure that the model can be optimized in terms of precision and accuracy of signal feature prediction:

[0112]

[0113] Classification loss is used to process the classification accuracy of different signal feature types in the evaluation model, using cross entropy loss:

[0114]

[0115] Here, y i,c is the true label, is the predicted output, C is the number of classification categories, and N is the number of samples.

[0116] Regression loss, used to ensure accurate estimation of extracted signal features (such as spectral energy, etc.), uses mean square error (MSE):

[0117]

[0118] Here, f(x j ) is the predicted value, t j is the true value of the feature, and M is the number of regression features.

[0119] Sparse loss, introduce L1 regularization to enhance feature selection effect and reduce the influence of noise features:

[0120]

[0121] Here, ω k is the weight parameter of the model, and P is the total number of weight parameters.

[0122] Corresponding to the above method, this embodiment also provides a LED driving power supply testing system, such as Figure 2 As shown, including:

[0123] A signal input module is used to input a test signal to the LED driving power supply to be tested and collect an output signal on the corresponding LED load;

[0124] A denoising module, used for performing wavelet transform denoising on the output signal to obtain a denoised signal;

[0125] A feature extraction module, used to extract features from the denoised signal to obtain signal features;

[0126] The test module is used to input the signal characteristics into a trained power supply test evaluation model to obtain a test result.

[0127] The beneficial effects of the present invention are as follows:

[0128] (1) The present invention uses wavelet transform to denoise the output signal, which can effectively filter out high-frequency noise and interference, retain the main features of the signal, and thus improve the accuracy of signal feature extraction.

[0129] (2) The present invention obtains a variety of important signal features including mean, variance, spectrum characteristics, etc. by performing detailed feature extraction on the denoised signal, so that the test results are more comprehensive and accurate, and the ability to evaluate the performance of the LED driver power supply is improved.

[0130] (3) The present invention inputs the extracted signal features into a trained power supply test evaluation model, which can realize automated evaluation, reduce manual intervention, improve test efficiency, and reduce the possibility of human error.

[0131] (4) The present invention can adapt to different types of LED driving power supplies and improve the system's testing capabilities for diversified products by constructing classifiers for different signals.

[0132] (5) The present invention can generate test results of multiple key performance indicators including efficiency, power factor, harmonic distortion, etc. through comprehensive analysis of extracted features, and comprehensively evaluate the working status and performance of the LED driver power supply.

[0133] (6) The automated testing process of the present invention reduces the reliance on professionals and improves the efficiency of testing, thereby reducing testing time and economic costs, and is particularly suitable for popularization and application in small and medium-sized enterprises.

[0134] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0135] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for testing an LED driving power supply, characterized in that: include: Input a test signal to the LED driver to be tested, and collect the output signal on the corresponding LED load; Performing wavelet transform denoising on the output signal to obtain a denoised signal; Performing feature extraction on the denoised signal to obtain signal features; The signal characteristics are input into a trained power supply test evaluation model to obtain a test result.

2. The LED driving power supply testing method according to claim 1, characterized in that: The test signal includes: input voltage, input frequency, input power and input waveform.

3. The LED driving power supply testing method according to claim 1, characterized in that: Performing wavelet transform denoising on the output signal to obtain a denoised signal, comprising: Shifting the output signal to obtain a shifted signal; Performing wavelet decomposition on the translation signal to obtain a plurality of wavelet coefficients; Determine the filtering threshold according to the decomposition scale and length of the signal; Constructing a denoising function according to the filtering threshold; The denoising function is used to remove noise from the translation signal to obtain the denoised signal.

4. The LED driving power supply testing method according to claim 3, characterized in that: The formula for the filtering threshold is: Among them, λ is the filtering threshold, j is the decomposition scale of the translation signal, and d j To decompose the wavelet coefficients with a scale of j, N represents the length of the translation signal, and median represents the median operation.

5. The LED driving power supply testing method according to claim 4, characterized in that: Using the denoising function to remove the noise of the translation signal to obtain the denoised signal includes: The corresponding wavelet coefficients are removed by using the denoising function to obtain smoothed signals corresponding to each test signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient; Performing inverse translation on the smoothed signal to obtain respective inverse translation signals; An average of a plurality of inverse translation signals is obtained to obtain the denoised signal.

6. The LED driving power supply testing method according to claim 1, characterized in that: Performing feature extraction on the denoised signal to obtain signal features includes: Performing normalization processing on the denoised signal to obtain a normalized signal; Segmenting the normalized signal to obtain multiple segments of signals to be extracted; Performing time domain extraction and frequency domain extraction on each segment of the signal to be extracted to obtain features to be processed; the features to be processed include: signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency characteristics; Feature selection, dimension reduction and formatting are performed on the features to be processed to obtain the signal features.

7. The LED driving power supply testing method according to claim 6, characterized in that: The method for constructing the power supply test evaluation model includes: Obtain a preset feature data set; the feature data set includes signal mean, signal variance, signal kurtosis, signal skewness, energy distribution of different frequency components and main frequency feature data and corresponding annotation evaluation results; Performing data enhancement on the feature data set to obtain an enhanced data set; Construct an initial convolutional neural network; The initial convolutional neural network is trained according to the feature data sets of each signal type to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, a fifth classifier and a sixth classifier; Cascading the trained first classifier, second classifier, third classifier, fourth classifier, fifth classifier and sixth classifier to obtain a classification network; The LSTM neural network and the fully connected layer are sequentially connected after the classification network to obtain the power supply test evaluation model.

8. The LED driving power supply testing method according to claim 7, characterized in that: The power test evaluates the loss function of the model The formula is: in, is the classification loss, Among them, y i,c is the true label, is the predicted output, C is the number of classification categories, and N is the number of samples; is the regression loss, Among them, f(x j ) is the predicted value, t j is the true value of the feature, M is the number of regression features; is the sparse loss; ω k is the weight parameter of the model, P is the total number of weight parameters, α, β, and γ are the weight coefficients of classification loss, regression loss, and sparse loss, respectively.

9. A LED drive power supply test system, characterized in that: include: A signal input module is used to input a test signal to the LED driving power supply to be tested and collect an output signal on the corresponding LED load; A denoising module, used for performing wavelet transform denoising on the output signal to obtain a denoised signal; A feature extraction module, used to extract features from the denoised signal to obtain signal features; The test module is used to input the signal characteristics into a trained power supply test evaluation model to obtain a test result.

Images