Fault prediction and health management system and method for LED lamp beads

Abstract

The invention discloses a fault prediction and health management system and method for LED lamp beads, and the method comprises the following steps: collecting spectrum data through an enhanced spectrum sampling technology, and dynamically adjusting sampling parameters through an optimization algorithm to improve the data precision; external environment parameters are collected in real time, and a correlation model of the spectrum and the environment is constructed; in the data analysis process, abnormal data are detected and corrected, the evaluation accuracy is ensured, a model is dynamically adjusted based on a multi-level feedback correction mechanism, comprehensive spectrum data are generated, finally, the performance of the LED lamp beads is comprehensively analyzed and evaluated by constructing a multi-dimensional photoelectric characteristic coupling model and inputting the data into the model, and a comprehensive health management report is provided. According to the invention, the accuracy and adaptability of evaluation are effectively improved, and the method is suitable for LED lamp bead performance evaluation under various complex environment conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of performance evaluation, in particular to a fault prediction and health management system and method for LED lamp beads. BACKGROUND

[0002] With the wide application of LED lamp beads in key fields such as lighting and display, reliable fault prediction and health management of LED lamp beads has become the key to ensure the stable operation of the system. An effective PHM system not only needs to evaluate the current performance, but also needs to predict the future performance degradation trend, quantify the health status, and provide early fault warning.

[0003] The prior art has the following significant limitations in realizing effective PHM of LED lamp beads: 1. Lack of effective fault prediction capability, traditional methods rely on end-point testing or simple aging models, which can only perform rough life estimation and cannot perform dynamic and high-precision fault prediction based on real-time operation data, resulting in potential risks that cannot be discovered in time.

[0004] 2. Health status cannot be quantitatively evaluated, existing technologies usually only provide isolated performance parameters (such as luminous flux, color temperature), but lack a health status index that can integrate multi-dimensional information (such as photoelectric properties, environmental influences, historical data), making it impossible for managers to intuitively and quantitatively judge the health status of the lamp beads.

[0005] 3. Poor adaptability to complex working conditions, existing models (such as linear decay models) have a significant decline in prediction accuracy when facing dynamic coupling effects of temperature, humidity and other environmental factors, and cannot truly reflect the health degradation trajectory of LEDs under actual complex working conditions.

[0006] 4. The potential of data-driven has not been tapped, although a large amount of spectral and environmental data can be collected, existing technologies lack effective algorithms and models to convert these multi-dimensional, non-linear data into forward-looking health management information.

[0007] Therefore, how to construct a system and method that can integrate multi-source data and utilize advanced algorithm models to achieve accurate fault prediction and quantitative health management of LED lamp beads is a technical problem that needs to be solved by those skilled in the art. SUMMARY

[0008] One purpose of the present application is to provide a failure prediction and health management system and method for LED lamp beads, the present application combines spectrum collection optimization, dynamic coupling of environmental parameters, multi-level feedback correction and nonlinear performance analysis, and describes in detail the implementation method of evaluating the photoelectric performance, durability and environmental adaptability of LED lamp beads under complex environmental conditions, which has the advantages of strong adaptability, high evaluation accuracy and comprehensive analysis.

[0009] According to the failure prediction and health management method for LED lamp beads provided by the embodiment of the present application, the following steps are included: S1, using quantum dot enhanced spectrum sampling technology, combining the spectrum characteristics of quantum dot materials, sampling the spectrum signal emitted by the LED lamp beads to generate original spectrum data; S2, applying a Bayesian optimization algorithm to the original spectrum data to dynamically adjust the sampling parameters, optimize the sampling conditions, and generate optimized spectrum data; S3, real-time collection of external environmental characteristic parameters of the LED lamp beads during operation, construction of a spectrum and environmental characteristic correlation model, generation of environment-related spectrum data through multi-dimensional coupling of external environmental characteristic parameters and optimized spectrum data; S4, modeling the environment-related spectrum data based on a Gaussian process regression algorithm, extracting the dynamic change characteristics of photoelectric performance, and generating photoelectric performance prediction data; S5, multi-level real-time feedback correction based on the photoelectric performance prediction data, analysis of the influence of environmental disturbance and abnormal data on performance evaluation, real-time correction and generation of corrected comprehensive spectrum data; S6, construction of a dynamic multi-dimensional photoelectric characteristic coupling model, input of the corrected comprehensive spectrum data into the dynamic multi-dimensional photoelectric characteristic coupling model, health state evaluation and failure prediction of the LED lamp beads, and generation of a health management report containing health state index, predicted remaining life or failure risk level.

[0010] Optionally, S2 specifically includes: S21, the original spectrum data obtained by the quantum dot enhanced spectrum sampling technology is defined as a data set , the initial sampling parameter set is set from the data set , , wherein represents the sampling frequency, represents the sampling time, represents the signal-to-noise ratio target range; S22, constructing a sampling parameter optimization objective function , wherein , represents the current sampling parameter set, and the sampling parameter optimization objective function is an error metric corresponding to the current sampling parameter set and sampling integrity , combined with a dynamic non-linear adjustment factor is calculated: ); wherein, and are weight factors, is a dynamic adjustment factor, represents the signal fluctuation range under the current sampling parameter set , calculated by the signal standard deviation of ; S23, maximize the sampling parameter optimization objective function using a Bayesian optimization algorithm to select the next sampling parameter set ; S24, apply the selected next sampling parameter set to sample the original spectral data to generate new spectral data , and calculate the error measure and the sampling integrity of the new spectral data ; S25, repeat and , iteratively update the sampling parameter set until the sampling parameter optimization objective function converges to reach a preset maximum number of iterations ; S26, output the optimized sampling parameter set and the corresponding optimized spectral data , wherein satisfies the optimal requirements of the error measure of the spectral data and the sampling integrity.

[0011] Optionally, the S22 specifically comprises: S221, based on the initial sampling parameter set and the statistical characteristics of the original spectral data , combined with the dynamic weights of the sampling frequency , the sampling time and the signal-to-noise ratio target , construct the error measure , wherein represents the current sampling parameter set; S222, combined with the signal coverage characteristics of the original spectral data , refine the sampling integrity to describe the coverage effect of the current sampling parameter set on the spectral signal; S223. Error metric based on definition and sampling integrity , combined with the dynamic nonlinear adjustment factor , construct sampling parameter optimization objective function .

