Configuration Method, Device, Equipment and Storage Medium of LED Driver Chip

Abstract

The present application relates to the technical field of LED driver chips, and discloses a configuration method, device, equipment and storage medium for an LED driver chip. The method includes: performing power cycle tests and data collection on multiple LED driver chips in an LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driver chip; performing feature extraction and feature fusion to obtain an overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors; predicting the remaining useful life through a BP neural network model to obtain remaining useful life prediction data; and performing parameter configuration on the LED module according to the remaining useful life prediction data to generate a target chip parameter configuration strategy. The present application improves the accuracy of performance parameter configuration of the LED driver chip.

Description

Technical Field

[0001] The present application relates to the technical field of LED driver chips, and particularly relates to a configuration method, device, equipment and storage medium of an LED driver chip. Background Art

[0002] During the use of an LED driver chip, it will face challenges of high temperature and current load, and these conditions will accelerate the aging and performance degradation of the chip. Especially in the power cycle test, the periodic changes of temperature and current cause continuous thermal stress on the chip material, which will lead to unstable performance and shortened lifespan. At present, the research on how to accurately predict the remaining lifespan of an LED driver chip and adjust parameters according to the prediction data to extend the lifespan is not sufficient.

[0003] In addition, the existing testing methods for LED driver chips mainly focus on performance testing under static conditions, and lack a dynamic and real-time monitoring and adjustment mechanism. This limits the ability of the driver chip to maintain optimal performance in a changing working environment. In practical applications, due to the lack of effective prediction and adjustment mechanisms, the LED driver chip fails before reaching the theoretical lifespan, increasing the maintenance cost and affecting the user experience. Summary of the Invention

[0004] The present application provides a configuration method, device, equipment and storage medium of an LED driver chip, and the present application improves the accuracy of configuring performance parameters of the LED driver chip.

[0005] In a first aspect, the present application provides a configuration method of an LED driver chip, and the configuration method of the LED driver chip includes:

[0006] Conduct a power cycle test and data acquisition on multiple LED driver chips in an LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driver chip;

[0007] Extract and fuse features from the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively extract and fuse features from the local thermal resistance data and the local temperature data to obtain multiple local thermal resistance temperature feature vectors;

[0008] Input the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors into a preset BP neural network model for predicting the remaining service life to obtain remaining service life prediction data;

[0009] Configure parameters of the LED module according to the remaining service life prediction data to generate a target chip parameter configuration strategy.

[0010] In a second aspect, the present application provides a configuration device for an LED driving chip. The configuration device for the LED driving chip includes:

[0011] A test module, configured to perform power cycle testing and data acquisition on multiple LED driving chips in an LED module, so as to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driving chip;

[0012] An extraction module, configured to perform feature extraction and feature fusion on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively perform feature extraction and feature fusion on the local thermal resistance data and the local temperature data to obtain a plurality of local thermal resistance temperature feature vectors;

[0013] A prediction module, configured to input the overall thermal resistance temperature feature vector and the plurality of local thermal resistance temperature feature vectors into a preset BP neural network model for remaining service life prediction, so as to obtain remaining service life prediction data;

[0014] A configuration module, configured to perform parameter configuration on the LED module according to the remaining service life prediction data to generate a target chip parameter configuration strategy.

[0015] In a third aspect of the present application, a computer device is provided, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned configuration method for the LED driving chip.

[0016] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, and when the instructions are run on a computer, the computer is caused to execute the above-mentioned configuration method for the LED driving chip.

[0017] In the technical solution provided by this application, by performing power cycle tests and data acquisition on multiple driving chips in the LED module, thermal resistance and temperature data regarding the overall and local parts of the chips can be obtained. These data are crucial for understanding the performance of the chips under actual working conditions. This method can reveal in detail the thermal characteristics and potential failure modes of the chips under different working states, providing accurate basic data for subsequent data analysis and feature extraction. Through feature extraction and feature fusion technologies, key information helpful for predicting the chip lifespan can be refined from complex temperature and thermal resistance data. These processing procedures use statistical and machine learning technologies to enhance the information density of the data, making subsequent lifespan prediction more accurate and reliable. The obtained feature vectors accurately reflect the current state and possible future performance of the chips, making the lifespan prediction highly operable and practically valuable. Inputting the extracted feature vectors into a pre-set BP neural network model for lifespan prediction not only improves the prediction accuracy but also enables dynamic adjustment of chip parameters according to the prediction results, achieving the optimization of the performance of LED driving chips and the extension of their lifespan. This configuration strategy based on intelligent learning algorithms can effectively cope with variable usage conditions and environmental factors, ensuring the long-term stability and reliability of the LED driving system, and thereby improving the accuracy of the performance parameter configuration of LED driving chips. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 Schematic diagram of an embodiment of the configuration method of the LED driving chip in the embodiment of this application;

[0020] Figure 2 Schematic diagram of an embodiment of the configuration device of the LED driving chip in the embodiment of this application. Detailed Embodiments

[0021] The embodiments of the present application provide a configuration method, device, equipment and storage medium for an LED driving chip. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and the above-mentioned drawings of the present application are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the term "comprising" or "having" and any variation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or equipment comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or equipment.

[0022] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the configuration method of the LED driving chip in the embodiments of the present application includes:

[0023] Step 101, perform power cycle testing and data acquisition on multiple LED driving chips in the LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driving chip;

[0024] It can be understood that the execution subject of the present application can be a configuration device for an LED driving chip, or a terminal or a server, and specific limitations are not made here. The embodiments of the present application will be described by taking the server as the execution subject as an example.

[0025] Specifically, power cycle tests and data acquisition are performed on multiple LED driver chips in the LED module. The power cycle test of the LED module is carried out with preset current and voltage to ensure the consistency and accuracy of the test environment. By simulating the voltage and current changes of the LED module during actual use, the stress states under different working environments can also be simulated, so as to more accurately evaluate the performance and stability of the LED driver chip. During the test, the module thermal resistance data and module temperature data of the LED module are collected, and at the same time, the chip thermal resistance data and chip temperature data are collected separately for each LED driver chip. The collected data often contains various noises, which may be caused by the instability of the test environment, voltage fluctuations or other electronic interferences. Effective data denoising is performed on the module thermal resistance data and module temperature data, and outliers and interferences in the data are removed through filters or statistical methods to ensure data quality. Data interpolation is carried out to estimate the possible values of missing data based on the existing data points to ensure the integrity and continuity of the overall data. Similarly, data denoising and data interpolation are respectively performed on the chip thermal resistance data and chip temperature data of each LED driver chip. Through denoising and interpolation, more accurate and reliable local thermal resistance data and local temperature data are obtained.

