High-brightness light source temperature real-time monitoring method, device and equipment

Abstract

The invention relates to the technical field of semiconductor light emitting diodes, and provides a high-brightness light source temperature real-time monitoring method, device and equipment, and the method comprises the steps: carrying out the spectrum collection of output light of a high-brightness light source, obtaining real-time spectrum data and a corresponding time stamp, carrying out the temperature prediction of the real-time spectrum data, and obtaining a current temperature value and prediction confidence; and comparing the current temperature value with a preset threshold value to obtain a temperature deviation value, adjusting and monitoring the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence coefficient, obtaining real-time working state data and a temperature change trend, and performing parameter adjustment on the deep learning model to obtain an updated deep learning model. Based on the updated deep learning model, temperature prediction is carried out on subsequent real-time spectral data, efficient and accurate temperature monitoring is achieved, and the problems that in practical application, in the using process of a temperature sensor, the precision is not high, the response speed is low, and the sensor is prone to being damaged are solved.

Description

Technical Field

[0001] The present application relates to the technical field of semiconductor light emitting diodes, and in particular to a method, device and equipment for real-time temperature monitoring of high-brightness light sources. Background Art

[0002] With the continuous advancement of high-brightness light source technology, light source equipment has been widely used in many application fields, such as stage lighting, automotive headlights and industrial lighting. While providing efficient lighting, these high-brightness light source equipment also faces temperature management issues caused by long-term operation or high power output. Ensuring that the light source temperature is stable within a safe range is not only crucial to the performance and life of the equipment, but also an important part of ensuring user safety.

[0003] In the related technical means, traditional temperature sensors are used for monitoring. These sensors directly measure the operating temperature of the light source through physical contact or non-contact methods, and feed the data back to the control system. In this way, the system can adjust the working state of the light source in real time, such as reducing power output or increasing heat dissipation, to maintain the temperature of the light source within a preset safe range.

[0004] Regarding the above technical solution, although the temperature monitoring of high-brightness light sources can be achieved through the use of traditional temperature sensors, in actual applications, the temperature sensors have problems such as low accuracy, slow response speed and easy damage to the sensors themselves during use. Summary of the invention

[0005] In order to improve the problems of low accuracy, slow response speed and easy damage of temperature sensors during use in practical applications, the present application provides a method, device and equipment for real-time monitoring of the temperature of high-brightness light sources using spectral data and deep learning models, which can achieve more efficient and accurate temperature management by predicting the temperature changes of the light source.

[0006] The present application provides a method for real-time monitoring the temperature of a high-brightness light source, comprising: performing spectral acquisition on the output light of the high-brightness light source to obtain real-time spectral data and corresponding time tags; performing temperature prediction on the real-time spectral data by using a preset deep learning model to obtain a current temperature value and a prediction confidence; comparing the current temperature value with a preset safety temperature threshold to obtain a temperature deviation value, and adjusting the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain adjusted driving parameters; using the time tags, synchronously monitoring the adjusted driving parameters to obtain real-time working status data and temperature change trends; adjusting the parameters of the deep learning model according to the real-time working status data and the temperature change trends to obtain an updated deep learning model; and performing temperature prediction on subsequent real-time spectral data based on the updated deep learning model to achieve real-time monitoring of the temperature of the high-brightness light source.

[0007] As a preferred solution, the step of performing spectral acquisition on the output light of the high-brightness light source to obtain real-time spectral data and corresponding time tags includes: using a high-resolution spectrum analyzer and a high-speed data acquisition module to synchronously perform spectral acquisition on the output light of the high-brightness light source to obtain original spectral data and an initial time signal, performing dark current correction and baseline correction on the original spectral data to obtain corrected spectral data and baseline drift parameters; comparing the corrected spectral data with a preset reference spectrum to extract a spectral offset and a spectral distortion coefficient, performing spectral correction on the corrected spectral data using the spectral offset and the spectral distortion coefficient to obtain standardized spectral data, and fusing the standardized spectral data with the baseline drift parameters to obtain real-time spectral feature data; synchronizing the initial time signal with the analysis time of the high-resolution spectrum analyzer through a time synchronization algorithm to obtain an accurate time tag, and associating the real-time spectral feature data with the accurate time tag to obtain real-time spectral data and a corresponding time tag.

[0008] As a preferred embodiment, the deep learning model includes a convolutional neural network model, a first long short-term memory network model and a second long short-term memory network model. The step of performing temperature prediction on the real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence includes: inputting the real-time spectral data into the convolutional neural network model, extracting feature response values, and using the feature response values ​​to calculate feature weights to obtain a main feature set and a secondary feature set; inputting the main feature set into the first long short-term memory network model to obtain a main temperature prediction value and a main feature confidence, and inputting the secondary feature set into the second long short-term memory network model to obtain a secondary temperature prediction value and a secondary feature confidence; performing weighted fusion of the main temperature prediction value and the secondary temperature prediction value based on the main feature confidence and the secondary feature confidence to obtain a fused temperature prediction value and a comprehensive confidence, using a residual network to perform error correction on the fused temperature prediction value to obtain a current temperature value and a correction factor, and combining the comprehensive confidence with the correction factor to calculate the prediction confidence.

