Intelligent dimming control method and system based on LED driver

Abstract

The invention relates to the technical field of intelligent dimming control, and discloses an intelligent dimming control method and system based on an LED driver, and the method comprises the steps: carrying out the current and light intensity data collection of the LED driver under different dimming depth values, and obtaining a response characteristic data set; inputting the response characteristic data set into a convolutional block attention module and convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; the ratio of the light intensity change rate to the current change rate is calculated, and a piecewise linear dimming sensitivity index is obtained through piecewise linear function fitting; constructing a first-order inertia controller and a brightness feedback loop, and generating a brightness self-compensation control strategy; according to the LED driver, the brightness change response is predicted, and the PWM dimming control signal is generated according to the trajectory smoothing algorithm and the PID controller, so that the adaptive capacity of the LED driver to external interference and parameter change is enhanced, and the stability and reliability of the LED driver are improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent dimming control, and in particular to an intelligent dimming control method and system based on an LED driver. Background Art

[0002] The intelligent dimming function of LED lighting system has become the target technology to meet different lighting needs, including adaptive ambient brightness adjustment, scene mode switching, circadian rhythm lighting and other diversified applications. However, due to the nonlinear characteristics of LED devices themselves and the complex relationship between current and luminous flux conversion, traditional LED dimming control technology faces many challenges, especially in achieving precise dimming and ensuring the stability of brightness output.

[0003] Traditional LED driver dimming technology mainly uses PWM modulation or constant current source regulation. These methods generally have the problem that output current control and brightness adjustment are independent of each other, which may cause conflicts between the two in actual applications, affecting dimming performance. Especially in the process of large-scale dimming, due to the diversity of lighting needs, the nonlinear harmonic distortion of LED drivers is unpredictable, making traditional open-loop control methods unable to effectively cope with it. In addition, LED dimming systems based on PWM modulation often cause transient current shock problems during high-frequency switching, which not only shortens the life of LED light sources, but also produces obvious flickering, affecting lighting quality and user experience. Summary of the invention

[0004] The present application provides an intelligent dimming control method and system based on an LED driver, thereby enhancing the adaptability of the LED driver to external interference and parameter changes, and improving the stability and reliability of the LED driver.

[0005] In a first aspect, the present application provides an intelligent dimming control method based on an LED driver, and the intelligent dimming control method based on an LED driver includes: The current and light intensity data of the LED driver are collected at different dimming depth values ​​to obtain a response characteristic data set; Inputting the response characteristic data set into the convolutional block attention module and the convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; The ratio of the light intensity change rate to the current change rate is calculated based on the LED response characteristic prediction model, and a piecewise linear dimming sensitivity index is obtained by piecewise linear function fitting; Constructing a first-order inertia controller and a brightness feedback loop according to the piecewise linear dimming sensitivity index, and generating a brightness self-compensation control strategy; The brightness change response is predicted according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and a PWM dimming control signal is generated according to a trajectory smoothing algorithm and a PID controller.

[0006] A second aspect of the present application provides an intelligent dimming control system based on an LED driver, the intelligent dimming control system based on an LED driver comprising: The acquisition module is used to collect the current and light intensity data of the LED driver at different dimming depth values ​​to obtain a response characteristic data set; A training module, used for inputting the response characteristic data set into a convolutional block attention module and a convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; A calculation module, used to calculate the ratio of the light intensity change rate to the current change rate based on the LED response characteristic prediction model, and obtain a piecewise linear dimming sensitivity index by fitting a piecewise linear function; A construction module, used to construct a first-order inertia controller and a brightness feedback loop according to the piecewise linear dimming sensitivity index, and generate a brightness self-compensation control strategy; A generation module is used to predict the brightness change response according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and to generate a PWM dimming control signal according to a trajectory smoothing algorithm and a PID controller.

[0007] Compared with the prior art, the present application has the following beneficial effects: by constructing a LED response characteristic prediction model based on CBAM-CNN and a piecewise linear dimming sensitivity index, high-precision prediction and control of LED brightness changes are achieved, especially in the low dimming depth area. The cascade control structure of the first-order inertia controller and the brightness feedback loop is adopted to solve the problem that the current control and brightness adjustment are independent of each other and may conflict with each other in the traditional method, and the current stability is significantly improved. Based on the dynamic inertia coefficient adjustment strategy and the cross-interval disturbance-free switching mechanism, the LED flickering phenomenon is effectively eliminated. At the same time, by accurately controlling the current transient impact, combining the adaptive PID control and trajectory smoothing algorithm, the current stress of the LED during the dimming process is reduced, and the service life of the LED light source is extended. In addition, the method uses a comprehensive objective function of multi-objective optimization to ensure the brightness adjustment accuracy while taking into account the optimization of energy efficiency ratio, and does not require an additional dimming signal processing circuit, which simplifies the system structure and reduces the hardware cost and complexity. Through the closed-loop self-optimization system of data acquisition-model update-strategy optimization-control execution, adaptive dimming for different LED device characteristics and environmental conditions is achieved, and the versatility and adaptability of the system are improved. The brightness change trajectory design based on B-spline curve fitting and satisfying dynamic constraints ensures a smooth transition of the dimming process. Through real-time monitoring by self-correcting coupling mechanism and state observer, the adaptability of the LED driver to external interference and parameter changes is enhanced, and the stability and reliability of the LED driver are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] 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.

[0009] 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.