[0026] Step 102: Extract features and fuse features from the overall thermal resistance data and overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively extract features and fuse features from the local thermal resistance data and local temperature data to obtain multiple local thermal resistance temperature feature vectors;

[0027] Specifically, by calculating the mean and standard deviation of the overall thermal resistance data and the overall temperature data, an overview data description is obtained. The mean provides the central position of the data, while the standard deviation describes the degree of dispersion of the data points around the mean. According to the calculated overall thermal resistance mean and standard deviation, multiple initial overall thermal resistance eigenvalue are extracted from the overall thermal resistance data. The initial eigenvalues can be selected through certain predetermined criteria or algorithms. For example, select those values that exceed or are lower than a certain multiple of the standard deviation to help the system identify extreme or abnormal data, which may imply potential performance problems or failure risks. The overall thermal resistance standard deviation is used to further screen these eigenvalues, retaining those eigenvalues that can represent the overall behavior of the system and excluding random fluctuations or atypical behaviors in the data. When processing the overall temperature data, multiple initial overall temperature eigenvalues are extracted based on the mean and standard deviation of the overall temperature data, and the overall temperature standard deviation is used to screen these eigenvalues to ensure that the finally obtained eigenvalues are representative and relevant. For the local thermal resistance and temperature data of each LED driver chip, the mean and standard deviation of the local thermal resistance and temperature data are calculated. Using the local thermal resistance mean and standard deviation, multiple initial local thermal resistance eigenvalues are extracted from the data of each chip, and these eigenvalues reflect the performance and stability of each chip under different working conditions. Screening is performed through the local thermal resistance standard deviation, and only those eigenvalues that best represent the characteristics of each chip are retained. For the local temperature data, the same eigenvalue extraction and screening method based on the mean and standard deviation is also used to ensure that the obtained eigenvectors are both comprehensive and specific. The target overall thermal resistance eigenvalues and overall temperature eigenvalues are subjected to weight calculation and vector mapping to generate an overall thermal resistance temperature eigenvector. This vector reflects the thermal performance and temperature performance of the LED module as a whole and is used for subsequent data analysis and system optimization. Similarly, weight calculation and vector mapping are performed on the target local thermal resistance eigenvalues and local temperature eigenvalues of each LED driver chip to generate multiple local thermal resistance temperature eigenvectors.

[0028] The overall coefficient of variation between multiple target overall thermal resistance eigenvalues ​​and multiple target overall temperature eigenvalues ​​is calculated according to the overall thermal resistance mean, the overall temperature mean, the overall thermal resistance standard deviation and the overall temperature standard deviation. The coefficient of variation is a statistical measure that describes the degree of data dispersion and reflects the degree of fluctuation of the eigenvalue relative to its mean. The corresponding first weight data is set according to the overall coefficient of variation. The setting of weight data directly affects the construction of subsequent eigenvectors and the prediction performance of the data model. The multiple target overall thermal resistance eigenvalues ​​and multiple overall temperature eigenvalues ​​are normalized, and the influence of the dimension is eliminated by adjusting the data scale to ensure that different data features can be processed at the same scale in the model, so as to obtain multiple normalized overall thermal resistance eigenvalues ​​and multiple normalized overall temperature eigenvalues. According to the set first weight data, the normalized overall thermal resistance eigenvalues ​​and overall temperature eigenvalues ​​are weighted. The weighted operation emphasizes the importance of certain features by weighted average or other means, so that the model pays more attention to those variables that have a decisive influence on the prediction target. After completing the weighted operation, the weighted eigenvalues ​​are merged into an overall thermal resistance temperature eigenvector by vector splicing. The coefficient of variation of the local eigenvalue is obtained by calculating the local thermal resistance mean, local temperature mean, local thermal resistance standard deviation and local temperature standard deviation. The local coefficient of variation provides an important indicator to measure the performance differences of each LED driver chip. Based on this coefficient, the corresponding second weight data is set for each chip. The setting of the weight takes into account the performance differences between each chip, so that the subsequent analysis can adjust the parameters of each chip in a targeted manner. The target local thermal resistance eigenvalue and local temperature eigenvalue are normalized and weighted by the second weight data. After the eigenvalues ​​of each LED driver chip are weighted, they are combined into multiple local thermal resistance temperature eigenvectors through vector splicing technology. The local eigenvector reflects the specific working status and performance of each chip.

[0029] Step 103, inputting the overall thermal resistance temperature characteristic vector and the multiple local thermal resistance temperature characteristic vectors into a preset BP neural network model to perform remaining service life prediction, and obtaining remaining service life prediction data;

[0030] Specifically, the overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors are input into a pre-set BP neural network model. This model has a multi-layer structure, including an input layer, multiple first hidden layers, a second hidden layer, an attention mechanism layer, and an output layer. The input layer receives the overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors respectively. The multiple first hidden layers process each local thermal resistance temperature feature vector. In the first hidden layer, each local feature vector undergoes multiple transformations and high-dimensional feature extractions. Through the weights and activation functions in the network, more abstract and information-rich high-dimensional feature vectors are extracted from the original local feature vectors. At the same time, through the second hidden layer, high-dimensional feature extraction is performed on the overall thermal resistance temperature feature vector to obtain an overall high-dimensional feature vector representing the overall performance state. The local high-dimensional feature vectors and the overall high-dimensional feature vector are input into the attention mechanism layer. The attention mechanism layer performs feature fusion by learning the relative importance between different features, enabling the network to focus more on the information most helpful for the prediction task. The remaining useful life is calculated for the attention fusion feature vector through the ReLU function in the output layer, obtaining the remaining useful life prediction data. The ReLU function ensures the non-linearity of the output and the effectiveness of the model, making the prediction results both stable and highly reliable.