[0012] Optionally, the S3 specifically includes: S31. Real-time collection of external environmental characteristic parameters of LED lamp beads during operation, and organization of the collected external environmental characteristic parameters into an environmental data set , for environmental data sets Each parameter in is normalized to obtain a normalized environmental data set ; S32, based on optimized spectral data and normalized environmental data sets , construct a preliminary correlation model between spectral data and external environmental characteristic parameters, use the multiple linear regression method to describe the preliminary relationship between the optimized spectral data and external environmental characteristic parameters and obtain the corrected spectral value: ; in, is the corrected spectral value, To optimize the original spectral values ​​in the spectral data, is the regression coefficient of the external environment characteristic parameter, For the Normalized environmental data sets The characteristic function of To optimize the weight of spectral data, To optimize the characteristic function of spectral data, is the number of characteristic parameters of the external environment; S33. Using a preliminary correlation model between the spectral data and the external environmental characteristic parameters, and employing a multidimensional nonlinear regression method, a multidimensional coupling model of the spectral data and the external environmental characteristic parameters is constructed, wherein the goal of the multidimensional coupling model is to minimize the spectral value correction error; S34, collect and normalize the normalized environmental data set in real time and optimize spectral data Input multidimensional coupling model to generate environmental correlation spectral data .

[0013] Optionally, the S32 specifically includes: S321. Based on normalized environmental data set , extract the characteristics of each external environment characteristic parameter and define the characteristic function , which is used to describe the linear influence of external environmental characteristic parameters on spectral values: ; in, For the Normalized environmental data sets The characteristic function of The first External environmental characteristic parameters, and Respectively with The linear regression coefficient related to the external environment characteristic parameters; S322, combined with optimized spectral data The distribution characteristics of the spectral data are defined to optimize the characteristic function : ; in, To optimize the spectral data at wavelength The spectral value distribution at To optimize the weighting factor of spectral data, and Indicates the effective boundary points of the spectral signal; S323, based on the defined characteristic function and , construct a preliminary correlation model between spectral data and external environmental characteristic parameters, and use the multiple linear regression method to describe the linear relationship between the optimized spectral data and external environmental characteristic parameters; S324, Optimizing the parameters of the multivariate linear regression method using the least squares method and , the fitting goal is to minimize the error of the regression model.

[0014] Optionally, the S33 specifically includes: S331. Construct a multidimensional nonlinear regression model of spectral data and external environmental characteristic parameters , the input of the multidimensional nonlinear regression model is the normalized environmental data set and optimize spectral data , the output is the corrected spectral value, and the multidimensional nonlinear regression model objective is to minimize the spectral value correction error: ; in, is the objective function of the multidimensional nonlinear regression model, is the set of model parameters, For the The corrected spectral values, For the A normalized environmental data set, For the Optimized spectral data, Represents the total number of sample points used to build and train the multidimensional nonlinear regression model; S332. Initialize model parameter set , including the weight vector and the bias vector , the convergence speed of the optimization process; S333, Objective function of multidimensional nonlinear regression model based on gradient descent method Optimize and update the model parameter set , each iteration is performed by gradient descent; S334. Determine the objective function of the multidimensional nonlinear regression model The iteration is terminated when one of the following conditions is met: The objective function value of the iteration The absolute difference between the objective function values ​​of the iterations is less than the preset convergence threshold , or the current number of iterations exceeds the maximum number of iterations , it is determined that the convergence condition is met and the iteration is stopped; S335, the optimized model parameter set Substitute into the multidimensional nonlinear regression model , generating the final multidimensional coupled model.

[0015] Optionally, the S5 specifically includes: S51, based on the photoelectric performance prediction data , build anomaly detection function , detect and mark outliers in the data. When the absolute value of the deviation between a data point and the mean of the optoelectronic performance prediction data set is greater than the product of the preset anomaly detection threshold multiple and the standard deviation, the data point is judged to be abnormal and the abnormal mark value is 1. If the absolute value of the deviation does not exceed the preset anomaly detection threshold coefficient, the data point is judged to be normal and the abnormal mark value is 0; S52: For the detected abnormal data, The data points are corrected using the weighted neighborhood interpolation method, and the corrected data points are combined with the normal data points to generate the corrected photoelectric performance prediction data. , the corrected data points The calculation formula is: ; in, For the The neighborhood set of data points, Neighborhood set The data points, is the weight coefficient, according to Data points and The distance between data points calculate; S53, based on the corrected photoelectric performance prediction data and normalized environmental data sets , combined with experimentally determined reference spectral data , building a multi-level feedback correction model The goal of the multi-level feedback correction model is to minimize the corrected comprehensive spectral data With reference spectral data The error between S54. Optimizing the multi-level feedback correction model using gradient descent method Parameter set , to minimize the multi-level feedback correction model The objective function After the optimization is completed, the corrected photoelectric performance prediction data set and normalized environmental data sets Input the multi-level feedback correction model to generate the corrected comprehensive spectral data .

[0016] Optionally, the S6 specifically includes: S61, based on the corrected comprehensive spectral data , normalized environmental data set , and a set of performance reference indicators extracted from the performance database , define the input variables and output variables of the dynamic multi-dimensional optoelectronic characteristic coupling model, where the input variables include 、 and ,The output variables include the health management report Y, which contains the health status index, the predicted remaining life or the failure risk level; S62. Constructing a characteristic function set of a dynamic multi-dimensional optoelectronic characteristic coupling model , used to describe the corrected comprehensive spectral data and normalized environmental data sets Health management report Nonlinear effects of: ; in, Indicates the A characteristic function, , , is the weight coefficient of the characteristic function, 、 and respectively represent the feature mapping functions of the corrected comprehensive spectral data, the normalized environment data set and the performance reference index set; S63, using the characteristic function set constructing a dynamic multi-dimensional optoelectronic characteristic coupling model by nonlinear fitting The dynamic multi-dimensional optoelectronic characteristic coupling model aims to predict the health status index, the remaining life or the failure risk level by nonlinear fitting; S64, health management report The calculation of the corrected comprehensive spectral data , the normalized environment data set and the performance reference index set input the dynamic multi-dimensional optoelectronic characteristic coupling model to generate a health management report Y* containing the health status index, the predicted remaining life or the failure risk level.