[0010] Figure 1 is a flow chart of an intelligent dimming control method based on an LED driver provided in an embodiment of the present invention; Figure 2 It is a schematic block diagram of the structure of an intelligent dimming control system based on an LED driver provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0011] 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.

[0012] 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 be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0013] 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.

[0014] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1 In the embodiment of the present application, an embodiment of the intelligent dimming control method based on the LED driver includes: Step 100, collecting current and light intensity data of the LED driver at different dimming depth values ​​to obtain a response characteristic data set; It is understandable that the execution subject of the present application may be an intelligent dimming control system based on an LED driver, or a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0015] Specifically, the current sampling circuit, light intensity detection circuit and data storage module are connected through the LED driver to ensure the real-time collection of current and light intensity data under different dimming depth values. Among them, the current sampling circuit adopts a high-precision current sensor to convert the real-time current signal in the LED driving circuit into a measurable voltage signal to ensure that the accuracy of current measurement is not less than 0.1mA, and the light intensity detection circuit is composed of a photodiode and a light intensity conversion unit, and the linearity error of its light intensity detection is controlled within 1% to ensure the reliability of the measured data. The data storage module includes a high-precision analog-to-digital conversion unit and a memory, and the analog-to-digital conversion accuracy is not less than 10 bits, so as to ensure that the precision of the collected data meets the requirements of intelligent dimming control. After completing the hardware construction of the data acquisition system, the dimming depth value sequence is set to be evenly distributed in the range of 0% to 100%, and the value interval is set to 1% to obtain a higher dimming resolution and ensure the integrity and continuity of data acquisition. After setting the dimming depth value, the transient response measurement is performed on each dimming depth value to obtain the dynamic characteristics of the LED driver under different dimming conditions. At each dimming depth value, the current and light intensity response data of the LED driver are collected in the range of 10ms to 1000ms, and the sampling frequency is set to 10kHz to ensure sufficient time resolution to capture the key features of the transient process. During the data collection process, attention is paid to the rise time, fall time and steady-state value of the LED current, and the change characteristics of the light intensity are recorded, including the light intensity rise time, fall time, steady-state value and fluctuation range. Through the analysis of these data, key characteristic parameters are extracted to form a characteristic parameter set, in which the current rise time and fall time reflect the response speed of the LED driver, the steady-state current value characterizes the steady-state characteristics of the driver, and the light intensity change parameter is used to evaluate the light output stability and dimming nonlinear characteristics of the LED light source. At the same time, in order to analyze the impact of PWM modulation on the LED driver, the frequency range and duty cycle range of the PWM signal are set to form a PWM parameter test matrix. The PWM frequency range is set between 100Hz and 20kHz to cover the dimming requirements in different application scenarios, and the PWM duty cycle is set between 0.1% and 99.9% to ensure that the dimming system accurately controls the light output. By traversing different combinations of frequencies and duty cycles, the LED driver is tested multiple times, the current and light intensity response data under each PWM parameter combination are recorded, and the corresponding characteristic parameters are extracted. On this basis, a mapping database of dimming parameters and system response characteristics is constructed. The database associates the data of four dimensions, namely dimming depth, PWM parameters, current response characteristics, and light intensity response characteristics, to fully reflect the nonlinear characteristics and harmonic distortion of the LED driver under different dimming conditions.By integrating all the above experimental data and calculation results, a complete response characteristic data set is obtained, which includes the relationship between dimming depth and current, light intensity, and also includes the dynamic response characteristics under different PWM frequencies and duty cycles.

[0016] Step 200: Input the response characteristic data set into the convolutional block attention module and the convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; Specifically, the response characteristic data set is preprocessed to ensure that the data is suitable for neural network training. The data is normalized to convert the dimming depth, current parameters, light intensity parameters and PWM control signals into standardized values ​​from 0 to 1 to avoid the problem of unstable gradient update caused by too large a data range and ensure that data of different feature dimensions are trained on the same scale. Statistical analysis methods are used to remove outliers, based on standard deviation threshold or box plot analysis, so as to improve the credibility of the data and the generalization ability of the model. In order to ensure the temporal consistency of the input data, time series alignment processing is performed, and the current and light intensity data under different dimming depths and PWM parameters are aligned at the same time node by interpolation or sliding window methods. After the data preprocessing operation, the training data obtained includes input features, namely dimming depth, PWM parameters, current waveform characteristics and light intensity waveform characteristics, and output labels, namely the current response characteristics and light intensity change trend of the LED driver. A model structure based on the convolutional block attention module and convolutional neural network is constructed. Create a channel attention submodule, whose structure consists of parallel computing units of maximum pooling and average pooling. The maximum pooling layer extracts the most global and significant information in the feature map, while the average pooling layer can provide overall background information. The outputs of these two pooling methods are then fused by a multi-layer perceptron to calculate channel weights, thereby enhancing the focus on key channel features. Create a spatial attention submodule, which uses the fusion strategy of maximum pooling and average pooling to extract feature information in the spatial dimension, then extracts features through a seven-by-seven convolution kernel, and uses a sigmoid activation function to generate a spatial attention map to highlight the features of key time periods during the dimming process. After completing the construction of the channel attention submodule and the spatial attention submodule, they are combined in series to form a convolutional block attention module, which enables the model to more accurately identify key feature information in the dimming signal during the learning process. While constructing the attention mechanism, the main structure of the convolutional neural network is built, and four convolutional blocks are designed. Each convolutional block contains two three-by-three convolutional layers to extract local spatial features. At the same time, a batch normalization layer is added after each convolutional layer to improve the training stability of the model, and the ReLU activation function is used to introduce nonlinear characteristics to ensure that the network can effectively fit the complex dynamic response characteristics of the LED driver. In order to enhance the feature extraction capability, the maximum pooling layer is used between each convolutional block for dimensionality reduction to reduce the amount of calculation and improve the generalization ability of the model. Two fully connected layers are set at the back end of the convolutional neural network, one of which is used to output the current prediction value of the LED driver, and the other is used to output the light intensity prediction value, so as to ensure that the model simultaneously learns the changing laws of current and light intensity during LED dimming, and provides multi-task learning capabilities.In order to improve the model's ability to identify key features, the previously constructed convolutional block attention module is embedded after the main structure of the convolutional network, so that the neural network can automatically pay attention to the response characteristics of the key time period in the time series of the dimming signal, and can adaptively adjust the feature extraction method for different PWM parameters and dimming depths, thereby improving the learning effect of the model and forming a CBAM-CNN combined architecture. The mean square error loss function and Adam optimizer are used to input the training input features and output labels into the CBAM-CNN combined architecture for training to measure the deviation between the predicted value and the true value, and drive the model to continuously optimize the weight parameters to improve the prediction accuracy. The Adam optimizer is used in the training process, which adjusts the learning rate according to the gradient adaptively, thereby improving the convergence speed and avoiding the local optimal trap, ensuring that the model can achieve optimal performance within a limited number of training rounds. In order to prevent the model from overfitting, the early stopping strategy is applied during the training process, that is, when the loss function of the validation set fails to significantly decrease in multiple consecutive training rounds, the training is terminated in advance to avoid the model from overfitting the training data and affecting the generalization ability. After 500 rounds of training, the final LED response characteristic prediction model can accurately predict the LED driver current response and light intensity change trend under any dimming depth.