[0031] The overall thermal resistance temperature feature vector is input into the second hidden layer, which is composed of three layers of unidirectionally connected neurons. Through the first layer of unidirectionally connected neurons, high-dimensional feature extraction is performed on the overall thermal resistance temperature feature vector. The neurons initially transform the input feature vector through weights and activation functions, extracting new high-dimensional features to form the first thermal resistance temperature feature vector. The first thermal resistance temperature feature vector is concatenated with the original overall thermal resistance temperature feature vector to generate the first concatenated feature vector. The original data and the high-dimensional data obtained after the first layer of processing are fused to enhance the model's ability to capture the correlations between features. The concatenated feature vector is input into the second layer of unidirectionally connected neurons for high-dimensional feature extraction to generate the second thermal resistance temperature feature vector. The second thermal resistance temperature feature vector is concatenated with the initial overall thermal resistance temperature feature vector to form the second concatenated feature vector. The second concatenated feature vector is input into the third layer of unidirectionally connected neurons for high-dimensional feature extraction to obtain the third thermal resistance temperature feature vector. Through the continuous processing of these three layers, new dimensions and depths are added to the original feature vector at each step, making the final feature vector rich in multi-level information and analysis depth. Weighted fusion is performed on the first, second, and third thermal resistance temperature feature vectors. The influence of each feature vector in the final output is adjusted through weight coefficients to ensure that the model can fully consider the unique contributions of the features extracted from different levels. Through weighted fusion, the results of multi-level feature extraction are integrated to generate an overall high-dimensional feature vector representing the thermal performance and reliability state of the entire LED driver chip.

[0032] Step 104: Configure the parameters of the LED module according to the remaining service life prediction data to generate a target chip parameter configuration strategy.

[0033] Specifically, initialize the parameter adjustment ranges of multiple LED driver chips according to the remaining service life prediction data, define the boundaries within which each chip parameter can be adjusted during subsequent optimization, ensuring the rationality and feasibility of the adjustment process. Based on the particle swarm optimization algorithm, perform an initial configuration of the parameters of the LED module to generate an initial chip parameter configuration strategy. The particle swarm optimization algorithm is an optimization tool based on swarm intelligence that searches for the optimal solution by simulating the social behavior of bird flocks and is suitable for dealing with complex optimization problems. The algorithm simulates each individual (i.e., particle) in a group exploring the optimal solution in the solution space. Each particle represents a possible chip parameter configuration strategy, and the position of the particle is iteratively updated to approach the optimal configuration. According to the parameter adjustment ranges of each LED driver chip, dynamically and randomly generate multiple first chip parameter configuration strategies. Calculate the fitness value of each generated configuration strategy, which reflects how well the strategy adapts to the current system requirements and operating conditions. Divide the multiple first chip parameter configuration strategies into groups according to the fitness values to form multiple target parameter configuration strategy groups. The grouping is based on the performance of the strategies, grouping strategies with similar performance together, aiming to optimize the subsequent processing efficiency and effect. Generate multiple second chip parameter configuration strategies based on the multiple target parameter configuration strategy groups. Perform parameter optimization solution on the second chip parameter configuration strategies to generate the final target chip parameter configuration strategy. The algorithm comprehensively considers various factors, such as energy consumption, performance, long-term stability, etc., to ensure that the final strategy can not only extend the service life of the LED driver chip but also maintain or improve its performance.

[0034] In the embodiments of the present application, by performing power cycle tests and data acquisition on multiple driver chips in the LED module, thermal resistance and temperature data of the overall and local parts of the chips can be obtained. These data are crucial for understanding the performance of the chips under actual working conditions. This method can reveal in detail the thermal characteristics and potential failure modes of the chips under different working states, providing accurate basic data for subsequent data analysis and feature extraction. Through feature extraction and feature fusion techniques, key information helpful for predicting the chip lifespan can be refined from the complex temperature and thermal resistance data. These processing procedures use statistical and machine learning techniques to enhance the information density of the data, making the subsequent lifespan prediction more accurate and reliable. The obtained feature vectors accurately reflect the immediate state and possible future performance of the chips, making the lifespan prediction highly operable and practically valuable. Inputting the extracted feature vectors into a preset BP neural network model for lifespan prediction not only improves the prediction accuracy but also enables dynamic adjustment of chip parameters according to the prediction results, realizing the optimization of the performance of LED driver chips and the extension of their lifespan. This configuration strategy based on intelligent learning algorithms can effectively cope with variable usage conditions and environmental factors, ensuring the long-term stability and reliability of the LED drive system, and thus improving the accuracy of the performance parameter configuration of LED driver chips.

[0035] In a specific embodiment, the process of executing step 101 may specifically include the following steps:

[0036] (1) Perform power cycle tests on the LED module using preset current and voltage. The LED module includes multiple LED driver chips;

[0037] (2) Collect the module thermal resistance data and module temperature data of the LED module, and respectively collect the chip thermal resistance data and chip temperature data of each LED driver chip;

[0038] (3) Perform data denoising and data interpolation on the module thermal resistance data and module temperature data respectively to obtain the overall thermal resistance data and overall temperature data of the LED module;

[0039] (4) Perform data denoising and data interpolation on the chip thermal resistance data and chip temperature data of each LED driver chip respectively to obtain the local thermal resistance data and local temperature data of each LED driver chip.

[0040] Specifically, a power cycle test is performed on an LED module containing multiple LED driver chips using a preset current and voltage. For example, the current is set to 350 milliamperes and the voltage is set to 3.3 volts to simulate the operating state of the LED under normal operating conditions and at the same time simulate the performance of the LED under different loads. Meanwhile, thermal resistance data and temperature data of the entire LED module are collected, as well as thermal resistance and temperature data for each individual LED driver chip in the module. Sensors are placed at key positions on the LED module and each LED driver chip to monitor the thermal behavior and temperature changes of each component under different operating conditions, thereby evaluating their thermal stability and potential overheating risks. The data collected usually contains a certain amount of noise, which may originate from the test environment, electronic noise of the device itself, or operational errors. Denoise this data. Denoising can be achieved through various techniques, such as using a low-pass filter or moving average method, to remove random fluctuations and outliers in the data, making the data smoother and more reliable. For example, if an abnormal fluctuation suddenly appears in the temperature data during the collection process, which may be caused by environmental interference or measurement error, the moving average method can effectively smooth the short-term fluctuations. Data interpolation is performed on the module thermal resistance data and the module temperature data. Data interpolation methods, such as linear interpolation or spline interpolation, can estimate missing data values between known data points to ensure the integrity and continuity of the data. Data denoising and data interpolation are respectively performed on the chip thermal resistance data and the chip temperature data of each LED driver chip to obtain local thermal resistance data and local temperature data for each LED driver chip. Local data processing enables the system to understand the specific performance and potential problems of each chip in more detail. For example, if the temperature of a certain chip continuously remains higher than that of other chips, this may indicate a defect or insufficient heat dissipation in that chip.