[0017] Optionally, a failure prediction and health management system for LED lamp beads includes the following modules: Spectrum sampling module: using quantum dot enhanced spectrum sampling technology to collect the spectral data of LED lamp beads to generate original spectral data; Sampling parameter optimization module: dynamically adjusting the sampling parameters by Bayesian optimization algorithm to generate optimized spectral data; Environment characteristic acquisition module: real-time acquisition of external environment characteristic parameters of LED lamp beads during operation, and normalization processing of the collected external environment characteristic data to generate a normalized environment data set; Spectrum and environment coupling model construction module: based on the optimized spectral data and the normalized environment data set, constructing a preliminary correlation model and a multi-dimensional coupling model of spectral and environmental characteristic parameters; Abnormality detection and correction module: abnormality detection of optoelectronic performance prediction data, marking abnormal data through statistical analysis, and correcting abnormal data by weighted neighborhood interpolation method to generate corrected optoelectronic performance prediction data; Multi-level feedback correction module: based on the corrected optoelectronic performance prediction data and the normalized environment data set, constructing a multi-level real-time feedback correction model to optimize the corrected comprehensive spectral data; Dynamic multi-dimensional optoelectronic characteristic coupling module: constructing a dynamic multi-dimensional optoelectronic characteristic coupling model, integrating the corrected comprehensive spectral data, the normalized environment data set and the performance reference index set to predict the optoelectronic performance, durability and environmental adaptability index of LED lamp beads; A health management report generation module: input the corrected comprehensive spectral data into a dynamic multi-dimensional optoelectronic property coupling model to generate a health management report of the LED lamp bead, which contains a health status index, a predicted remaining life or a failure risk level, etc.

[0018] The beneficial effects of the present application are: (1) The present application realizes dynamic sampling optimization of the optoelectronic performance of the LED lamp bead by combining quantum dot enhanced spectral sampling technology, Bayesian optimization algorithm and high-efficiency data processing method, can adapt to complex environmental changes and capture small changes in performance, thereby significantly improving the accuracy and adaptability of the evaluation. It provides a high-fidelity data basis for establishing a precise health status baseline, and solves the problem that the existing technology cannot be fine-managed due to insufficient data accuracy.

[0019] (2) The present application realizes accurate prediction of the performance degradation trajectory of the LED lamp bead under complex working conditions by constructing a dynamic coupling model of spectrum and environment and combining machine learning algorithms such as Gaussian process regression, solving the problem of insufficient prediction ability of traditional models.

[0020] (3) The present application realizes quantitative evaluation and comprehensive management of the health status of the LED lamp bead by constructing a dynamic multi-dimensional optoelectronic property coupling model, can output multi-dimensional management information including health index and failure risk level, solving the problem that the existing technology can only provide single performance parameter and cannot perform comprehensive health evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, which together with the embodiments of the present application, serve to explain the present application, and do not constitute a limitation of the present application. In the drawings: Figure 1 A general framework diagram of a failure prediction and health management method for LED lamp beads is proposed. DETAILED DESCRIPTION

[0022] The present application will now be described in further detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, and only illustrate the basic structure of the present application in a schematic manner, and therefore only show the components related to the present application.

[0023] Reference Figure 1 A failure prediction and health management method for LED lamp beads includes the following steps: S1, using quantum dot enhanced spectral sampling technology, combining the spectral properties of quantum dot materials, sampling the spectral signal emitted by the LED lamp bead to generate original spectral data; The embodiment utilizes quantum dot enhanced spectral sampling technology to collect the emission spectrum of the LED lamp bead with high sensitivity. Specifically, the selective enhancement of different wavelength light signals is realized by embedding a sensor array of quantum dot materials, thereby improving the response capability of weak spectral components. This method not only significantly expands the dynamic range of spectral sampling, but also maintains high signal-to-noise ratio and stability of the data in high background noise or complex environment, solving the problem of low collection accuracy of traditional spectral sampling under low illumination or interference conditions. By highly matching the narrowband selectivity of quantum dot materials with the spectral characteristics of LEDs, the scheme realizes high-fidelity acquisition of original spectral data, providing accurate and rich basic data for subsequent performance modeling and dynamic optimization, significantly enhancing the robustness and adaptability of the overall evaluation system in actual application scenarios.

[0024] S2, applying a Bayesian optimization algorithm to the original spectral data to dynamically adjust the sampling parameters, optimize the sampling conditions, and generate optimized spectral data; S3, real-time collection of external environmental characteristic parameters of the LED lamp bead during operation, construction of a spectral and environmental characteristic correlation model, generation of environmental correlation spectral data through multi-dimensional coupling of external environmental characteristic parameters and optimized spectral data; S4, modeling of the environmental correlation spectral data based on a Gaussian process regression algorithm, extraction of dynamic change characteristics of optoelectronic performance, and generation of optoelectronic performance prediction data; The embodiment adopts a Gaussian process regression algorithm to model the environmental correlation spectral data, and realizes learning and prediction of the implicit relationship between spectral data and external environmental changes by constructing a non-parametric Bayesian model. The system first inputs the environmental correlation spectral data processed by multi-dimensional coupling, dynamically adjusts the sensitivity of the model to different wavelengths and environmental factors using the covariance function of the Gaussian process, thereby realizing continuous prediction of optoelectronic performance indicators (such as brightness, color temperature, power response, etc.) over time or environmental changes. This approach does not require a pre-set function structure, has high flexibility and uncertainty expression capability, and can effectively capture the nonlinear change trend of optoelectronic performance under complex environmental conditions. This method significantly improves the forward-looking and adaptability of LED performance evaluation, providing accurate and dynamic data support for subsequent anomaly detection and correction mechanisms.

[0025] S5, multi-level real-time feedback correction based on the optoelectronic performance prediction data, analysis of the influence of environmental disturbances and abnormal data on performance evaluation, implementation of real-time correction, and generation of corrected comprehensive spectral data; S6, construction of a dynamic multi-dimensional optoelectronic characteristic coupling model, input of the corrected comprehensive spectral data into the dynamic multi-dimensional optoelectronic characteristic coupling model, health state evaluation and fault prediction of the LED lamp bead, and generation of a health management report containing health state index, predicted remaining life, or fault risk level.