[0009] As a preferred solution, the step of comparing the current temperature value with a preset safety temperature threshold to obtain a temperature deviation value, adjusting the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain adjusted driving parameters includes: comparing the current temperature value with the preset upper and lower limits of the safety temperature threshold to obtain an over-temperature deviation value and an under-temperature deviation value, determining the temperature adjustment requirement and the adjustment priority according to the over-temperature deviation value and the under-temperature deviation value in combination with the prediction confidence; processing the temperature adjustment requirement and the adjustment priority using a fuzzy control algorithm to obtain an initial current adjustment amount, comparing the initial current adjustment amount with historical driving current data, and calculating an adaptation coefficient based on the comparison result; optimizing the initial current adjustment amount using the adaptation coefficient to obtain an optimized current adjustment amount, applying the optimized current adjustment amount to the current driving current, calculating a current adjustment curve, and generating a driving control instruction according to the current adjustment curve; sending the driving control instruction to a preset driving control module to adjust the driving current of the high-brightness light source to obtain adjusted driving parameters.

[0010] As a preferred solution, the step of using the time mark to synchronously monitor the adjusted driving parameters to obtain real-time working status data and temperature change trends includes: acquiring the execution time of the adjusted driving parameters through a preset data acquisition module to obtain an execution time mark, and using the time mark to time align and synchronize the execution time mark to obtain synchronous driving data; collecting the operating status parameters of the high-brightness light source to obtain operating status data, and correlating and analyzing the operating status data with the synchronous driving data to obtain real-time working status data and correlation factors; and using the correlation factors to perform trend analysis on the real-time working status data to obtain temperature change trends.

[0011] As a preferred scheme, the step of adjusting the parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model includes: extracting key parameters in the real-time working status data and combining them with the temperature change trend to generate a model adjustment data set and feature adjustment parameters; using an online incremental learning algorithm to input the model adjustment data set into the deep learning model, updating the weight parameters and bias parameters of the deep learning model, and obtaining a preliminary optimized deep learning model; using the feature adjustment parameters to reselect the main feature set and the secondary feature set of the preliminary optimized deep learning model, updating the feature subspace, and obtaining an updated deep learning model.

[0012] As a preferred scheme, after the step of using the feature adjustment parameters to reselect the main feature set and the secondary feature set of the preliminarily optimized deep learning model, updating the feature subspace, and obtaining the updated deep learning model, it includes: using a preset validation set to evaluate the updated deep learning model to obtain the accuracy, stability and generalization ability indicators of the updated deep learning model; if the accuracy, stability and generalization ability indicators of the updated deep learning model meet the preset requirements, saving the updated deep learning model parameters and structure to generate a new model version; deploying the new model version in the system to replace the old deep learning model to achieve online update.

[0013] The present application also provides a high-brightness light source temperature real-time monitoring device, including: an acquisition module, used to perform spectral acquisition on the output light of the high-brightness light source to obtain real-time spectral data and corresponding time tags; a prediction module, used to perform temperature prediction on the real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence; a comparison module, used to compare the current temperature value with a preset safety temperature threshold to obtain a temperature deviation value, and adjust the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter; a monitoring module, used to synchronously monitor the adjusted driving parameter using the time tag to obtain real-time working status data and temperature change trend; an adjustment module, used to adjust the parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model; a monitoring module, used to perform temperature prediction on subsequent real-time spectral data based on the updated deep learning model to achieve real-time monitoring of the temperature of the high-brightness light source.

[0014] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the method for real-time temperature monitoring of a high-brightness light source described above is implemented.

[0015] Compared with the prior art, the present application has the following beneficial effects: high accuracy and fast response speed. By collecting spectral data and time stamps in real time and using a trained deep learning model to predict temperature, the current temperature value and prediction confidence can be quickly obtained; according to the temperature deviation value and prediction confidence, the driving current of the light source is adjusted in real time to keep the temperature within a safe range, and then the temperature change trend is obtained by synchronously monitoring the adjusted driving parameters and working status data, and the parameters of the deep learning model are optimized based on these data to continuously improve the prediction accuracy and adaptability; based on the updated deep learning model, the temperature of subsequent real-time spectral data is predicted to achieve efficient and accurate real-time temperature monitoring, and improve the actual application. In the process of using temperature sensors, there are problems such as low accuracy, slow response speed and easy damage of the sensor itself. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0017] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0018] Figure 1 It is a flow chart of a method for real-time temperature monitoring of a high-brightness light source provided by an embodiment of the present invention; Figure 2 is a schematic block diagram of the structure of a device for real-time monitoring temperature of a high-brightness light source provided by an embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention.

[0019] Description of reference numerals: 10. High-brightness light source temperature real-time monitoring device; 11. Acquisition module; 12. Prediction module; 13. Comparison module; 14. Monitoring module; 15. Adjustment module; 16. Monitoring module; 20. Electronic equipment; 21. Memory; 22. Processor. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0022] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0023] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0024] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0025] Embodiment 1: like Figure 1 As shown, the method for real-time monitoring temperature of a high-brightness light source provided in the embodiment of the present application includes steps S100 to S600.