[0041] In a specific embodiment, the process of performing step 102 may specifically include the following steps:

[0042] (1) Calculate the mean values of the overall thermal resistance data and the overall temperature data respectively to obtain the overall thermal resistance mean value and the overall temperature mean value, and calculate the standard deviations of the overall thermal resistance data and the overall temperature data respectively to obtain the overall thermal resistance standard deviation and the overall temperature standard deviation;

[0043] (2) Extract eigenvalue features from the overall thermal resistance data based on the overall thermal resistance mean value to obtain multiple initial overall thermal resistance eigenvalue features, and screen the multiple initial overall thermal resistance eigenvalue features based on the overall thermal resistance standard deviation to obtain multiple target overall thermal resistance eigenvalue features; and extract eigenvalue features from the overall temperature data based on the overall temperature mean value to obtain multiple initial overall temperature eigenvalue features, and screen the multiple initial overall temperature eigenvalue features based on the overall temperature standard deviation to obtain multiple target overall temperature eigenvalue features;

[0044] (3)Perform mean operations on the local thermal resistance data and local temperature data respectively to obtain the local thermal resistance mean and local temperature mean, and calculate the standard deviations of the local thermal resistance data and local temperature data respectively to obtain the local thermal resistance standard deviation and local temperature standard deviation;

[0045] (4)Extract eigenvalue from the local thermal resistance data according to the local thermal resistance mean to obtain multiple initial local thermal resistance eigenvalues of each LED driver chip, and screen the eigenvalues based on the local thermal resistance standard deviation to obtain multiple target local thermal resistance eigenvalues of each LED driver chip; and extract eigenvalue from the local temperature data according to the local temperature mean to obtain multiple initial local temperature eigenvalues of each LED driver chip, and screen the eigenvalues based on the local temperature standard deviation to obtain multiple target local temperature eigenvalues of each LED driver chip;

[0046] (5)Perform weight calculation and vector mapping on the multiple target overall thermal resistance eigenvalues and multiple target overall temperature eigenvalues to obtain the overall thermal resistance temperature eigenvector; and perform weight calculation and vector mapping on the multiple target local thermal resistance eigenvalues and multiple target local temperature eigenvalues to obtain multiple local thermal resistance temperature eigenvectors.

[0047] Specifically, perform mean operations on the overall thermal resistance data and the overall temperature data respectively to obtain the overall thermal resistance mean and the overall temperature mean, and calculate the standard deviations of the overall thermal resistance data and the overall temperature data respectively to obtain the overall thermal resistance standard deviation and the overall temperature standard deviation. The mean and standard deviation are quantitative indicators describing the central tendency and dispersion degree of the data. For example, assume there is an LED module containing ten LED driver chips, and the thermal resistance and temperature data of each chip under different power conditions are continuously recorded. Perform mean operations on all the thermal resistance data and temperature data of the entire module. Suppose the obtained overall thermal resistance mean is 5.0 Ω and the overall temperature mean is 50 °C. These two means provide a summary of the overall performance. Calculate the standard deviations of the overall thermal resistance and temperature data. Suppose the obtained overall thermal resistance standard deviation is 0.5 Ω and the overall temperature standard deviation is 5 °C. These two standard deviations reflect the fluctuation magnitudes of each data point relative to its mean. Based on the overall thermal resistance mean and standard deviation, extract eigenvalue of the overall thermal resistance data. Identify those special data points, such as abnormally high or low thermal resistance values, which may indicate potential faults or design defects. For example, select all data points greater than the mean plus twice the standard deviation as the initial overall thermal resistance eigenvalues. Similarly, perform the same processing on the overall temperature data to extract the initial overall temperature eigenvalues. Calculate the mean and standard deviation of the local thermal resistance and temperature data of each LED driver chip. Each chip may exhibit different thermal resistance and temperature characteristics due to manufacturing differences, positions, or other environmental factors. Based on the local statistical data, perform eigenvalue extraction and screening to extract the initial local thermal resistance eigenvalues and temperature eigenvalues of each chip, and screen them through the corresponding standard deviations to ensure the significance and relevance of the selected eigenvalues. Calculate the weights and perform vector mapping on the screened overall and local eigenvalues. Convert these eigenvalues into a format that can be used for further analysis and model training. For example, assign a weight to each eigenvalue based on its relative importance or the magnitude of its impact on the module performance, and map these weighted eigenvalues to a higher-dimensional space to form eigenvectors. The overall eigenvector may be composed of a combination of the overall thermal resistance and temperature eigenvalues, while the local eigenvector describes the specific situation of each LED driver chip.

[0048] In a specific embodiment, the process of performing weight calculation and vector mapping on multiple target overall thermal resistance eigenvalues and multiple target overall temperature eigenvalues to obtain the overall thermal resistance temperature eigenvector; and performing weight calculation and vector mapping on multiple target local thermal resistance eigenvalues and multiple target local temperature eigenvalues to obtain multiple local thermal resistance temperature eigenvectors may specifically include the following steps:

[0049] (1) Calculate the overall coefficient of variation between multiple target overall thermal resistance eigenvalues and multiple target overall temperature eigenvalues based on the mean of the overall thermal resistance, the mean of the overall temperature, the standard deviation of the overall thermal resistance, and the standard deviation of the overall temperature, and set the corresponding first weight data according to the overall coefficient of variation;

[0050] (2) Perform feature normalization on multiple target overall thermal resistance eigenvalues and multiple target overall temperature eigenvalues to obtain multiple normalized overall thermal resistance eigenvalues and multiple normalized overall temperature eigenvalues;

[0051] (3) Perform weighted operations and vector concatenation on multiple normalized overall thermal resistance eigenvalues and multiple normalized overall temperature eigenvalues according to the first weight data to obtain an overall thermal resistance temperature feature vector;

[0052] (4) Calculate the local coefficient of variation between multiple target local thermal resistance eigenvalues and multiple target local temperature eigenvalues based on the mean of the local thermal resistance, the mean of the local temperature, the standard deviation of the local thermal resistance, and the standard deviation of the local temperature, and set the second weight data corresponding to each LED driver chip according to the local coefficient of variation;

[0053] (5) Perform feature normalization on multiple target local thermal resistance eigenvalues and multiple target local temperature eigenvalues to obtain multiple normalized local thermal resistance eigenvalues and multiple normalized local temperature eigenvalues;

[0054] (6) Perform weighted operations and vector concatenation on multiple normalized local thermal resistance eigenvalues and multiple normalized local temperature eigenvalues according to the second weight data of each LED driver chip to obtain multiple local thermal resistance temperature feature vectors.