[0017] Step 300: Calculate the ratio of the light intensity change rate to the current change rate based on the LED response characteristic prediction model, and obtain a piecewise linear dimming sensitivity index by piecewise linear function fitting; It should be noted that the LED response characteristic prediction model is used to predict the dimming depth points to obtain the LED current response curve and light intensity response curve. The prediction model can accurately simulate the dynamic response of the current and light intensity of the LED under different dimming depths and PWM parameters. By inputting a series of equally spaced dimming depth points, such as from 0% to 100%, with an interval set to 0.1%, the corresponding LED current response curve and light intensity response curve are generated, where the current response curve describes the current change trend of the LED driver under different dimming depths, and the light intensity response curve describes the corresponding light intensity output change. The current change rate between adjacent points is calculated according to the LED current response curve, and the light intensity change rate between adjacent points is calculated according to the light intensity response curve to quantify the instantaneous change rate of current and light intensity during the dimming process. By performing discrete differentiation on the current response curve, that is, calculating the current increment between each dimming depth point and the adjacent point, and dividing it by the corresponding dimming depth increment, the current change rate data set is obtained. Similarly, the light intensity response curve is subjected to discrete differentiation to obtain the light intensity change rate data set. The light intensity change rate data set and the current change rate data set are divided point-to-point. For each dimming depth point, the light intensity change rate is divided by the current change rate to calculate the dimming sensitivity curve, which describes the sensitivity of the dimming depth change to the light intensity change. In this curve, the larger the sensitivity value, the smaller the current change can cause a larger light intensity change at this dimming depth, while the area with smaller sensitivity indicates that the light intensity responds to the current more slowly. In order to analyze the change law of dimming sensitivity, the slope change characteristics of the dimming sensitivity curve are analyzed, and the dimming depth range is divided into multiple sensitivity segmented intervals according to different change trends. By calculating the slope of the dimming sensitivity curve, the change trend of the sensitivity in different dimming depth intervals is judged. When the slope changes suddenly, it can be used as the boundary point of the interval division. When the slope of the sensitivity curve is large, it means that the light intensity responds more violently to the current change in this interval, while when the slope is small, the dimming stability of the system is higher. Multiple sensitivity segmented intervals are divided according to these change characteristics. Linear fitting is performed on the sensitivity curve in each segmented interval to construct a piecewise linear dimming sensitivity model. A piecewise linear function is used to fit each sensitivity segmented interval so that the dimming sensitivity of each interval is represented by a simple linear function, thereby reducing the computational complexity and improving the interpretability of the model. During the fitting process, the parameter set of the piecewise linear function is determined, including the slope and intercept of each interval, and these parameters are stored for subsequent dimming control optimization. The interval sensitivity coefficient, sensitivity change rate, sensitivity extreme point and sensitivity stability interval are calculated based on the piecewise linear function parameter set, where the interval sensitivity coefficient represents the average sensitivity level in a specific dimming depth interval, and the sensitivity change rate describes the rate of change of sensitivity with dimming depth, which is used to judge the dimming stability of a certain interval.The sensitivity extreme point is used to identify the peak position of the sensitivity during the dimming process, that is, at certain specific dimming depths, the LED's light output is the most sensitive or the most stable, and the sensitivity stability range is defined as the dimming depth range where the sensitivity change rate is lower than the preset threshold, indicating that the output of the dimming system is relatively stable within this range. By calculating these key indicators, the response characteristics of the LED driver at different dimming depths are fully described. By integrating the dimming sensitivity curve, sensitivity segmentation interval, piecewise linear function parameter set, and various sensitivity evaluation indicators, a piecewise linear dimming sensitivity index is constructed.