[0026] Step S100: spectrally collect the output light of the high-brightness light source to obtain real-time spectral data and corresponding time tags.

[0027] In this step, a high-precision spectrometer is used to collect the output light of the high-brightness light source to ensure that the collected spectral data has sufficient resolution and sensitivity; specifically, the light of the high-brightness light source is guided to the spectrometer through optical fiber, and the spectral signal output by the spectrometer is converted into a digital signal through a data acquisition card to generate real-time spectral data and time stamps.

[0028] For example, using a specific model of spectrometer and data acquisition card can improve the speed and accuracy of spectral data acquisition.

[0029] Step S200: Use a preset deep learning model to perform temperature prediction on real-time spectral data to obtain a current temperature value and a prediction confidence level.

[0030] In this step, the collected real-time spectral data is analyzed using a trained deep learning model to predict the current temperature of the high-brightness light source. Specifically, the real-time spectral data is input into the deep learning model, and the current temperature value and the corresponding prediction confidence are output through the multi-layer neural network structure of the model.

[0031] For example, using a trained convolutional neural network (CNN) model can effectively improve the accuracy and reliability of temperature prediction.

[0032] Step S300: Compare the current temperature value with the preset safety temperature threshold to obtain a temperature deviation value, and adjust the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter.

[0033] In this step, the current temperature value is compared with the preset safety temperature threshold in real time through the control system to calculate the temperature deviation value; specifically, the driving current of the high-brightness light source is adjusted according to the temperature deviation value and the prediction confidence to ensure that the temperature is controlled within a safe range.

[0034] For example, when the temperature deviation value is large and the prediction confidence is high, the driving current is rapidly reduced to reduce the light source temperature.

[0035] Step S400: Use time stamps to synchronously monitor the adjusted drive parameters to obtain real-time working status data and temperature change trends.

[0036] In this step, the adjusted driving parameters are synchronized with the collected real-time working status data using time markers to monitor the working status and temperature change trend of the high-brightness light source; specifically, the driving parameters and working status data are recorded in real time through the control system, and the temperature change trend is analyzed.

[0037] For example, by monitoring data, we can discover the periodic characteristics and abnormal fluctuations of temperature changes and adjust the control strategy in time.

[0038] Step S500: adjust the parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model.

[0039] In this step, the parameters of the deep learning model are optimized and adjusted according to the working status data and temperature change trends collected in real time to improve the prediction accuracy of the model; specifically, by continuously updating the model's training data and optimization algorithm, the deep learning model can maintain efficient prediction capabilities under different working conditions.

[0040] For example, through incremental learning methods, model parameters are updated in real time to enhance the adaptability and robustness of the model.

[0041] Step S600: Based on the updated deep learning model, temperature prediction is performed on subsequent real-time spectral data to achieve real-time monitoring of the temperature of the high-brightness light source.

[0042] In this step, the updated deep learning model is used to analyze and predict the subsequently collected real-time spectral data to ensure the accuracy and timeliness of temperature monitoring; specifically, the real-time spectral data is input into the updated model to output the new current temperature value and prediction confidence.

[0043] For example, the updated model can more accurately reflect the current working environment and light source status, and improve the reliability of temperature monitoring.

[0044] In this embodiment, the output light of the high-brightness light source is spectrally collected to obtain real-time spectral data and corresponding time tags. Then, the real-time spectral data is temperature predicted using a preset deep learning model to obtain the current temperature value and the prediction confidence. Next, the current temperature value is compared with the preset safety temperature threshold to obtain the temperature deviation value, and the driving current of the high-brightness light source is adjusted according to the temperature deviation value and the prediction confidence to obtain the adjusted driving parameters. The adjusted driving parameters are synchronously monitored using time tags to obtain real-time working status data and temperature change trends. Finally, according to the real-time working status data and temperature change trends, the deep learning model is parameter adjusted to update the model, and the subsequent real-time spectral data is temperature predicted based on the updated deep learning model, thereby achieving real-time monitoring of the temperature of the high-brightness light source.

[0045] By introducing deep learning models and real-time spectral data, the temperature of high-brightness light sources is accurately predicted and controlled, effectively improving the response speed and accuracy of temperature monitoring. At the same time, through comprehensive consideration of temperature deviation values ​​and prediction confidence, the driving parameter adjustment of high-brightness light sources is optimized to avoid equipment damage and performance degradation caused by excessive temperature. In addition, the deep learning model is dynamically adjusted through real-time working status data and temperature change trends to ensure the adaptability and reliability of the temperature monitoring system under different working conditions, and to improve the problems of low accuracy, slow response speed and easy damage of the temperature sensor itself during use in actual applications.

[0046] Embodiment 2: In step S100, a high-resolution spectrum analyzer and a high-speed data acquisition module are used to synchronously acquire the spectrum of the output light of the high-brightness light source to obtain raw spectrum data and an initial time signal, and dark current correction and baseline correction are performed on the raw spectrum data to obtain corrected spectrum data and baseline drift parameters.