[0026] In this embodiment, S2 specifically includes: S21, the original spectral data obtained by the quantum dot enhanced optical sampling technology is defined as a data set , the data set is set as an initial sampling parameter set , wherein represents a sampling frequency, represents a sampling time, represents a signal-to-noise ratio target range; S22, a sampling parameter optimization objective function is constructed , wherein , represents the current sampling parameter set, and the sampling parameter optimization objective function is calculated by combining the error metric and the sampling integrity of the current sampling parameter set with a dynamic nonlinear adjustment factor : ); wherein, and are weight factors, is a dynamic adjustment factor, represents the signal fluctuation range under the current sampling parameter set , which is calculated by the signal standard deviation of ; The formula for the sampling parameter optimization objective function comprehensively considers two dimensions of sampling error and sampling integrity, forms a unified evaluation index through weighted addition, and introduces a dynamic nonlinear adjustment factor to adaptively adjust the optimization tendency according to the fluctuation degree of the current sampling data. The principle is that the sampling error reflects the accuracy of the data, the sampling integrity reflects the coverage degree of the data, and the dynamic factor automatically adjusts the relative weight of the two according to the change intensity of the spectral signal, so that the objective function can still maintain the rationality of the optimization direction under different spectral characteristics and environments. The system can more scientifically judge the pros and cons of the current sampling setting, provide a clear optimization target for subsequent Bayesian optimization, significantly improve the accuracy and representativeness of the sampling, and enhance the reflection ability of the spectral data on the performance changes of the LED lamp beads.

[0027] S23, the sampling parameter optimization objective function is maximized by using a Bayesian optimization algorithm, and the next sampling parameter set is selected; S24, the original spectral data is sampled using the selected next sampling parameter set to generate new spectral data , and the new spectral data Error metric and sampling integrity ; S25, repeat and , iteratively update the sampling parameter set , until the sampling parameters optimize the objective function Converges to the preset maximum number of iterations ; S26. Output optimized sampling parameter set and the corresponding optimized spectral data ,in Satisfy the optimal requirements of error measurement and sampling integrity of spectral data.

[0028] This implementation combines quantum dot-enhanced spectral sampling technology with a Bayesian optimization algorithm to dynamically optimize sampling parameters. By constructing an objective function that comprehensively considers error metrics and sampling integrity, this method enables real-time adjustment of sampling conditions and generates optimized, high-precision spectral data. This method significantly improves the accuracy and sensitivity of spectral data acquisition, providing more reliable data support for LED lamp performance evaluation.

[0029] In this embodiment, S22 specifically includes: S221, based on the initial sampling parameter set and raw spectral data The statistical characteristics of the sampling frequency , sampling time and signal-to-noise ratio targets Dynamic weights, construct error metrics ,in Indicates the current sampling parameter set; S222, combined with original spectral data Signal coverage characteristics, refined sampling integrity , used to describe the current sampling parameter set Coverage effect on spectral signals; S223. Error metric based on definition and sampling integrity , combined with the dynamic nonlinear adjustment factor , construct sampling parameter optimization objective function .

[0030] This implementation effectively combines the dynamic weights of sampling frequency, sampling time, and signal-to-noise ratio targets by constructing a dynamic error metric function and refining the sampling integrity index, generating an optimized set of sampling parameters. This not only improves the sampling accuracy of spectral data, but also enhances the sampling coverage effect, providing more accurate and comprehensive data support for photoelectric characteristic analysis.

[0031] In this embodiment, S3 specifically includes: S31. Real-time collection of external environmental characteristic parameters of LED lamp beads during operation, and organization of the collected external environmental characteristic parameters into an environmental data set , for environmental data sets Each parameter in is normalized to obtain a normalized environmental data set ; S32, based on optimized spectral data and normalized environmental data sets , construct a preliminary correlation model between spectral data and external environmental characteristic parameters, use the multiple linear regression method to describe the preliminary relationship between the optimized spectral data and external environmental characteristic parameters and obtain the corrected spectral value: ; in, is the corrected spectral value, To optimize the original spectral values ​​in the spectral data, is the regression coefficient of the external environment characteristic parameter, For the Normalized environmental data sets The characteristic function of To optimize the weight of spectral data, To optimize the characteristic function of spectral data, is the number of characteristic parameters of the external environment; This formula constructs a preliminary correlation model between spectral data and external environmental characteristic parameters. Its core principle is to map and associate the optimized spectral data with normalized environmental parameters through multiple linear regression. Specifically, the model transforms each environmental parameter through a characteristic function and performs a weighted combination with the characteristics in the spectral data to finally generate a corrected spectral value. It not only takes into account the distribution characteristics of the spectrum itself, but also introduces the linear influence of environmental factors on spectral changes, thereby more realistically reflecting the spectral response of LED lamp beads in the actual operating environment. The effect is to achieve basic modeling of environmental impacts without increasing the complexity of the model, providing initial conditions and modeling directions for subsequent nonlinear coupling models, and improving the overall system's adaptability to external interference and modeling accuracy.

[0032] S33. Using a preliminary correlation model between the spectral data and the external environmental characteristic parameters, and employing a multidimensional nonlinear regression method, a multidimensional coupling model of the spectral data and the external environmental characteristic parameters is constructed, wherein the goal of the multidimensional coupling model is to minimize the spectral value correction error; S34, collect and normalize the normalized environmental data set in real time and optimize spectral data Input multidimensional coupling model to generate environmental correlation spectral data .

[0033] This implementation uses real-time acquisition and normalization of environmental characteristic parameters, combined with optimized spectral data, to construct a preliminary correlation model between the spectrum and environmental characteristic parameters. Dynamic coupling is achieved through multidimensional nonlinear regression. This method accurately reflects the dynamic impact of environmental changes on spectral data, generating more adaptable and accurate environmentally correlated spectral data, thereby improving the precision and comprehensiveness of LED lamp performance evaluation.

[0034] In this embodiment, S32 specifically includes: S321. Based on normalized environmental data set , extract the characteristics of each external environment characteristic parameter and define the characteristic function , which is used to describe the linear influence of external environmental characteristic parameters on spectral values: ; in, For the Normalized environmental data sets The characteristic function of The first External environmental characteristic parameters, and Respectively with The linear regression coefficient related to the external environment characteristic parameters; This formula is used to map each normalized external environmental parameter into a linear characteristic function, with the aim of describing the linear influence of the environmental variable on the spectral value. The principle is to introduce a slope coefficient and an offset constant for each environmental parameter, so that each environmental characteristic not only has its own weight, but also can show its driving ability on the spectral change trend in the model. The design of this linear characteristic mapping enables the model to clearly distinguish the intensity and direction of the effects of different environmental factors on the spectral response. Its effect is to establish a preliminary correlation between spectral data and environmental characteristics, providing a characteristic expression with actual physical meaning for subsequent multidimensional modeling, while also improving the sensitivity and explanatory power of the overall model to environmental disturbances.