[0055] Specifically, the coefficient of variation between the overall characteristic values is calculated based on the mean and standard deviation of the overall thermal resistance and temperature data. The coefficient of variation is an important statistical indicator for measuring the degree of data dispersion. The coefficient of variation is usually calculated by dividing the standard deviation by the mean. The coefficient of variation provides a dimensionless measure of dispersion, enabling comparison between different data sets. Suppose the mean of the overall thermal resistance of an LED module is 2.5 Ω and the standard deviation is 0.25 Ω, then its coefficient of variation is 0.25 Ω / 2.5 Ω = 0.1. Similarly, if the mean of the temperature is 50 °C and the standard deviation is 5 °C, its coefficient of variation is 5 °C / 50 °C = 0.1. The corresponding first weight data is set according to the coefficient of variation, and the influence of each characteristic value in the eigenvector is adjusted according to the weight. The setting of the weight may enhance or weaken the influence of certain characteristic values according to the magnitude of the coefficient of variation. For example, a higher coefficient of variation may mean that the characteristic value fluctuates more in the data set, so a smaller weight may be assigned to avoid its excessive influence on the output of the model. The selected multiple target overall thermal resistance characteristic values and temperature characteristic values are normalized to standardize the data into a common range, usually between 0 and 1 or between -1 and 1, so as to eliminate the influence between data of different magnitudes and make the model training more stable. For example, normalization is performed through the minimum and maximum values, or the Z-score normalization method is used. The normalized overall thermal resistance and temperature characteristic values are weighted according to the first weight data. Each characteristic value is multiplied by its corresponding weight, and then all the weighted characteristic values are combined by vector concatenation to form an overall thermal resistance temperature eigenvector. Similarly, the same processing steps are performed on the local thermal resistance and temperature data of each LED driver chip. The local coefficient of variation is calculated and the second weight data is set, and these weights will be personalized adjusted according to the specific performance of each chip. Subsequently, feature normalization and weighted operations are performed, and multiple local thermal resistance temperature eigenvectors are obtained through vector concatenation. Each local eigenvector describes the specific performance of the corresponding chip.

[0056] In a specific embodiment, the process of executing step 103 may specifically include the following steps:

[0057] (1) Input the overall thermal resistance temperature eigenvector and multiple local thermal resistance temperature eigenvectors into a preset BP neural network model. The BP neural network model includes: an input layer, multiple first hidden layers, a second hidden layer, an attention mechanism layer, and an output layer;

[0058] (2) Receive the overall thermal resistance temperature eigenvector and multiple local thermal resistance temperature eigenvectors through the input layer respectively;

[0059] (3) Perform high-dimensional feature extraction on each local thermal resistance temperature eigenvector through multiple first hidden layers to obtain multiple local high-dimensional eigenvectors;

[0060] (4) Through the second hidden layer, high-dimensional feature extraction is performed on the overall thermal resistance temperature feature vector to obtain an overall high-dimensional feature vector;

[0061] (5) Input multiple local high-dimensional feature vectors and the overall high-dimensional feature vector into the attention mechanism layer for feature fusion to obtain an attention fusion feature vector;

[0062] (6) Through the ReLU function in the output layer, calculate the remaining useful life of the attention fusion feature vector to obtain remaining useful life prediction data.

[0063] Specifically, input the overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors into a pre-set BP neural network model. The BP neural network model includes: an input layer, multiple first hidden layers, a second hidden layer, an attention mechanism layer, and an output layer. Receive the overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors through the input layer respectively. The overall feature vector contains data on the overall thermal performance and temperature stability of the LED module, while each local feature vector describes the specific performance of a single LED driver chip. Through multiple first hidden layers, high-dimensional feature extraction is performed on each local thermal resistance temperature feature vector respectively to capture deeper and subtle patterns that may affect the performance of the driver chip. Different activation functions, such as Sigmoid or Tanh, may be used in each layer to extract the non-linear relationships in the data, thereby generating multiple local high-dimensional feature vectors. At the same time, through the second hidden layer, high-dimensional feature extraction is performed on the overall thermal resistance temperature feature vector to generate an overall high-dimensional feature vector representing the entire LED module, reflecting the overall behavior and possible long-term performance of the entire module. Input multiple local high-dimensional feature vectors and the overall high-dimensional feature vector into the attention mechanism layer for feature fusion. The attention mechanism optimizes the integration of information by dynamically adjusting the network's attention degree to different feature vectors. Enabling the network to identify the most critical information among numerous features and generate a fused feature vector. Through the ReLU function in the output layer, calculate the remaining useful life of the attention fusion feature vector to obtain remaining useful life prediction data. The ReLU function can enhance the model's prediction ability and avoid the problem of gradient disappearance. Through data processing and intelligent learning, the BP neural network model can predict the performance of each LED module in future operations, providing information on its possible maintenance requirements and life prediction.