[0047] By using a high-resolution spectrometer to capture the output spectrum of a high-brightness light source, the accuracy and resolution of the spectral data can be ensured; specifically, in combination with a high-speed data acquisition module, real-time acquisition of spectral data and synchronous generation of the initial time signal can be achieved, thereby providing accurate time marking of the original spectral data.

[0048] For example, using the SR-750-B1 high-resolution spectrometer from Andor Technology and the PXI-5154 high-speed data acquisition module from NI can significantly improve the speed and accuracy of spectral data acquisition.

[0049] The corrected spectral data is compared with the preset reference spectrum, the spectral offset and the spectral distortion coefficient are extracted, the spectral offset and the spectral distortion coefficient are used to perform spectral correction on the corrected spectral data to obtain standardized spectral data, and the standardized spectral data is fused with the baseline drift parameter to obtain real-time spectral feature data.

[0050] By comparing the corrected spectral data with the reference spectrum, the spectral offset and spectral distortion coefficient are extracted and corrected to standardize the spectral data and eliminate systematic errors. Specifically, the standardized spectral data are fused with the baseline drift parameters to generate real-time spectral feature data that can reflect the actual working status.

[0051] For example, by using MATLAB's spectral analysis toolbox, the comparison and correction process of spectral data can be automated to ensure the efficiency and accuracy of data processing.

[0052] The initial time signal is synchronized with the analysis time of the high-resolution spectrum analyzer through a time synchronization algorithm to obtain an accurate time tag, and the real-time spectrum feature data and the accurate time tag are associated to obtain real-time spectrum data and the corresponding time tag.

[0053] By adopting a time synchronization algorithm, the initial time signal is ensured to be precisely aligned with the analysis time of the spectrum analyzer to achieve accurate synchronization of the time mark of the data; specifically, a synchronous clock signal and timestamp technology are used to accurately mark the real-time spectral feature data and generate an associated time mark.

[0054] For example, using GPS synchronized clocks and high-precision timestamp modules can achieve high-precision time synchronization between different devices, thereby improving the timing consistency of data.

[0055] In step S200, the deep learning model includes a convolutional neural network model, a first long short-term memory network model, and a second long short-term memory network model; wherein the step of performing temperature prediction on real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence includes: The real-time spectral data is input into the convolutional neural network model, the characteristic response values ​​are extracted, and the characteristic weights are calculated using the characteristic response values ​​to obtain the main feature set and the secondary feature set.

[0056] Through the convolutional neural network model, convolution operation is performed on the input real-time spectral data to extract multi-level feature response values, and the main features and secondary features are screened out through weight calculation; specifically, the convolution kernel is used to perform local perception and feature extraction on the spectral data, and the pooling layer is used to reduce the dimension and abstract it, and finally the main feature set and the secondary feature set are output.

[0057] For example, using the convolutional neural network model in the TensorFlow framework, the entire process of feature extraction and response value calculation can be completed efficiently.

[0058] The main feature set is input into the first long short-term memory network model to obtain the main temperature prediction value and the main feature confidence, and the secondary feature set is input into the second long short-term memory network model to obtain the secondary temperature prediction value and the secondary feature confidence.

[0059] By inputting the main feature set into the first long short-term memory network model, analyzing its time series relationship, and outputting the main temperature prediction value and the main feature confidence; specifically, using the LSTM unit to capture and memorize the time series information in the feature set, the main temperature prediction value and the corresponding confidence are calculated. At the same time, the secondary feature set is input into the second long short-term memory network model to obtain the secondary temperature prediction value and the secondary feature confidence.

[0060] For example, the LSTM model in the PyTorch framework can effectively process the time correlation in spectral data and improve the accuracy of temperature prediction.

[0061] Based on the main feature confidence and the secondary feature confidence, the main temperature prediction value and the secondary temperature prediction value are weightedly fused to obtain the fused temperature prediction value and the comprehensive confidence. The residual network is used to perform error correction on the fused temperature prediction value to obtain the current temperature value and the correction factor. The comprehensive confidence is combined with the correction factor to calculate the prediction confidence.

[0062] By adopting the weighted fusion method, the temperature prediction value is weighted based on the main feature confidence and the secondary feature confidence to generate a fused temperature prediction value and a comprehensive confidence; specifically, the fused temperature prediction value is further corrected through the residual network to obtain a more accurate current temperature value and correction factor, and the final prediction confidence is calculated by combining the comprehensive confidence and the correction factor.

[0063] For example, using the residual network structure in Keras to refine and correct the predicted values ​​can significantly improve the accuracy of temperature prediction and the robustness of the model.

[0064] In step S300, the current temperature value is compared with the preset upper and lower limits of the safety temperature threshold to obtain the over-temperature deviation value and the under-temperature deviation value. According to the over-temperature deviation value and the under-temperature deviation value, combined with the prediction confidence, the temperature adjustment requirement and adjustment priority are determined.