[0035] S322, combined with optimized spectral data The distribution characteristics of the spectral data are defined to optimize the characteristic function : ; in, To optimize the spectral data at wavelength The spectral value distribution at to optimize the weight factor of the spectral data, and representing the effective boundary points of the spectral signal; The formula defines an integral type of characteristic function for extracting the spectral energy distribution characteristics in the entire wavelength range. Specifically, the spectral signal is integrated in the effective wavelength range to obtain a numerical characteristic that represents the overall spectral performance. This characteristic not only considers the spectral intensity of different wavelength points, but also expresses their continuity and total amount through the integral form, more comprehensively reflecting the global properties of the spectral data. By introducing an adjustable weight factor, the importance of the characteristic can be adjusted according to different evaluation tasks. This design effectively improves the perception accuracy of the model for spectral changes and provides a key input basis for subsequent environment coupling and performance evaluation models.

[0036] S323, based on the defined characteristic function and , a preliminary correlation model of spectral data and external environment characteristic parameters is constructed, and a multivariate linear regression method is used to describe the linear relationship between the optimized spectral data and the external environment characteristic parameters; S324, using least squares method to optimize the parameters of the multivariate linear regression method and , the fitting target is to minimize the error of the regression model.

[0037] In this embodiment, by defining the characteristic function of the environment characteristic and the optimized spectral data, combining multivariate linear regression and least squares optimization, a preliminary correlation model of spectral data and environment characteristic parameters is constructed, which realizes accurate description of the linear influence of environment characteristic on spectral value and error minimization, significantly improves the accuracy of data fitting and the reliability of the model.

[0038] In this embodiment, S33 specifically includes: S331, constructing a multi-dimensional nonlinear regression model of spectral data and external environment characteristic parameters , the input of the multi-dimensional nonlinear regression model is the normalized environment data set and the optimized spectral data , the output is the corrected spectral value, and the target of the multi-dimensional nonlinear regression model is to minimize the correction error of the spectral value: ; wherein, is the objective function of the multi-dimensional nonlinear regression model, is the model parameter set, is the corrected spectral value, is the normalized environment data set, For the first Optimized spectral data, represents the total number of sample points used to build and train the multi-dimensional nonlinear regression model; The formula builds a multi-dimensional nonlinear regression model for accurately capturing the complex dynamic coupling relationship between spectral data and external environmental characteristic parameters. The model takes normalized environmental parameters and optimized spectral data as joint input, learns the nonlinear variation trend of spectral response under different environmental conditions, and outputs corrected spectral values. The core principle is to establish a trainable function structure, so that the model has flexible fitting capability when dealing with high-dimensional and multi-variable input. By comparing the deviation between the predicted output and the reference spectral value and continuously optimizing the model parameters, the system can minimize the overall error, thereby significantly improving the accuracy and stability of the spectral data. This method breaks through the limitations of traditional linear models and more effectively adapts to spectral changes under various complex environmental disturbances, providing a more representative and robust input basis for subsequent performance evaluation.

[0039] S332, initialize the model parameter set , including the weight vector and the bias vector , the convergence speed of the optimization process; S333, optimize the objective function of the multi-dimensional nonlinear regression model based on the gradient descent method, update the model parameter set , and perform each iteration by gradient descent; S334, judge the convergence of the objective function of the multi-dimensional nonlinear regression model, terminate the iteration when one of the following conditions is met, if the absolute difference between the objective function value of the th iteration and the objective function value of the th iteration is less than the preset convergence threshold , or the current iteration number exceeds the maximum iteration number , it is determined that the convergence condition is met, and the iteration is stopped; S335, substitute the optimized model parameter set into the multi-dimensional nonlinear regression model to generate the final multi-dimensional coupling model.

[0040] This embodiment realizes the deep coupling of spectral data and external environmental characteristic parameters by building and optimizing a multi-dimensional nonlinear regression model, which can dynamically correct spectral values and minimize errors. By introducing gradient descent optimization and convergence judgment mechanism, not only the training efficiency of the model is improved, but also the accuracy and stability of the correction result are ensured, providing reliable technical support for LED lamp bead performance evaluation.

[0041] In this embodiment, S5 specifically includes: S51, based on the photoelectric performance prediction data , build anomaly detection function , detect and mark outliers in the data. When the absolute value of the deviation between a data point and the mean of the optoelectronic performance prediction data set is greater than the product of the preset anomaly detection threshold multiple and the standard deviation, the data point is judged to be abnormal and the abnormal mark value is 1. If the absolute value of the deviation does not exceed the preset anomaly detection threshold coefficient, the data point is judged to be normal and the abnormal mark value is 0; S52: For the detected abnormal data, The data points are corrected using the weighted neighborhood interpolation method, and the corrected data points are combined with the normal data points to generate the corrected photoelectric performance prediction data. , the corrected data points The calculation formula is: ; in, For the The neighborhood set of data points, Neighborhood set The data points, is the weight coefficient, according to Data points and The distance between data points calculate; This formula uses a weighted neighborhood interpolation method to repair it. The specific approach is to select the normal data points adjacent to the data point in the data sequence or feature space, calculate the weighted average of the data point and these neighboring points, and use the result as the replacement value for the abnormal point. The weight coefficient is set according to the relative distance or similarity. The closer the distance and the higher the similarity, the greater the weight. In order to ensure the physical consistency of the repair result, the interpolation calculation adopts normalization processing, and the weighted sum is divided by the total weight so that the repair value is always within a reasonable range. The effect of this method is that, without destroying the continuity of the data structure, it effectively eliminates the interference of abnormal fluctuations on the model evaluation results, thereby improving the data quality and the robustness of the overall evaluation.

[0042] S53, based on the corrected photoelectric performance prediction data and normalized environmental data sets , combined with experimentally determined reference spectral data , building a multi-level feedback correction model The goal of the multi-level feedback correction model is to minimize the corrected comprehensive spectral data With reference spectral data The error between S54, optimizing the multi-level feedback correction model using gradient descent method of the parameter set to minimize the objective function of the multi-level feedback correction model , after optimization, input the corrected photoelectric performance prediction data set and the normalized environment data set into the multi-level feedback correction model to generate corrected comprehensive spectral data .

[0043] The embodiment eliminates the abnormal influence in the photoelectric performance prediction data by abnormality detection and weighted neighborhood interpolation correction, constructs a multi-level feedback correction model combining normalized environment data and reference spectral data, optimizes the model parameters by gradient descent method, realizes high-precision correction of comprehensive spectral data, and significantly improves the accuracy and consistency of performance evaluation.