[0064] In a specific embodiment, the process of performing high-dimensional feature extraction on the overall thermal resistance temperature feature vector through the second hidden layer to obtain an overall high-dimensional feature vector may specifically include the following steps:

[0065] (1) Input the overall thermal resistance temperature feature vector into the second hidden layer, which includes: the first-layer unidirectionally connected neurons, the second-layer unidirectionally connected neurons, and the third-layer unidirectionally connected neurons;

[0066] (2) Perform high-dimensional feature extraction on the overall thermal resistance temperature feature vector through the first-layer unidirectionally connected neurons to obtain the first thermal resistance temperature feature vector;

[0067] (3) Concatenate the first thermal resistance temperature feature vector and the overall thermal resistance temperature feature vector to obtain the first concatenated feature vector, and input the first concatenated feature vector into the second-layer unidirectionally connected neurons for high-dimensional feature extraction to obtain the second thermal resistance temperature feature vector;

[0068] (4) Concatenate the second thermal resistance temperature feature vector and the overall thermal resistance temperature feature vector to obtain the second concatenated feature vector, and input the second concatenated feature vector into the third-layer unidirectionally connected neurons for high-dimensional feature extraction to obtain the third thermal resistance temperature feature vector;

[0069] (5) Perform weighted fusion on the first thermal resistance temperature feature vector, the second thermal resistance temperature feature vector, and the third thermal resistance temperature feature vector to obtain the overall high-dimensional feature vector.

[0070] Specifically, the overall thermal resistance temperature feature vector is input into the second hidden layer, which includes: the first-layer unidirectionally connected neurons, the second-layer unidirectionally connected neurons, and the third-layer unidirectionally connected neurons. For example, assume there is a feature vector based on temperature and thermal resistance data, which represents the overall working state of an LED module, including information such as average temperature, maximum temperature, average thermal resistance, and thermal resistance fluctuation. When the overall thermal resistance temperature feature vector is input into the first-layer unidirectionally connected neurons of the second hidden layer, through the processing of this layer, the overall thermal resistance temperature feature vector is transformed into the first thermal resistance temperature feature vector. Nonlinear activation functions such as ReLU or Tanh are applied to help the model capture complex patterns and nonlinear relationships in the data. The first thermal resistance temperature feature vector and the overall thermal resistance temperature feature vector are concatenated to obtain the first concatenated feature vector. This strategy enables the network to learn new abstract features while retaining part of the original information, which helps in more refined feature extraction in subsequent layers. The first concatenated feature vector is input into the second-layer unidirectionally connected neurons to perform high-dimensional extraction of the features, generating the second thermal resistance temperature feature vector. Each process of extraction and concatenation gradually expands the "field of vision" of the model, enhancing the depth and accuracy of its prediction. Similarly, the second thermal resistance temperature feature vector is concatenated with the original overall thermal resistance temperature feature vector to generate the second concatenated feature vector, which is then input into the third-layer unidirectionally connected neurons to obtain the third thermal resistance temperature feature vector. The first, second, and third thermal resistance temperature feature vectors are weighted and fused to form an overall high-dimensional feature vector. In this process, each feature vector may be assigned different weights according to its importance in the model prediction. For example, if the third-layer feature vector is considered more critical in the prediction, it may be assigned a higher weight. The weighted fusion ensures that the model can fully consider the comprehensive influence of the features extracted from each layer when making the final prediction.

[0071] In a specific embodiment, the process of executing step 104 may specifically include the following steps:

[0072] (1) Initialize the parameter adjustment range for multiple LED driver chips according to the remaining service life prediction data to obtain the parameter adjustment range for each LED driver chip;

[0073] (2) Based on the particle swarm optimization algorithm, perform parameter initialization configuration for the LED module to obtain the initial chip parameter configuration strategy;

[0074] (3) Dynamically and randomly generate the initial chip parameter configuration strategy according to the parameter adjustment range of each LED driver chip to obtain multiple first chip parameter configuration strategies;

[0075] (4) Calculate the fitness value of each first chip parameter configuration strategy respectively, and perform population division on multiple first chip parameter configuration strategies according to the fitness value to obtain multiple target parameter configuration strategy populations;

[0076] (5) Generate multiple second chip parameter configuration strategies according to multiple target parameter configuration strategy populations;

[0077] (6) Perform parameter optimization solution on multiple second chip parameter configuration strategies to generate a target chip parameter configuration strategy.

[0078] Specifically, initialize the parameter adjustment range for multiple LED driver chips according to the remaining service life prediction data, and define the boundaries of parameter adjustment in the subsequent optimization process. For example, if the prediction data shows that the service life of a certain chip decreases sharply at high temperatures, the adjustment range of its temperature parameter will be set to a lower temperature range. The parameter adjustment range of each chip is personalized according to its predicted life and performance characteristics. Use the particle swarm optimization algorithm to perform initial parameter configuration on the LED module. Particle swarm optimization is an algorithm that simulates social behavior and optimizes by simulating the process of a group of particles searching for the optimal solution in the solution space. Each particle represents a set of possible parameter configurations, and the particle updates its position and velocity by tracking the historical optimal positions of the individual and the group. For example, initially, the particles are randomly distributed within the defined parameter range, and then they are gradually adjusted according to their own experience and the group experience to approach the optimal configuration. According to the parameter adjustment range of each LED driver chip, dynamically and randomly generate the initial chip parameter configuration strategy to obtain multiple first chip parameter configuration strategies through a randomization algorithm, where each strategy attempts to explore new possibilities within the given parameter adjustment range to discover parameter combinations that may bring better performance. Perform fitness evaluation on each first chip parameter configuration strategy. The fitness value is usually based on the performance of this configuration in simulation or actual operation, such as energy efficiency, stability, and expected life, etc. According to the fitness value, perform population division on the first chip parameter configuration strategies and classify them into multiple target parameter configuration strategy populations. Based on each target parameter configuration strategy population, generate multiple second chip parameter configuration strategies. Through genetic algorithm operations such as crossover and mutation to introduce new mutations and increase the diversity of solutions, which helps the algorithm jump out of local optima and find the global optimal solution. Perform parameter optimization solution on the second chip parameter configuration strategies, and finally generate a target chip parameter configuration strategy. Through mathematical modeling and calculation processes, ensure that the setting of each parameter is the optimal choice under the current process and technical conditions.