[0065] By comparing the current temperature value with the upper and lower limits of the safety temperature threshold, the over-temperature deviation value and under-temperature deviation value are accurately calculated; specifically, based on the prediction confidence, the adjustment requirements and priorities of over-temperature and under-temperature conditions are comprehensively evaluated to determine the optimal temperature adjustment solution.

[0066] For example, using numerical analysis tools to perform real-time calculations of temperature deviations can help quickly make decisions on temperature regulation strategies to ensure stable operation of high-brightness light sources.

[0067] The temperature regulation requirements and regulation priorities are processed by a fuzzy control algorithm to obtain an initial current regulation amount, which is then compared with historical drive current data, and an adaptation coefficient is calculated based on the comparison result.

[0068] By applying the fuzzy control algorithm, the temperature regulation requirements and priorities are comprehensively processed to generate the initial current regulation amount; specifically, by comparing the initial current regulation amount with the historical driving current data, the adaptation coefficient is calculated to improve the adaptability and stability of the regulation scheme.

[0069] For example, using the fuzzy control toolbox in Simulink, you can quickly build and optimize a fuzzy control system to ensure accurate calculation of the regulation variable.

[0070] The initial current regulation amount is optimized by using the adaptation coefficient to obtain an optimized current regulation amount, the optimized current regulation amount is applied to the current driving current, a current regulation curve is calculated, and a driving control instruction is generated according to the current regulation curve.

[0071] By using the adaptation coefficient, the initial current regulation amount is further optimized to generate a more adaptable current regulation amount; specifically, the optimized current regulation amount is applied to the current driving current, and the current regulation curve is calculated and generated, thereby deriving the corresponding driving control instruction.

[0072] For example, using the control system design tools in Matlab, current regulation parameters can be accurately optimized and drive control instructions can be generated in real time.

[0073] The driving control instruction is sent to a preset driving control module to adjust the driving current of the high-brightness light source to obtain the adjusted driving parameters.

[0074] By sending the generated driving control instruction to the driving control module of the high-brightness light source, the driving current of the light source is adjusted in real time to respond to the current temperature regulation requirements; specifically, the driving current is adjusted in real time by the driving control module to ensure that the light source works stably within a safe temperature range.

[0075] For example, using a dedicated high-precision current controller can achieve precise adjustment of the driving current, thereby ensuring the stability and safety of high-brightness light sources.

[0076] In step S400, the execution time of the adjusted driving parameters is acquired through a preset data acquisition module to obtain an execution time mark, and the execution time mark is time-aligned and synchronized using the time mark to obtain synchronous driving data.

[0077] By using the data acquisition module, the execution time of the adjusted driving parameters is recorded and a corresponding execution time mark is generated; specifically, the execution time mark is aligned and synchronized using the time mark, thereby ensuring the consistency and accuracy of the driving data.

[0078] For example, using LabVIEW's real-time data acquisition system, you can efficiently complete the acquisition and synchronization tasks of execution time stamps.

[0079] The operating status parameters of the high-brightness light source are collected to obtain operating status data, and the operating status data is correlated and analyzed with the synchronous drive data to obtain real-time working status data and correlation factors.

[0080] By collecting the operating status parameters of the high-brightness light source, including key parameters such as the voltage, current, and temperature of the light source, the operating status data of the equipment is obtained in real time; specifically, these operating status data are correlated and analyzed with the synchronous drive data to identify the relationship between the drive parameters and the light source performance, and generate real-time working status data and correlation factors.

[0081] For example, using NI's PXIe-4081 high-precision digital multimeter, it is possible to simultaneously collect multiple parameters such as voltage and current, and perform comprehensive analysis with synchronous drive data.

[0082] The correlation factors are used to perform trend analysis on the real-time working status data to obtain the temperature change trend.

[0083] By utilizing correlation factors, the real-time working status data is analyzed to identify and predict the temperature change trend; specifically, statistical analysis and algorithm modeling techniques are applied to comprehensively process historical and real-time data to generate a temperature change trend graph.

[0084] For example, using Python's pandas and matplotlib libraries for data analysis and visualization can clearly present the periodicity and trend of temperature changes.

[0085] In step S500, key parameters in the real-time working status data are extracted and combined with the temperature change trend to generate a model adjustment data set and feature adjustment parameters.

[0086] By extracting key parameters from real-time working status data and combining them with temperature change trends, data sets and feature adjustment parameters for model adjustment are generated; specifically, real-time data is filtered and aggregated to identify the most valuable features for model optimization, and adjustment data sets and parameters are generated.

[0087] For example, using SQL database systems to query and extract large-scale data can effectively handle complex data integration tasks.

[0088] Using the online incremental learning algorithm, the model adjustment data set is input into the deep learning model, the weight parameters and bias parameters of the deep learning model are updated, and a preliminary optimized deep learning model is obtained.

[0089] By applying the online incremental learning algorithm, the generated model adjustment dataset is used to update and optimize the deep learning model and adjust the weight and bias parameters of the model. Specifically, by gradually learning and adjusting the model parameters, it is ensured that the model can reflect new data features in a timely manner and improve the prediction accuracy.

[0090] For example, using the incremental learning module of scikit-learn, model parameters can be updated batch by batch, avoiding full retraining and improving model update efficiency.