[0018] Step 400: construct a first-order inertia controller and a brightness feedback loop according to the piecewise linear dimming sensitivity index, and generate a brightness self-compensation control strategy; Specifically, a first-order inertia controller is constructed and its mathematical model is established to describe the dynamic response characteristics of the LED drive system at different dimming depths. The transfer characteristics of the controller are defined so that it can adjust the rate of change of the LED current within different dimming depth intervals and ensure the stability of the light intensity output. In order to make the response of the controller more adaptable to the sensitivity changes at different dimming depths, the inertia coefficient adjustment strategy of the controller is designed according to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index. Since the dimming sensitivity index has been divided into multiple sensitivity segmented intervals, a piecewise function is constructed based on the sensitivity characteristics of different intervals so that the inertia coefficient can be dynamically adjusted with the change of the dimming depth. For example, in the interval with higher sensitivity, the LED light output is more sensitive to the current change. At this time, increasing the inertia coefficient reduces the response speed of the system and reduces the fluctuation caused by the sudden change of light intensity. In the interval with lower sensitivity, the inertia coefficient is appropriately reduced to speed up the dynamic response of the system and improve the dimming accuracy and real-time performance. The dynamic adjustment strategy of the inertia coefficient needs to be combined with the size of the sensitivity value and optimize the dimming characteristics of the system through different value rules to ensure that the system can achieve the optimal control effect at different dimming depths. While designing the inertia controller, a brightness feedback loop is constructed to achieve real-time monitoring and compensation adjustment of the output brightness of the LED light source. The brightness feedback loop is composed of a photoelectric detection unit, which includes a photoelectric sensor and a signal conditioning circuit. The light intensity signal of the LED light source is converted into a measurable electrical signal, and compared with the target brightness signal to calculate the brightness error signal. The error signal represents the deviation between the current brightness of the LED and the set brightness, and is used for subsequent brightness compensation control. After the brightness error signal is calculated, compensation processing is performed, and a current adjustment signal is generated through the compensation network. The parameters of the compensation network are optimized according to the segmented linear dimming sensitivity index to ensure that the best brightness compensation effect can be achieved in different dimming depth ranges. In order to improve the adaptive ability of dimming control, the controller parameters are dynamically adjusted according to the dimming depth of the current working point, as well as the sensitivity extreme point and the sensitivity stable range information, and an adaptive control parameter table is constructed. The parameter table is used to store the optimal control parameters under different dimming depths, including inertia coefficient, gain, time constant and related parameters of the compensation network. When the LED driver is at different dimming depths, the system automatically searches for the corresponding control parameters and makes real-time adjustments to ensure that the system maintains stable dynamic response characteristics within the full dimming range. For example, near the extreme sensitivity point, the LED light output responds most violently to current changes. The inertia coefficient is appropriately increased to suppress light intensity fluctuations. In the sensitivity stability range, since the system response is relatively stable, the inertia coefficient is reduced to improve the control accuracy and energy efficiency of the dimming system. In order to achieve closed-loop control of the above control strategy, a cascade control structure is constructed, in which the brightness feedback loop is used as the outer loop, and the first-order inertia controller is used as the inner loop, and the coupling control of the inner and outer loops is achieved through the current adjustment signal.In this structure, the outer loop detects the brightness output of the LED light source and generates a brightness error signal, while the inner loop controls the driving current through the current adjustment signal to ensure that the system can operate stably at different dimming depths. At the same time, the dynamic inertia coefficient adjustment strategy and adaptive control parameter table are introduced in the cascade control structure, so that the system can automatically adjust the control parameters according to the characteristics of different dimming depths, thereby optimizing the dynamic response performance of the system. The designed brightness self-compensation control strategy can achieve precise adjustment of the brightness of the LED light source while ensuring the stability of the current control, and improve the energy efficiency ratio of the system through the adaptive adjustment mechanism, so that the LED driver can maintain efficient and stable operation in the full dimming range.

[0019] The cascade control structure of the brightness feedback loop and the first-order inertia controller is constructed, and the signal flow topology relationship is established, so that the brightness command signal enters the brightness feedback loop, and the current command signal is input into the first-order inertia controller to ensure that the system accurately adjusts the LED light output within the full dimming range. Through this cascade control method, the coordinated optimization of brightness adjustment and current control is realized, in which the brightness feedback loop is used to detect the actual brightness of the LED light source in real time and calculate the error signal between the actual brightness and the target brightness, while the first-order inertia controller adjusts the current of the LED driver by receiving the current command signal, thereby achieving efficient and stable brightness control. In the processing of the brightness error signal, a gain regulator is designed, and the gain value of the gain regulator is dynamically adjusted based on the interval sensitivity coefficient in the segmented linear dimming sensitivity index to adapt to the system characteristics under different dimming depths. In the high-sensitivity range, the brightness change has a more drastic effect on the current, so the gain value is reduced to suppress overshoot and fluctuation. In the low-sensitivity range, the dimming accuracy of the system is required to be higher. At this time, the gain value is appropriately increased to enhance the system's responsiveness to the brightness error signal. In the process of realizing brightness self-compensation, a feedforward-feedback structure is established between the current adjustment signal and the actual current signal of the LED driver to improve the stability and anti-interference ability of the system. Among them, the feedforward part is used to directly calculate the current adjustment signal according to the brightness target, while the feedback part uses the actual current signal to correct the current adjustment result to compensate for the nonlinear error in the system. The dynamic inertia coefficient adjustment strategy is introduced in the feedback link, so that the system can dynamically adjust the inertia coefficient according to the sensitivity characteristics of the current dimming depth to ensure that the optimal current control effect can be achieved under different working conditions. A reasonable transition strategy is designed to avoid the sudden change of control parameters from affecting the stability of the system. Based on the sensitivity extreme point information in the adaptive control parameter table, a disturbance-free switching mechanism is designed for the sensitivity mutation interval, in which the sliding average method is used to smoothly adjust the control parameters to ensure the continuity of parameter changes. Through this mechanism, a smooth transition is achieved between different dimming intervals, avoiding brightness jumps and flickering during the dimming process, and improving user experience and the stability of the lighting system. A state observer is added to the coupling node of the brightness feedback loop and the first-order inertia controller to monitor the corresponding relationship between the current command and the actual brightness in real time, and self-correct the control parameters during system operation. The self-correction coupling mechanism continuously monitors the changing trend of LED light output to determine whether the system deviates from the optimal control trajectory, and automatically adjusts the control parameters when deviation is detected to compensate for changes in system characteristics caused by environmental changes, device aging and other factors, thereby improving the long-term stability of the dimming system.By integrating the brightness feedback loop, first-order inertia controller, adaptive gain adjustment mechanism, dynamic inertia coefficient adjustment strategy, cross-interval disturbance-free switching mechanism and self-correction coupling mechanism, and performing stability analysis and boundary condition limitation on the entire system to ensure that the system can maintain stable operation under different working conditions, a complete brightness self-compensation control strategy is finally formed.