[0079] The configuration method of the LED driver chip in the embodiment of the present application is described above. Next, the configuration device of the LED driver chip in the embodiment of the present application will be described. Please refer to Figure 2, an embodiment of the configuration device of the LED driving chip in the embodiment of the present application includes:

[0080] A test module 201, configured to perform power cycle tests and data acquisition on multiple LED driving chips in the LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driving chip;

[0081] An extraction module 202, configured to perform feature extraction and feature fusion on the overall thermal resistance data and overall temperature data to obtain an overall thermal resistance temperature feature vector, and perform feature extraction and feature fusion on the local thermal resistance data and local temperature data respectively to obtain multiple local thermal resistance temperature feature vectors;

[0082] A prediction module 203, configured to input the overall thermal resistance temperature feature vector and multiple local thermal resistance temperature feature vectors into a preset BP neural network model for remaining service life prediction to obtain remaining service life prediction data;

[0083] A configuration module 204, configured to perform parameter configuration on the LED module according to the remaining service life prediction data to generate a target chip parameter configuration strategy.

[0084] Through the collaborative cooperation of the above-mentioned components, by performing power cycle tests and data acquisition on multiple driving chips in the LED module, the thermal resistance and temperature data of the whole and local parts of the chip can be obtained, which are the key to understanding the performance of the chip under actual working conditions. This method can reveal in detail the thermal characteristics and potential failure modes of the chip under different working states, providing accurate basic data for subsequent data analysis and feature extraction. Through feature extraction and feature fusion technologies, the key information helpful for predicting the chip life can be extracted from the complex temperature and thermal resistance data. These processing processes use statistical and machine learning technologies to enhance the information density of the data, making the subsequent life prediction more accurate and reliable. The obtained feature vectors accurately reflect the current state and possible future performance of the chip, making the life prediction highly operable and practical. Inputting the extracted feature vectors into a preset BP neural network model for life prediction not only improves the prediction accuracy, but also can dynamically adjust the chip parameters according to the prediction results, realizing the optimization of the performance of the LED driving chip and the extension of its life. This configuration strategy based on intelligent learning algorithms can effectively cope with changing usage conditions and environmental factors, ensure the long-term stability and reliability of the LED driving system, and thus improve the accuracy of the performance parameter configuration of the LED driving chip.

[0085] The present application also provides a computer device, which includes a memory and a processor. Computer-readable instructions are stored in the memory. When the computer-readable instructions are executed by the processor, the processor executes the steps of the configuration method of the LED driving chip in the above-mentioned embodiments.

[0086] The present application also provides a computer-readable storage medium. The computer-readable storage medium can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer executes the steps of the configuration method of the LED driving chip.

[0087] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described system, system and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0088] 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 such an understanding, the technical solution of the present application, in essence, 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 and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

[0089] As described above, the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A configuration method for an LED driving chip, characterized in that, The configuration method of the LED driver chip includes: Performing power cycle tests and data acquisition on multiple LED driver chips in the LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driver chip; Performing feature extraction and feature fusion on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively performing feature extraction and feature fusion on the local thermal resistance data and the local temperature data to obtain multiple local thermal resistance temperature feature vectors; Inputting the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors into a preset BP neural network model for remaining useful life prediction to obtain remaining useful life prediction data; Performing parameter configuration on the LED module according to the remaining useful life prediction data to generate a target chip parameter configuration strategy.

2. The configuration method of the LED driving chip according to claim 1, characterized in that The performing power cycle tests and data acquisition on multiple LED driver chips in the LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driver chip, includes: Performing power cycle tests on the LED module using a preset current and voltage, where the LED module includes multiple LED driver chips; Collecting the module thermal resistance data and module temperature data of the LED module, and respectively collecting the chip thermal resistance data and chip temperature data of each LED driver chip; Respectively performing data denoising and data interpolation on the module thermal resistance data and the module temperature data to obtain the overall thermal resistance data and overall temperature data of the LED module; Respectively performing data denoising and data interpolation on the chip thermal resistance data and chip temperature data of each LED driver chip to obtain the local thermal resistance data and local temperature data of each LED driver chip.

3. The configuration method of the LED driving chip according to claim 1, wherein, The performing feature extraction and feature fusion on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively performing feature extraction and feature fusion on the local thermal resistance data and the local temperature data to obtain multiple local thermal resistance temperature feature vectors, includes: Respectively performing mean operations on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance mean and an overall temperature mean, and respectively performing standard deviation calculations on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance standard deviation and an overall temperature standard deviation; Performing eigenvalue extraction on the overall thermal resistance data according to the overall thermal resistance mean to obtain multiple initial overall thermal resistance eigenvalues, and performing eigenvalue screening on the multiple initial overall thermal resistance eigenvalues according to the overall thermal resistance standard deviation to obtain multiple target overall thermal resistance eigenvalues; and performing eigenvalue extraction on the overall temperature data according to the overall temperature mean to obtain multiple initial overall temperature eigenvalues, and performing eigenvalue screening on the multiple initial overall temperature eigenvalues according to the overall temperature standard deviation to obtain multiple target overall temperature eigenvalues; Perform mean operations on the local thermal resistance data and the local temperature data respectively to obtain the local thermal resistance mean and the local temperature mean, and calculate the standard deviations of the local thermal resistance data and the local temperature data respectively to obtain the local thermal resistance standard deviation and the local temperature standard deviation; Extract eigenvalue of the local thermal resistance data according to the local thermal resistance mean to obtain multiple initial local thermal resistance eigenvalues of each LED driver chip, and screen the multiple initial local thermal resistance eigenvalues according to the local thermal resistance standard deviation to obtain multiple target local thermal resistance eigenvalues of each LED driver chip; and extract eigenvalue of the local temperature data according to the local temperature mean to obtain multiple initial local temperature eigenvalues of each LED driver chip, and screen the multiple initial local temperature eigenvalues according to the local temperature standard deviation to obtain multiple target local temperature eigenvalues of each LED driver chip; Perform weight calculation and vector mapping on the multiple target overall thermal resistance eigenvalues and the multiple target overall temperature eigenvalues to obtain an overall thermal resistance temperature eigenvector; and perform weight calculation and vector mapping on the multiple target local thermal resistance eigenvalues and the multiple target local temperature eigenvalues to obtain multiple local thermal resistance temperature eigenvectors.