[0091] The feature adjustment parameters are used to reselect the main feature set and the secondary feature set of the initially optimized deep learning model, update the feature subspace, and obtain the updated deep learning model.

[0092] By utilizing feature adjustment parameters, the initially optimized deep learning model is further optimized, the main feature set and the secondary feature set are reselected and adjusted, and the feature subspace of the model is updated; specifically, the feature representation and subspace structure of the model are optimized through the feature selection algorithm to ensure that the updated model has stronger predictive ability and adaptability.

[0093] For example, applying the L1 regularization method for feature selection can effectively reduce redundant features and improve the robustness of the model.

[0094] Among them, after the step of reselecting the main feature set and the secondary feature set of the initially optimized deep learning model by using the feature adjustment parameters and updating the feature subspace to obtain the updated deep learning model, it includes: The updated deep learning model is evaluated using the preset validation set to obtain the accuracy, stability, and generalization ability indicators of the updated deep learning model.

[0095] By using the preset validation set, the updated deep learning model is comprehensively evaluated to test the accuracy, stability and generalization ability of the model. Specifically, the model is quantitatively evaluated through standard evaluation indicators such as accuracy, F1 value, AUC, etc. to ensure its reliability in practical applications.

[0096] For example, using TensorFlow's validation tools, you can systematically evaluate the performance of an updated model to ensure that it meets the preset requirements.

[0097] If the accuracy, stability, and generalization ability indicators of the updated deep learning model meet the preset requirements, the updated deep learning model parameters and structure are saved to generate a new model version.

[0098] By analyzing the evaluation results, if the various indicators of the model meet the preset requirements, the updated model parameters and structure will be saved to generate a new model version; specifically, persistent storage technology is used to save the model's weight parameters, network structure and other information to ensure the traceability and reusability of the model version.

[0099] For example, saving model data in HDF5 format can facilitate migration and deployment between different systems.

[0100] Deploy the new model version in the system to replace the old deep learning model and implement online update.

[0101] By deploying the new model version to the actual application system to replace the old deep learning model, we ensure that the system can apply the latest model in real time for prediction and control; specifically, we use automated deployment tools to implement online model updates to maintain the continuity and efficiency of the system.

[0102] For example, using Docker container technology, deep learning models can be quickly deployed and updated in production environments, ensuring a smooth transition and efficient operation of the system.

[0103] In this embodiment, first, by using a high-resolution spectrum analyzer and a high-speed data acquisition module, the output light of the high-brightness light source is synchronously spectrum collected to ensure the accuracy of the spectrum data and the synchronization of the time mark. In addition, by performing dark current correction and baseline correction on the original spectrum data, and comparing it with the reference spectrum, the spectrum offset and spectrum distortion coefficient are extracted, and the spectrum correction and standardization are further performed to obtain high-precision real-time spectrum feature data. The application of the time synchronization algorithm enables the initial time signal to be accurately aligned with the spectrum analysis time, ensuring the accuracy of the data mark, and by associating the real-time spectrum feature data with the time mark, the real-time spectrum data and the corresponding time mark are generated.

[0104] Secondly, the convolutional neural network model is used to extract the characteristic response value of the real-time spectral data, and the long short-term memory network model is used to process the main and secondary feature sets to obtain the main and secondary temperature prediction values ​​and their confidences respectively. The predicted values ​​are corrected by weighted fusion and residual network, and finally the current temperature value and prediction confidence are generated to ensure the high accuracy and reliability of temperature prediction. In the temperature adjustment step, the temperature adjustment requirements and priorities are determined by comparing the current temperature value with the preset safety temperature threshold, and the fuzzy control algorithm and adaptation coefficient are used to optimize the initial current adjustment amount, generate the current adjustment curve and drive control instructions, and ensure the effective control of the temperature of the high-brightness light source. The execution time of the adjusted drive parameters is obtained through the data acquisition module, and combined with the operating status parameters of the high-brightness light source, correlation analysis and trend prediction are performed to generate real-time working status data and temperature change trends.

[0105] Finally, in the optimization process of the deep learning model, key parameters are extracted, and the online incremental learning algorithm and feature adjustment parameters are used to dynamically adjust and update the model to ensure its efficient prediction ability under different working conditions. Finally, by verifying and evaluating the updated deep learning model to ensure its accuracy, stability and generalization ability, the old model is replaced to achieve online update, thereby achieving accurate real-time monitoring and effective management of the temperature of high-brightness light sources.

[0106] Embodiment 3: like Figure 2 As shown, the present application also provides a high-brightness light source temperature real-time monitoring device 10, including a collection module 11, a prediction module 12, a comparison module 13, a monitoring module 14, an adjustment module 15 and a monitoring module 16.

[0107] The acquisition module 11 is mainly used to acquire the spectrum of the output light of the high-brightness light source to obtain real-time spectrum data and corresponding time tags.

[0108] The prediction module 12 is mainly used to perform temperature prediction on real-time spectral data through a preset deep learning model to obtain the current temperature value and prediction confidence.