[0020] Step 500: predict the brightness change response according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and generate a PWM dimming control signal according to the trajectory smoothing algorithm and the PID controller.

[0021] Specifically, the change process between the target brightness value and the current brightness value is modeled to ensure the smoothness and controllability of the dimming process. The transition time parameters and the brightness change function are set so that the brightness adjustment of the LED can be completed within a certain time range and avoid visual discomfort caused by sudden changes. By inputting these parameters into the LED response characteristic prediction model, a brightness change prediction sequence is obtained, which describes the expected change trajectory of the LED light output during a given dimming depth adjustment process and provides key time series data for subsequent dimming control optimization. A comprehensive objective function is constructed based on the brightness change prediction sequence and the brightness self-compensation control strategy. The objective function is used to balance the accuracy, smoothness and energy efficiency of brightness adjustment and provide evaluation criteria for trajectory optimization. In order to make the brightness change trajectory smoother, a trajectory smoothing algorithm is used to perform B-spline curve fitting on the brightness change prediction sequence. The B-spline curve is a smooth interpolation method that effectively suppresses the drastic fluctuations that may occur during the brightness change process by adjusting the control point density and tension coefficient, thereby ensuring a smooth transition of the dimming trajectory. During the fitting process, the setting of the control point density is adaptively adjusted according to the rate of change of the dimming depth. For example, the control point density is increased in the interval where the brightness changes faster to improve the smoothness of the trajectory, while the control point density is appropriately reduced in the interval where the brightness changes slower to reduce the computational complexity. At the same time, the optimization of the tension coefficient enables the adjustment rate of the brightness change curve in different time periods to transition naturally, avoiding the problem of sudden brightness jumps or discontinuities. Through B-spline curve fitting, a smooth brightness control trajectory is obtained, which can accurately describe the gradual change process from the target brightness value to the current brightness value and meet the visual comfort requirements. The smooth brightness control trajectory is differentiated to extract the speed and acceleration components during the brightness change process. The speed component reflects the rate of change of brightness over time, while the acceleration component describes the dynamic adjustment of the brightness change rate. Constraints are screened for the speed and acceleration components to limit the brightness change rate and brightness change acceleration to ensure the smoothness of the dimming process. For example, in the low dimming depth range, LEDs are more sensitive to current changes, so a lower upper limit of the brightness change rate is set to prevent excessive brightness fluctuations. In the high dimming depth range, the change rate limit is appropriately relaxed to improve the response speed of the system. After constraint screening, the brightness change trajectory that meets the dynamic constraints is obtained. Based on the brightness change trajectory that meets the dynamic constraints, the adaptive PID control parameters of the PID controller are calculated to ensure the stability and accuracy of the dimming control. The parameters of the PID controller include three gain coefficients: proportional, integral, and differential. The settings of these parameters directly affect the response characteristics of the system. In order to adapt to the changing laws of different dimming depth ranges, according to the dynamic characteristics of the brightness change trajectory, an adaptive parameter adjustment strategy is adopted, that is, the proportional gain is appropriately reduced in the high sensitivity range to reduce overshoot and oscillation, and the integral gain is increased in the low sensitivity range to improve the steady-state accuracy of the system.In the case of a large rate of change in brightness, the differential gain is appropriately increased to enhance the system's ability to predict the trend of changes, thereby avoiding lags in the brightness adjustment process. The adaptive PID parameter adjustment strategy ensures that the PID controller can maintain good control performance in the full dimming range and improves the system's dynamic response capability. After the PID parameter calculation is completed, it is applied to the tracking control of the brightness change trajectory that meets the dynamic constraints to generate real-time control values ​​of the PWM frequency and duty cycle. The frequency and duty cycle of the PWM signal are the core control parameters of the LED dimming system, where the frequency affects the stability of the dimming and the duty cycle directly determines the light output of the LED. When generating the PWM control signal, dynamic limiting and dead zone control are performed to prevent the PWM signal from having too high or too low a duty cycle, thereby avoiding nonlinear distortion or flickering problems of the LED under extreme working conditions. For example, in the low brightness range, due to the high sensitivity of the LED to the duty cycle, a smaller minimum duty cycle limit is set to avoid failure of brightness adjustment, while in the high brightness range, it is necessary to prevent the duty cycle from being too high, causing the LED to overheat or reduce its service life. In addition, the introduction of dead zone control prevents the PWM signal from switching frequently within a small change range, thereby reducing the switching loss of the drive circuit and improving the energy efficiency of the system. By performing dynamic limiting and dead zone control, an optimized PWM dimming control signal is obtained, which can accurately match the LED brightness change requirements and ensure the smoothness, response speed and system stability of the dimming process.