[0044] In the embodiment, S6 specifically includes: S61, based on the corrected comprehensive spectral data , the normalized environment data set , and the performance reference index set extracted from the performance database , define the input variables and output variables of the dynamic multi-dimensional photoelectric characteristic coupling model, wherein the input variables include 、 and , and the output variables include the health management report Y, which contains the health state index, the predicted remaining life or the fault risk level; S62, construct a characteristic function set of the dynamic multi-dimensional photoelectric characteristic coupling model , which is used to describe the nonlinear influence of the corrected comprehensive spectral data and the normalized environment data set on the health management report : ; wherein, represents the th characteristic function, , , is the weight coefficient of the characteristic function, 、 and respectively represent the feature mapping functions of the corrected comprehensive spectral data, the normalized environment data set and the performance reference index set; ​The formula constructs a set composed of multiple characteristic functions for comprehensively expressing the nonlinear influence of spectral data, environmental data, and performance reference indicators on the LED lamp bead health management report. Each characteristic function is composed of three parts, which respectively extract the corrected spectral features, normalized environmental features, and performance reference indicators, and assign independent weight coefficients to them. This design allows the system to flexibly adjust the influence of different data sources on the final evaluation results according to their importance and behavior patterns. The principle is to abstract and transform data of different dimensions through a feature mapping function, so that the complex and variable input data have a unified expression structure, thereby providing a highly adjustable input vector for the subsequent nonlinear fitting model. This method significantly enhances the expression and generalization ability of the model, making the performance evaluation results more close to the actual performance, and improving the accuracy and flexibility of the system in multi-source data fusion modeling.

[0045] S63, utilizing the characteristic function set constructing a dynamic multi-dimensional optoelectronic characteristic coupling model for nonlinear fitting The dynamic multi-dimensional optoelectronic characteristic coupling model aims to predict the health status index, remaining life, or failure risk level through nonlinear fitting. S64, health management report The calculation of the health management report Y* will input the corrected comprehensive spectral data , the normalized environmental data set , and the performance reference indicator set into the dynamic multi-dimensional optoelectronic characteristic coupling model to generate a health management report Y* containing the health status index, predicted remaining life, or failure risk level.

[0046] The present embodiment uses the dynamic multi-dimensional optoelectronic characteristic coupling model to input the corrected comprehensive spectral data, normalized environmental data, and performance reference indicator set as inputs, and performs nonlinear fitting with the characteristic function set to accurately predict optoelectronic performance, durability, and environmental adaptability indicators, achieving comprehensive, accurate, and dynamic adaptability of performance evaluation, and providing a comprehensive solution for LED lamp bead performance analysis.

[0047] In the present embodiment, a failure prediction and health management system for LED lamp beads includes the following modules: Spectrum sampling module: uses quantum dot enhanced spectrum sampling technology to collect spectral data of LED lamp beads to generate original spectral data; Sampling parameter optimization module: dynamically adjusts the sampling parameters through the Bayesian optimization algorithm to generate optimized spectral data; An environmental characteristic collection module: real-time collection of external environmental characteristic parameters of the LED lamp bead in operation, and normalization processing of the collected external environmental characteristic data to generate a normalized environmental data set; A spectrum and environment coupling model construction module: based on the optimized spectrum data and the normalized environmental data set, a preliminary association model and a multi-dimensional coupling model of the spectrum and the environmental characteristic parameters are constructed; An anomaly detection and correction module: anomaly detection is performed on the photoelectric performance prediction data, abnormal data is marked through statistical analysis, and the abnormal data is corrected using a weighted neighborhood interpolation method to generate corrected photoelectric performance prediction data; A multi-level feedback correction module: based on the corrected photoelectric performance prediction data and the normalized environmental data set, a multi-level real-time feedback correction model is constructed to optimize the corrected comprehensive spectrum data; A dynamic multi-dimensional photoelectric characteristic coupling module: a dynamic multi-dimensional photoelectric characteristic coupling model is constructed, and the corrected comprehensive spectrum data, the normalized environmental data set and the performance reference index set are integrated to predict the photoelectric performance, durability and environmental adaptability indicators of the LED lamp bead; A health management report generation module: the corrected comprehensive spectrum data is input into the dynamic multi-dimensional photoelectric characteristic coupling model to generate a health management report of the LED lamp bead, which includes a health status index, a predicted remaining life or a failure risk level.

[0048] In order to verify the actual application effect of the present application in the performance evaluation of LED lamp beads, the present application is applied to the product performance test laboratory of a large LED lamp bead manufacturing enterprise. The enterprise needs to test the performance of multiple batches of LED lamp beads, including comprehensive evaluation of multi-dimensional indicators such as photoelectric performance, durability and environmental adaptability. The traditional evaluation method is based on fixed sampling conditions and linear models, which cannot adapt to complex environmental changes and batch testing requirements, resulting in insufficient evaluation accuracy and low efficiency. In order to solve these problems, the enterprise introduces the dynamic multi-dimensional photoelectric characteristic coupling model performance evaluation method of the present application to comprehensively optimize the testing process.

[0049] In the experiment, the laboratory selected 50 groups of LED lamp bead samples, covering different specifications and models, and the test environment included three typical scenes: normal temperature, high humidity, and high temperature. The spectral data collection equipment used a high-sensitivity quantum dot spectrometer to record the spectral signal in real time, and the environmental parameter monitoring system collected external characteristic data such as environmental temperature and humidity, and light intensity. The laboratory collected spectral data from 50 samples under different test environments. Through quantum dot enhanced spectral sampling technology, the sampling parameters were adjusted in real time to optimize the sampling accuracy, and the original spectral data set was generated. Subsequently, based on the Bayesian optimization algorithm, the sampling parameters were adjusted to ensure that the sampling data reached a high signal-to-noise ratio under multiple environmental conditions. Through the abnormality detection module, abnormal data caused by equipment fluctuations or environmental disturbances were identified and corrected using the weighted neighborhood interpolation method to ensure data reliability.

[0050] The laboratory coupled the corrected spectral data with environmental parameters through a dynamic multi-dimensional photoelectric characteristic coupling model, and comprehensively analyzed the influence of photoelectric performance and environmental characteristics on LED lamp bead performance. In the model construction stage, the system used a dynamic characteristic function to fit the corrected spectral data, and combined with a feedback correction model to adjust the evaluation process in real time. The input of the model includes corrected comprehensive spectral data, normalized environmental data set and reference performance index, and the output is key evaluation results such as photoelectric performance, durability and environmental adaptability.