[0109] The comparison module 13 is mainly used to compare the current temperature value with the preset safety temperature threshold value to obtain a temperature deviation value, and adjust the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter.

[0110] The monitoring module 14 is mainly used to synchronously monitor the adjusted driving parameters using time stamps to obtain real-time working status data and temperature change trends.

[0111] The adjustment module 15 is mainly used to adjust the parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model.

[0112] The monitoring module 16 is mainly used to perform temperature prediction on subsequent real-time spectral data based on the updated deep learning model, so as to realize real-time monitoring of the temperature of the high-brightness light source.

[0113] In this embodiment, the output light of the high-brightness light source is spectrally collected by the acquisition module 11 to obtain real-time spectrum data and corresponding time tags, thereby ensuring the real-time and accuracy of the light source temperature information. The prediction module 12 predicts the temperature of the real-time spectrum data through a preset deep learning model, provides the current temperature value and the prediction confidence, and ensures the high accuracy and reliability of the prediction. The comparison module 13 compares the current temperature value with the preset safe temperature threshold, calculates the temperature deviation value, and accurately adjusts the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence, thereby obtaining the adjusted driving parameters and ensuring that the light source works stably within the safe temperature range. The monitoring module 14 uses the time tag to synchronously monitor the adjusted driving parameters, obtains the real-time working status data and the temperature change trend, and provides continuous temperature monitoring capabilities. The adjustment module 15 adjusts the parameters of the deep learning model according to the real-time working status data and the temperature change trend, and continuously updates the model to adapt to the temperature changes under different working conditions. Finally, the monitoring module 16 predicts the temperature of the subsequent real-time spectrum data based on the updated deep learning model, ensuring that the system can monitor the temperature changes of the high-brightness light source in real time and achieve efficient and accurate temperature management and control.

[0114] It should be noted that technicians in the relevant technical field can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned high-brightness light source temperature real-time monitoring method embodiment, and will not be repeated here.

[0115] Embodiment 4: like Figure 3 As shown, the present application also provides an electronic device 20, including a memory 21 and a processor 22, the memory 21 stores a computer program that can be run on the processor 22, and when the processor 22 executes the computer program, the real-time monitoring method of the high-brightness light source temperature of Example 1 is implemented.

[0116] In this embodiment, an electronic device 20 including a memory 21 and a processor 22 is used to run a computer program stored in the memory 21, and the processor 22 implements a method for real-time monitoring of the temperature of a high-brightness light source when executing the program. Specifically, the electronic device 20 performs spectrum acquisition on the output light of the high-brightness light source through the coordinated work of the acquisition module 11, the prediction module 12, the comparison module 13, the monitoring module 14, the adjustment module 15 and the monitoring module 16, and obtains spectrum data and time tags in real time. The real-time spectrum data is temperature predicted by the deep learning model to obtain the current temperature value and the prediction confidence, and the temperature is compared with the preset safety temperature threshold, the temperature deviation value is calculated, the driving current is accurately adjusted, and the safety of the light source temperature is ensured. The monitoring module 14 monitors the adjusted driving parameters and temperature change trends in real time, and provides feedback data for continuous optimization and updating of the model. In this way, the electronic device 20 can efficiently manage and control the temperature of the high-brightness light source, ensure that it operates within a safe range, extend the service life of the equipment and improve the performance stability, thereby realizing accurate real-time monitoring of the temperature of the high-brightness light source.

[0117] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time monitoring of temperature of a high-brightness light source, characterized in that: include: Collect the spectrum of the output light of the high-brightness light source to obtain real-time spectrum data and corresponding time stamps; Performing temperature prediction on the real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence; Comparing the current temperature value with a preset safety temperature threshold to obtain a temperature deviation value, and adjusting the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter; Using the time mark, synchronously monitoring the adjusted driving parameters to obtain real-time working status data and temperature change trends; Adjusting parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model; Based on the updated deep learning model, temperature prediction is performed on subsequent real-time spectral data to achieve real-time monitoring of the temperature of the high-brightness light source.

2. The method for real-time monitoring temperature of a high-brightness light source according to claim 1, characterized in that: The step of collecting the spectrum of the output light of the high-brightness light source to obtain real-time spectrum data and corresponding time tags includes: Using a high-resolution spectrum analyzer and a high-speed data acquisition module to synchronously collect the spectrum of the output light of the high-brightness light source to obtain original spectrum data and an initial time signal, and performing dark current correction and baseline correction on the original spectrum data to obtain corrected spectrum data and baseline drift parameters; The corrected spectral data are compared with a preset reference spectrum, a spectral offset and a spectral distortion coefficient are extracted, the spectral offset and the spectral distortion coefficient are used to perform spectral correction on the corrected spectral data to obtain standardized spectral data, and the standardized spectral data are fused with the baseline drift parameter to obtain real-time spectral feature data; The initial time signal is synchronized with the analysis time of the high-resolution spectrum analyzer through a time synchronization algorithm to obtain an accurate time mark, and the real-time spectrum feature data and the accurate time mark are associated to obtain real-time spectrum data and a corresponding time mark.