[0022] In the embodiment of the present application, by constructing a LED response characteristic prediction model based on CBAM-CNN and a piecewise linear dimming sensitivity index, high-precision prediction and control of LED brightness changes are achieved, especially in the low dimming depth area. The cascade control structure of the first-order inertia controller and the brightness feedback loop is adopted to solve the problem that the current control and brightness adjustment are independent of each other and may conflict in the traditional method, and the current stability is significantly improved. Based on the dynamic inertia coefficient adjustment strategy and the cross-interval disturbance-free switching mechanism, the LED flickering phenomenon is effectively eliminated. At the same time, by accurately controlling the current transient impact, combining the adaptive PID control and trajectory smoothing algorithm, the current stress of the LED during the dimming process is reduced, and the service life of the LED light source is extended. In addition, the method uses a comprehensive objective function of multi-objective optimization to ensure the brightness adjustment accuracy while taking into account the optimization of energy efficiency ratio, and does not require an additional dimming signal processing circuit, which simplifies the system structure and reduces the hardware cost and complexity. Through the closed-loop self-optimization system of data acquisition-model update-strategy optimization-control execution, adaptive dimming for different LED device characteristics and environmental conditions is achieved, and the versatility and adaptability of the system are improved. The brightness change trajectory design based on B-spline curve fitting and satisfying dynamic constraints ensures a smooth transition of the dimming process. Through real-time monitoring by self-correcting coupling mechanism and state observer, the adaptability of the LED driver to external interference and parameter changes is enhanced, and the stability and reliability of the LED driver are improved.

[0023] In a specific embodiment, the process of executing step 100 may specifically include the following steps: Connect the current sampling circuit, the light intensity detection circuit and the data storage module through the LED driver, and set a value sequence for the dimming depth value to obtain multiple dimming depth values; Perform transient response measurement on each dimming depth value to obtain time domain response raw data, and extract current rise time, fall time, steady-state value and light intensity change parameters from the time domain response raw data to obtain a characteristic parameter set; The frequency range and duty cycle range are set for the PWM modulation mode to obtain a PWM parameter test matrix, and a mapping relationship between dimming parameters and system response characteristics is established based on the characteristic parameter set and the PWM parameter test matrix to obtain a response characteristic data set.

[0024] Specifically, the current sampling circuit, light intensity detection circuit and data storage module are connected through the LED driver to ensure accurate measurement of the transient response characteristics of the LED. Among them, the current sampling circuit uses a high-precision current sensor to convert the LED driving current into a measurable electrical signal in real time to ensure that the measurement accuracy reaches the milliampere level, and the light intensity detection circuit is composed of a photodiode and a signal conditioning unit, which can effectively capture the changes in the LED light output and convert the light signal into a voltage signal to ensure that the light intensity measurement error is controlled within 1%. The data storage module integrates a high-precision analog-to-digital conversion unit and a data storage device to ensure the integrity and accuracy of data acquisition. In order to systematically analyze the operating characteristics of the LED driver under different dimming depths, the value sequence of the dimming depth is set to cover the entire range from 0% to 100%, and the dimming step size is set, such as 1%, to obtain sufficiently fine data points while ensuring the efficiency of data acquisition. After completing the hardware configuration of the data acquisition system, the transient response measurement is performed on each dimming depth value to obtain the current and light intensity change process of the LED driver under different dimming conditions. At each dimming depth, the current and light intensity data of the LED driver will be collected within a set time window, such as 10 milliseconds to 1000 milliseconds, and the data will be recorded at a sampling frequency of 10kHz to ensure that the complete transient change characteristics are captured. In data collection, pay attention to the rise time, fall time and steady-state value of the current, and record the change characteristics of the light intensity, including the light intensity rise time, fall time, steady-state light intensity and the fluctuation range of the light intensity. Through these data, the key feature parameter set is extracted to quantify the transient response characteristics of the LED. The frequency range and duty cycle range are set for the PWM modulation method to analyze the impact of PWM parameters on the LED driving characteristics. The PWM dimming method changes the average current of the LED by adjusting the duty cycle to achieve brightness adjustment. Experiments are conducted at different PWM frequencies to ensure that the dimming system can maintain stable operation in different application scenarios. For example, the PWM frequency range is set to 100Hz to 20kHz, covering low-frequency to high-frequency dimming requirements, and the PWM duty cycle range is set to 0.1% to 99.9% to study the LED response characteristics under different duty cycles. During the test, different PWM parameter combinations are traversed, and the LED current and light intensity response data under each PWM setting are recorded. In order to establish the mapping relationship between PWM parameters and LED system response, the characteristic parameter set and PWM parameter test matrix are comprehensively analyzed to build a mathematical model to describe the relationship between dimming depth, current and light intensity. Through experimental data, piecewise functions or data fitting methods are used to build a more accurate model. For example, the current-light intensity characteristic function of the dimming system is defined as follows: ; in, Represents the LED light intensity, is the current, is the PWM duty cycle, and is a parameter that describes the nonlinearity of LED current-light output, and is the coefficient of the PWM duty cycle on the light output. This model is fitted by experimental data to obtain the best parameters for different LED drivers, so as to achieve accurate prediction of LED brightness changes. In order to improve the accuracy of PWM dimming, the mapping relationship between dimming parameters and system response characteristics is stored in the database, and the PWM control strategy is optimized by machine learning algorithm. For example, a neural network model is used to input PWM parameters and dimming depth, and output the predicted values ​​of LED current and light intensity changes, so as to achieve adaptive dimming control in practical applications.