[0051] During the evaluation process, the system recorded the performance indicators of multiple batches of samples under three environments in real time, and the following is part of the experimental data: Table 1 Comparison of photoelectric performance evaluation data under different environments Environmental conditions Photoelectric performance index (lumens) Durability index (hours) Environmental adaptability score (1-100) Normal temperature 950±5 25,000±500 95 High humidity (80%) 890±10 22,000±700 85 High temperature (60°C) 860±15 18,000±800 75 Table 2 Comparison before and after abnormal data detection and correction Data category Proportion of abnormal data (%) Average error before correction (%) Average error after correction (%) Normal temperature environment data 5.5 4.2 1.8 High humidity environment data 6.8 5.6 2.1 High temperature environment data 7.3 6.3 2.4 Based on the performance evaluation results in Table 1, the system further generates a health management report, in which the durability index (hours) is directly used as the core basis for predicting the remaining life, and at the same time, the system calculates the health state index by weighting the photoelectric performance index and environmental adaptability score according to the preset health evaluation algorithm. For example, under high temperature (60℃) conditions, lower performance index (860 lumens) and adaptability score (75 points) will jointly result in a lower health state index (such as 68 / 100) and a higher high-risk failure level. In this way, the present application converts multi-dimensional performance data into intuitive and decision-guiding health management information.

[0052] The experimental results show that, in a normal temperature environment, the photoelectric performance and environmental adaptability score of the LED lamp bead are the highest, and the high temperature environment has the most significant influence on the performance, and the photoelectric performance and durability are significantly reduced. At the same time, through the abnormal data detection and correction module, the error of abnormal data is reduced from an average of 5.3% to 2.1%, and the reliability of the data is significantly improved.

[0053] In order to further verify the advantages of the present application, the laboratory compared the performance evaluation efficiency before and after introducing the technology of the present application: Table 3 comparison data of evaluation efficiency Test scenario Average single evaluation time (seconds) Multi-environment comprehensive evaluation time (minutes) Data processing accuracy improvement (%) Traditional method 120 40 - Method of the present application 65 18 35 As can be seen from Table 3, the present application method shortens the single evaluation time from 120 seconds of the traditional method to 65 seconds, reduces the multi-environment comprehensive evaluation time from 40 minutes to 18 minutes, and at the same time, the data processing accuracy is improved by 35%. The significant efficiency and accuracy improvement makes large-scale sample evaluation possible, meeting the production needs of enterprises.

[0054] In summary, the present application realizes precise fault prediction and quantitative health management of LED lamp beads through a dynamic multi-dimensional photoelectric characteristic coupling model, solves the problem of low efficiency and inaccurate evaluation results caused by environmental changes, data abnormalities and single evaluation model in the traditional method, and provides efficient and reliable technical support for quality control and performance optimization of LED lamp beads.

[0055] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for fault prediction and health management of LED lamp beads, characterized in that: The following steps are involved: S1. Using quantum dot enhanced spectral sampling technology and combining the spectral characteristics of quantum dot materials, the spectral signal emitted by LED lamp beads is sampled to generate raw spectral data; S2. Applying the Bayesian optimization algorithm to the original spectral data to dynamically adjust the sampling parameters, optimize the sampling conditions, and generate optimized spectral data; S3, real-time collection of external environmental characteristic parameters of LED lamp beads during operation, construction of a spectrum and environmental characteristic correlation model, and generation of environment-related spectrum data through multi-dimensional coupling of external environmental characteristic parameters and optimized spectrum data; S4. Model the environmental correlation spectral data based on the Gaussian process regression algorithm, extract the dynamic change characteristics of the photoelectric performance, and generate photoelectric performance prediction data; S5. Perform multi-level real-time feedback correction based on the optoelectronic performance prediction data, analyze the impact of environmental disturbances and abnormal data on performance evaluation, implement real-time correction and generate corrected comprehensive spectral data; S6. Construct a dynamic multi-dimensional photoelectric characteristic coupling model, input the corrected comprehensive spectral data into the dynamic multi-dimensional photoelectric characteristic coupling model, perform health status assessment and fault prediction on LED lamp beads, and generate a health management report including health status index, predicted remaining life or fault risk level.

2. A method for fault prediction and health management of LED lamp beads according to claim 1, characterized in that: The S2 specifically includes: S21. The original spectral data obtained by quantum dot enhanced spectral sampling technology is defined as the dataset , from the dataset Set the initial sampling parameter set in ,in represents the sampling frequency, represents the sampling time, Indicates the target range of signal-to-noise ratio; S22. Construct sampling parameter optimization objective function ,in , Represents the current sampling parameter set. The sampling parameter optimization objective function is measured by the error corresponding to the current sampling parameter set. and sampling integrity , combined with the dynamic nonlinear adjustment factor Perform calculations; S23. Use Bayesian optimization algorithm to optimize the objective function of sampling parameters Maximize and select the next sampling parameter set ; S24: Apply the selected next sampling parameter set For the original spectral data Sampling to generate new spectral data , and calculate the new spectral data Error metric and sampling integrity ; S25, repeat and , iteratively update the sampling parameter set , until the sampling parameters optimize the objective function Converges to the preset maximum number of iterations ; S26. Output optimized sampling parameter set and the corresponding optimized spectral data ,in Satisfy the optimal requirements of error measurement and sampling integrity of spectral data.

3. A method for fault prediction and health management of LED lamp beads according to claim 2, characterized in that: The S22 specifically includes: S221, based on the initial sampling parameter set and raw spectral data The statistical characteristics of the sampling frequency , sampling time and signal-to-noise ratio targets Dynamic weights, construct error metrics ,in Indicates the current sampling parameter set; S222, combined with original spectral data Signal coverage characteristics, refined sampling integrity , used to describe the current sampling parameter set Coverage effect on spectral signals; S223. Error metric based on definition and sampling integrity , combined with the dynamic nonlinear adjustment factor , construct sampling parameter optimization objective function .

4. A method for fault prediction and health management of LED lamp beads according to claim 2, characterized in that: The S3 specifically includes: S31. Real-time collection of external environmental characteristic parameters of LED lamp beads during operation, and organization of the collected external environmental characteristic parameters into an environmental data set , for environmental data sets Each parameter in is normalized to obtain a normalized environmental data set ; S32, based on optimized spectral data and normalized environmental data sets , construct a preliminary correlation model between spectral data and external environmental characteristic parameters, use multiple linear regression method to describe the preliminary relationship between optimized spectral data and external environmental characteristic parameters and obtain the corrected spectral value; S33. Using a preliminary correlation model between the spectral data and the external environmental characteristic parameters, and employing a multidimensional nonlinear regression method, a multidimensional coupling model of the spectral data and the external environmental characteristic parameters is constructed, wherein the goal of the multidimensional coupling model is to minimize the spectral value correction error; S34, collect and normalize the normalized environmental data in real time and optimize spectral data Input multidimensional coupling model to generate environmental correlation spectral data .