4. The configuration method of the LED driving chip according to claim 3, wherein The performing weight calculation and vector mapping on the multiple target overall thermal resistance eigenvalues and the multiple target overall temperature eigenvalues to obtain an overall thermal resistance temperature eigenvector; and performing weight calculation and vector mapping on the multiple target local thermal resistance eigenvalues and the multiple target local temperature eigenvalues to obtain multiple local thermal resistance temperature eigenvectors includes: Calculate the overall coefficient of variation between the multiple target overall thermal resistance eigenvalues and the multiple target overall temperature eigenvalues according to the overall thermal resistance mean, the overall temperature mean, the overall thermal resistance standard deviation and the overall temperature standard deviation, and set corresponding first weight data according to the overall coefficient of variation; Perform feature normalization processing on the multiple target overall thermal resistance eigenvalues and the multiple target overall temperature eigenvalues to obtain multiple normalized overall thermal resistance eigenvalues and multiple normalized overall temperature eigenvalues; Perform weighted operation and vector splicing on the multiple normalized overall thermal resistance eigenvalues and the multiple normalized overall temperature eigenvalues according to the first weight data to obtain an overall thermal resistance temperature eigenvector; Calculate the local coefficient of variation between the multiple target local thermal resistance eigenvalues and the multiple target local temperature eigenvalues according to the local thermal resistance mean, the local temperature mean, the local thermal resistance standard deviation and the local temperature standard deviation, and set corresponding second weight data for each LED driver chip according to the local coefficient of variation; Perform feature normalization processing on the multiple target local thermal resistance eigenvalues and the multiple target local temperature eigenvalues to obtain multiple normalized local thermal resistance eigenvalues and multiple normalized local temperature eigenvalues; Perform weighted operations and vector concatenation on the multiple normalized local thermal resistance eigenvalues and the multiple normalized local temperature eigenvalues according to the second weight data of each LED driver chip to obtain multiple local thermal resistance temperature feature vectors.

5. The configuration method of the LED driving chip according to claim 1, characterized in that, Inputting the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors into a preset BP neural network model for remaining useful life prediction to obtain remaining useful life prediction data, including: Inputting the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors into a preset BP neural network model, the BP neural network model includes: an input layer, multiple first hidden layers, a second hidden layer, an attention mechanism layer, and an output layer; Receiving the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors respectively through the input layer; Performing high-dimensional feature extraction on each local thermal resistance temperature feature vector respectively through the multiple first hidden layers to obtain multiple local high-dimensional feature vectors; Performing high-dimensional feature extraction on the overall thermal resistance temperature feature vector through the second hidden layer to obtain an overall high-dimensional feature vector; Inputting the multiple local high-dimensional feature vectors and the overall high-dimensional feature vector into the attention mechanism layer for feature fusion to obtain an attention fusion feature vector; Calculating the remaining useful life of the attention fusion feature vector through the ReLU function in the output layer to obtain remaining useful life prediction data.

6. The configuration method of the LED driving chip according to claim 5, wherein The performing high-dimensional feature extraction on the overall thermal resistance temperature feature vector through the second hidden layer to obtain an overall high-dimensional feature vector includes: Inputting the overall thermal resistance temperature feature vector into the second hidden layer, the second hidden layer includes: a first layer of unidirectional connection neurons, a second layer of unidirectional connection neurons, and a third layer of unidirectional connection neurons; Performing high-dimensional feature extraction on the overall thermal resistance temperature feature vector through the first layer of unidirectional connection neurons to obtain a first thermal resistance temperature feature vector; Concatenating the first thermal resistance temperature feature vector and the overall thermal resistance temperature feature vector to obtain a first concatenated feature vector, and inputting the first concatenated feature vector into the second layer of unidirectional connection neurons for high-dimensional feature extraction to obtain a second thermal resistance temperature feature vector; Concatenating the second thermal resistance temperature feature vector and the overall thermal resistance temperature feature vector to obtain a second concatenated feature vector, and inputting the second concatenated feature vector into the third layer of unidirectional connection neurons for high-dimensional feature extraction to obtain a third thermal resistance temperature feature vector; Performing weighted fusion on the first thermal resistance temperature feature vector, the second thermal resistance temperature feature vector, and the third thermal resistance temperature feature vector to obtain an overall high-dimensional feature vector.

7. The configuration method of the LED driving chip according to claim 1, wherein Configuring parameters of the LED module according to the remaining useful life prediction data to generate a target chip parameter configuration strategy, including: Initializing the parameter adjustment range of the multiple LED driver chips according to the remaining useful life prediction data to obtain the parameter adjustment range of each LED driver chip; Based on the particle swarm optimization algorithm, initialize the parameters of the LED module to obtain an initial chip parameter configuration strategy; Dynamically and randomly generate the initial chip parameter configuration strategy according to the parameter adjustment range of each LED driver chip to obtain multiple first chip parameter configuration strategies; Calculate the fitness value of each first chip parameter configuration strategy respectively, and divide the multiple first chip parameter configuration strategies according to the fitness value to obtain multiple target parameter configuration strategy groups; Generate multiple second chip parameter configuration strategies according to the multiple target parameter configuration strategy groups; Perform parameter optimization solution on the multiple second chip parameter configuration strategies to generate a target chip parameter configuration strategy.

8. A configuration device for an LED driving chip, characterized in that, The configuration device of the LED driver chip includes: A test module for performing power cycle tests and data collection on multiple LED driver chips in the LED module to obtain the overall thermal resistance data and overall temperature data of the LED module, as well as the local thermal resistance data and local temperature data of each LED driver chip; An extraction module for performing feature extraction and feature fusion on the overall thermal resistance data and the overall temperature data to obtain an overall thermal resistance temperature feature vector, and respectively performing feature extraction and feature fusion on the local thermal resistance data and the local temperature data to obtain multiple local thermal resistance temperature feature vectors; A prediction module for inputting the overall thermal resistance temperature feature vector and the multiple local thermal resistance temperature feature vectors into a pre-set BP neural network model for remaining service life prediction to obtain remaining service life prediction data; A configuration module for performing parameter configuration on the LED module according to the remaining service life prediction data to generate a target chip parameter configuration strategy.

9. A computer device, characterized in that, The computer device includes: a memory and at least one processor, and instructions are stored in the memory; The at least one processor invokes the instructions in the memory so that the computer device executes the configuration method of the LED driver chip according to any one of claims 1-7.

10. A computer-readable storage medium, on which instructions are stored, characterized in that, When the instructions are executed by the processor, the configuration method of the LED driver chip according to any one of claims 1-7 is implemented.

Images