[0025] In a specific embodiment, the process of executing step 200 may specifically include the following steps: Perform data normalization, outlier filtering, and time series alignment on the response characteristic dataset to obtain training input features and output labels; Create a parallel structure of maximum pooling and average pooling and a multi-layer perceptron processing unit in the channel attention submodule, and create a maximum-average pooling fusion layer and a convolution kernel feature extraction unit in the spatial attention submodule; The channel attention submodule and the spatial attention submodule are combined in series to obtain the convolutional block attention module, and four convolutional blocks of the main structure of the convolutional neural network are constructed at the same time. Each convolutional block contains two 3×3 convolutional layers, a batch normalization layer and a ReLU activation function; Two fully connected layers are set in the convolutional neural network. The two fully connected layers are connected to the current prediction output terminal and the light intensity prediction output terminal respectively. At the same time, a convolutional block attention module is embedded after the convolutional block. The recognition ability of the target time period and target features in the dimming process is enhanced through the attention mechanism, and the CBAM-CNN combined architecture is obtained. The mean square error loss function and Adam optimizer are used to input the training input features and output labels into the CBAM-CNN combined architecture for training to obtain the LED response characteristics prediction model.

[0026] Specifically, the response characteristic data set is preprocessed to ensure that the data is suitable for neural network training. The data is normalized to eliminate the scale differences between different features so that the range of all input data is limited to the same numerical range, such as converting the dimming depth, current parameter, light intensity parameter and PWM control signal into standardized values ​​between 0 and 1, so as to avoid the problem of uneven learning speed caused by different feature value ranges during gradient update. In order to improve the reliability of the data, outlier filtering is performed, and the outliers in the data set are removed by statistical analysis methods, such as using the triple standard deviation rule or quantile-based methods to detect and remove data points that deviate from the normal range, thereby improving the generalization ability of the model. At the same time, in order to ensure that time series data can effectively perform pattern learning in the neural network, time series alignment processing is performed, and the current and light intensity data under different dimming depths and PWM parameters are aligned to a unified time base through interpolation or sliding window technology to ensure that the model can accurately learn the relationship between input features and output labels. After data preprocessing, the training data obtained includes input features, namely dimming depth, PWM parameters, current waveform features and light intensity waveform features, as well as output labels, namely the current response characteristics and light intensity change trend of the LED driver. After completing data preprocessing, a model structure based on the convolutional block attention module and convolutional neural network is constructed to effectively extract key features in the data and improve prediction accuracy. A channel attention submodule is created, which consists of parallel computing units of maximum pooling and average pooling, where the maximum pooling layer is used to extract global significant information in the feature map, while the average pooling layer provides overall background information. The results of the parallel computing are then fused through a multi-layer perceptron to calculate channel weights, thereby enhancing the focus on key channel features. At the same time, a spatial attention submodule is created, which uses a fusion strategy of maximum pooling and average pooling to extract feature information in the spatial dimension, then extracts features through a convolution kernel, and uses a nonlinear activation function to generate a spatial attention map to highlight the features of key time periods during the dimming process. After completing the construction of the channel attention submodule and the spatial attention submodule, they are combined in series to form a convolutional block attention module, so that the model can more accurately identify the key feature information in the LED response characteristic data. In order to enhance the feature extraction capability, the main structure of the convolutional neural network is built, and four convolutional blocks are designed. Each convolutional block contains two three-by-three convolutional layers. The output of each convolutional layer is expressed as: ; in, Representative The feature map of the layer, is the convolution kernel, * represents the convolution operation, is the bias term, is the ReLU activation function. Through convolution calculation, the network extracts local spatial features, and a batch normalization layer is added after each convolution layer to improve the training stability of the model. The maximum pooling layer is used between each convolution block for dimensionality reduction to reduce the amount of calculation and improve the generalization ability of the model. At the output end of the convolutional neural network, two fully connected layers are set, one of which is used to output the current prediction value of the LED driver, and the other is used to output the light intensity prediction value, so as to ensure that the model simultaneously learns the changing laws of current and light intensity during the LED dimming process and provides multi-task learning capabilities. At the same time, in order to improve the model's ability to recognize key features, the previously constructed convolutional block attention module is embedded after the main structure of the convolutional network, so that the neural network can automatically pay attention to the response characteristics of the key time period in the time series of the dimming signal, and can adaptively adjust the feature extraction method for different PWM parameters and dimming depths, thereby improving the learning effect of the model and forming a CBAM-CNN combined architecture. After completing the construction of the model architecture, the training process of the model is optimized, so the mean square error loss function is used. This loss function is used to measure the deviation between the model prediction value and the true value, and drive the model to continuously optimize the weight parameters to improve the prediction accuracy. At the same time, the Adam optimizer is used for gradient update. This optimizer combines the momentum method and the adaptive learning rate adjustment mechanism to dynamically adjust the learning rate according to the gradient change, thereby improving the convergence speed and reducing the instability of parameter updates. During the training process, in order to prevent the model from overfitting, an early stopping strategy is adopted, that is, when the loss function of the validation set fails to significantly decrease in multiple consecutive training rounds, the training is terminated in advance to avoid the model from overfitting the training data and affecting the generalization ability. During the entire model training process, the weights of the neural network are continuously optimized through back propagation, so that it can extract meaningful features from the input data and accurately predict the current response and light intensity change trend of the LED driver.