5. A method for fault prediction and health management of LED lamp beads according to claim 4, characterized in that: The S32 specifically includes: S321. Based on normalized environmental data set , extract the characteristics of each external environment characteristic parameter and define the characteristic function , used to describe the linear influence of external environmental characteristic parameters on spectral values; S322, combined with optimized spectral data The distribution characteristics of the spectral data are defined to optimize the characteristic function ; S323, based on the defined characteristic function and , construct a preliminary correlation model between spectral data and external environmental characteristic parameters, and use the multiple linear regression method to describe the linear relationship between the optimized spectral data and external environmental characteristic parameters; S324, Optimizing the parameters of the multivariate linear regression method using the least squares method and , the fitting goal is to minimize the error of the regression model.

6. A method for fault prediction and health management of LED lamp beads according to claim 4, characterized in that: The S33 specifically includes: S331. Construct a multidimensional nonlinear regression model of spectral data and external environmental characteristic parameters , the input of the multidimensional nonlinear regression model is the normalized environmental data set and optimize spectral data , the output is the corrected spectral value, and the goal of the multidimensional nonlinear regression model is to minimize the spectral value correction error; S332. Initialize model parameter set , including the weight vector and the bias vector , the convergence speed of the optimization process; S333, Objective function of multidimensional nonlinear regression model based on gradient descent method Optimize and update the model parameter set , each iteration is performed by gradient descent; S334. Determine the objective function of the multidimensional nonlinear regression model The iteration is terminated when one of the following conditions is met: The objective function value of the iteration The absolute difference between the objective function values ​​of the iterations is less than the preset convergence threshold , or the current number of iterations exceeds the maximum number of iterations , it is determined that the convergence condition is met and the iteration is stopped; S335, the optimized model parameter set Substitute into the multidimensional nonlinear regression model , generating the final multidimensional coupled model.

7. A method for fault prediction and health management of LED lamp beads according to claim 4, characterized in that: The S5 specifically includes: S51, based on the photoelectric performance prediction data , build anomaly detection function , detect and mark outliers in the data. When the absolute value of the deviation between a data point and the mean of the optoelectronic performance prediction data set is greater than the product of the preset anomaly detection threshold multiple and the standard deviation, the data point is judged to be abnormal and the abnormal mark value is 1. If the absolute value of the deviation does not exceed the preset anomaly detection threshold coefficient, the data point is judged to be normal and the abnormal mark value is 0; S52: For the detected abnormal data, The data points are corrected using the weighted neighborhood interpolation method, and the corrected data points are combined with the normal data points to generate the corrected photoelectric performance prediction data. ; S53, based on the corrected photoelectric performance prediction data and normalized environmental data sets , combined with experimentally determined reference spectral data , building a multi-level feedback correction model The goal of the multi-level feedback correction model is to minimize the corrected comprehensive spectral data With reference spectral data The error between S54. Optimizing the multi-level feedback correction model using gradient descent method Parameter set , to minimize the multi-level feedback correction model The objective function After the optimization is completed, the corrected photoelectric performance prediction data set and normalized environmental data sets Input the multi-level feedback correction model to generate the corrected comprehensive spectral data .

8. The method for fault prediction and health management of LED lamp beads according to claim 7, characterized in that: The S6 specifically includes: S61, based on the corrected comprehensive spectral data , normalized environmental data set , and a set of performance reference indicators extracted from the performance database , define the input variables and output variables of the dynamic multi-dimensional optoelectronic characteristic coupling model, where the input variables include 、 and ,The output variables include the health management report Y, which contains the health status index, the predicted remaining life or the failure risk level; S62. Constructing a characteristic function set of a dynamic multi-dimensional optoelectronic characteristic coupling model , used to describe the corrected comprehensive spectral data and normalized environmental data sets Health management report The nonlinear effect of S63. Using characteristic function sets Constructing a dynamic multidimensional optoelectronic characteristic coupling model with nonlinear fitting ,The goal of the dynamic multi-dimensional optoelectronic characteristic coupling model is to ,predict the health status index, remaining life or failure risk level through ,nonlinear fitting; S64. Health Management Report The calculation of the corrected comprehensive spectral data , normalized environmental data set and performance reference indicator set Input dynamic multi-dimensional optoelectronic characteristic coupling model , generate a health management report Y* containing health status index, predicted remaining life or failure risk level.

9. A fault prediction and health management system for LED lamp beads, used to implement a fault prediction and health management method for LED lamp beads according to any one of claims 1 to 8, characterized in that: Includes the following modules: Spectral sampling module: uses quantum dot enhanced spectral sampling technology to collect spectral data of LED lamp beads and generate original spectral data; Sampling parameter optimization module: dynamically adjusts sampling parameters through Bayesian optimization algorithm to generate optimized spectral data; Environmental characteristic acquisition module: collects the external environmental characteristic parameters of LED lamp beads in real time during operation, and normalizes the collected external environmental characteristic data to generate a normalized environmental data set; Spectral and environmental coupling model construction module: Based on the optimized spectral data and normalized environmental data sets, a preliminary correlation model and a multidimensional coupling model of spectral and environmental characteristic parameters are constructed; Anomaly detection and correction module: performs anomaly detection on the optoelectronic performance prediction data, marks the abnormal data through statistical analysis, and corrects the abnormal data using the weighted neighborhood interpolation method to generate the corrected optoelectronic performance prediction data; Multi-level feedback correction module: Based on the corrected optoelectronic performance prediction data and the normalized environmental data set, a multi-level real-time feedback correction model is constructed to optimize the corrected comprehensive spectral data; Dynamic multi-dimensional photoelectric characteristic coupling module: Build a dynamic multi-dimensional photoelectric characteristic coupling model, integrate the calibrated comprehensive spectrum data, normalized environmental data set and performance reference index set, and predict the photoelectric performance, durability and environmental adaptability indicators of LED lamp beads; Health management report generation module: Input the corrected comprehensive spectral data into the dynamic multi-dimensional photoelectric characteristic coupling model to generate a health management report for the LED lamp beads, which includes the health status index, predicted remaining life or failure risk level.

Images