3. The method for real-time monitoring temperature of a high-brightness light source according to claim 1, characterized in that: The deep learning model includes a convolutional neural network model, a first long short-term memory network model and a second long short-term memory network model. The step of performing temperature prediction on the real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence includes: Inputting the real-time spectral data into the convolutional neural network model, extracting characteristic response values, calculating characteristic weights using the characteristic response values, and obtaining a primary feature set and a secondary feature set; Inputting the primary feature set into the first long short-term memory network model to obtain a primary temperature prediction value and a primary feature confidence, and inputting the secondary feature set into the second long short-term memory network model to obtain a secondary temperature prediction value and a secondary feature confidence; The main temperature prediction value and the secondary temperature prediction value are weightedly fused based on the main feature confidence and the secondary feature confidence to obtain a fused temperature prediction value and a comprehensive confidence. The fused temperature prediction value is error corrected using a residual network to obtain a current temperature value and a correction factor. The comprehensive confidence is combined with the correction factor to calculate the prediction confidence.

4. The method for real-time monitoring temperature of a high-brightness light source according to claim 1, characterized in that: The step of comparing the current temperature value with a preset safety temperature threshold to obtain a temperature deviation value, and adjusting the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter includes: Compare the current temperature value with the preset upper and lower limits of the safety temperature threshold to obtain an over-temperature deviation value and an under-temperature deviation value, and determine the temperature adjustment requirement and adjustment priority according to the over-temperature deviation value and the under-temperature deviation value in combination with the prediction confidence; The temperature regulation requirement and the regulation priority are processed by a fuzzy control algorithm to obtain an initial current regulation amount, the initial current regulation amount is compared with historical driving current data, and an adaptation coefficient is calculated based on the comparison result; The initial current regulation amount is optimized by using the adaptation coefficient to obtain an optimized current regulation amount, the optimized current regulation amount is applied to the current driving current, a current regulation curve is calculated, and a driving control instruction is generated according to the current regulation curve; The driving control instruction is sent to a preset driving control module to adjust the driving current of the high-brightness light source to obtain an adjusted driving parameter.

5. The method for real-time monitoring temperature of a high-brightness light source according to claim 1, characterized in that: The step of synchronously monitoring the adjusted drive parameters using the time marker to obtain real-time working status data and temperature change trend includes: Acquire the execution time of the adjusted driving parameters through a preset data acquisition module to obtain an execution time mark, and use the time mark to time align and synchronize the execution time mark to obtain synchronous driving data; Collecting the operating state parameters of the high-brightness light source to obtain operating state data, and correlating and analyzing the operating state data with the synchronous drive data to obtain real-time working state data and correlation factors; The correlation factor is used to perform trend analysis on the real-time working status data to obtain the temperature change trend.

6. The method for real-time monitoring temperature of a high-brightness light source according to claim 1, characterized in that: The step of adjusting parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model includes: Extracting key parameters from the real-time working status data and combining them with the temperature change trend to generate a model adjustment data set and feature adjustment parameters; Using an online incremental learning algorithm, the model adjustment data set is input into the deep learning model, and the weight parameters and bias parameters of the deep learning model are updated to obtain a preliminarily optimized deep learning model; The feature adjustment parameters are used to reselect the main feature set and the secondary feature set of the initially optimized deep learning model, update the feature subspace, and obtain an updated deep learning model.

7. The method for real-time monitoring temperature of a high-brightness light source according to claim 6, characterized in that: After the step of reselecting the primary feature set and the secondary feature set of the initially optimized deep learning model by using the feature adjustment parameters, updating the feature subspace, and obtaining the updated deep learning model, the method further comprises: Using a preset validation set to evaluate the updated deep learning model, and obtain the accuracy, stability and generalization ability indicators of the updated deep learning model; If the accuracy, stability and generalization ability indicators of the updated deep learning model meet the preset requirements, the updated deep learning model parameters and structure are saved to generate a new model version; The new model version is deployed in the system to replace the old deep learning model and realize online update.

8. A high brightness light source temperature real-time monitoring device, characterized in that: include: The acquisition module is used to collect the spectrum of the output light of the high-brightness light source to obtain real-time spectrum data and corresponding time tags; A prediction module, used to perform temperature prediction on the real-time spectral data through a preset deep learning model to obtain a current temperature value and a prediction confidence; A comparison module, used for comparing the current temperature value with a preset safety temperature threshold value to obtain a temperature deviation value, and adjusting the driving current of the high-brightness light source according to the temperature deviation value and the prediction confidence to obtain an adjusted driving parameter; A monitoring module, for synchronously monitoring the adjusted driving parameters using the time marker to obtain real-time working status data and temperature change trend; An adjustment module, used to adjust parameters of the deep learning model according to the real-time working status data and the temperature change trend to obtain an updated deep learning model; The monitoring module is used to perform temperature prediction on subsequent real-time spectral data based on the updated deep learning model to achieve real-time monitoring of the temperature of the high-brightness light source.

9. An electronic device, characterized in that: The invention comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the processor implements the high-brightness light source temperature real-time monitoring method according to any one of claims 1 to 7 when executing the computer program.

Images