[0027] In a specific embodiment, the process of executing step 300 may specifically include the following steps: The dimming depth point is predicted using the LED response characteristic prediction model to obtain the LED current response curve and light intensity response curve; The current change rate between adjacent points is calculated according to the LED current response curve, and the light intensity change rate between adjacent points is calculated according to the light intensity response curve, so as to obtain the current change rate data set and the light intensity change rate data set at each dimming depth point; Performing a point-to-point division operation on the light intensity change rate data set and the current change rate data set to obtain a dimming sensitivity curve; According to the slope variation characteristics of the dimming sensitivity curve, the dimming depth range is divided into multiple intervals to obtain multiple sensitivity segmented intervals; A piecewise linear function is used to fit each sensitivity segment interval to obtain a piecewise linear function parameter set; The interval sensitivity coefficient, sensitivity change rate, sensitivity extreme point and sensitivity stability interval are calculated according to the piecewise linear function parameter set to obtain the piecewise linear dimming sensitivity index.

[0028] Specifically, the dimming depth points are predicted using the LED response characteristic prediction model to obtain the LED current response curve and the light intensity response curve. Due to the nonlinear characteristics of the LED drive system, the change rates of these two curves at different dimming depths are not uniform, so it is necessary to ensure that the prediction model has sufficient accuracy to accurately describe the response behavior of the LED current and light intensity under different dimming depth conditions. Assuming that the dimming depth range is from 0% to 100%, discrete sampling is performed with a step size of 1%. By inputting these dimming depth points into the trained LED response characteristic prediction model, the corresponding current and light intensity change trends are obtained respectively, thereby constructing the LED current response curve and the light intensity response curve. Based on the current and light intensity response curves, the current change rate and light intensity change rate between adjacent dimming depth points are calculated to quantify the transient change characteristics under different dimming depths. The current change rate is obtained by calculating the incremental ratio of the current response curve at adjacent depth points. The calculation method of the light intensity change rate is similar, that is, discrete differential operations are performed on the current response curve and the light intensity response curve respectively to construct the current change rate data set and the light intensity change rate data set. Point-to-point division operation is performed on the light intensity change rate data set and the current change rate data set to obtain the dimming sensitivity curve, that is, at each dimming depth point, the corresponding dimming sensitivity is expressed as: ; in, Indicates the dimming depth Dimming sensitivity at Represents light intensity, represents the current, and Respectively represent the light intensity increment and current increment of adjacent dimming depth points. The physical meaning of this formula is to measure the amplitude of light intensity change caused by unit current change. The larger the value, the more sensitive the LED is to the current change in the dimming depth range. Conversely, the LED light output in this area is relatively stable. The slope of the sensitivity curve is analyzed, and based on the characteristics of the slope change, the dimming depth range is divided into multiple sensitivity segmented intervals. The rate of change of sensitivity is determined by calculating the numerical derivative of the dimming sensitivity curve at each dimming depth point. For example, the sensitivity changes slowly in some intervals, while the sensitivity changes suddenly in other intervals. By setting the slope change threshold, different sensitivity segmented intervals can be identified. After completing the sensitivity segmentation, the dimming sensitivity curve in each interval is linearly fitted to construct a piecewise linear dimming sensitivity model. The purpose of using linear function fitting is to describe the relationship between dimming depth and dimming sensitivity in a simpler mathematical expression in each sensitivity interval, so as to reduce the computational complexity and improve the interpretability of the model. Assuming that in a certain sensitivity interval, the change of dimming sensitivity with dimming depth is approximately expressed as: ; in, is the dimming sensitivity within this range, is the sensitivity slope of this interval, is the intercept. The piecewise linear function parameters are fitted by the least squares method or the gradient descent method to ensure that they can accurately reflect the actual sensitivity changes in the interval. Through the piecewise linear function parameter set obtained by fitting, various key dimming performance indicators are calculated to fully characterize the dimming sensitivity characteristics under different dimming depths. When calculating the dimming performance indicators, the interval sensitivity coefficient is determined. This indicator is used to represent the average sensitivity level in a specific dimming depth interval. Its calculation is directly obtained through the linear function parameters of the interval; the sensitivity change rate is calculated. This indicator is used to describe the rate of change of sensitivity with dimming depth, so as to judge the dimming stability in different intervals; the sensitivity extreme point is identified, that is, the dimming depth position where the sensitivity peaks or valleys appear. This information helps to optimize the dimming control strategy, such as appropriately adjusting the PWM dimming strategy near the sensitivity extreme point to reduce the sudden change of light intensity; the sensitivity stability interval is calculated. This interval is defined as the dimming depth range where the sensitivity change rate is lower than the preset threshold, indicating that the response of the LED light output to the current change in this interval is relatively stable. These areas are usually important targets for dimming control strategy optimization. A piecewise linear dimming sensitivity index system is constructed by integrating the dimming sensitivity curve, sensitivity segmentation interval, piecewise linear function parameter set and various sensitivity evaluation indicators.

[0029] In a specific embodiment, the process of executing step 400 may specifically include the following steps: Define the transfer function of the first-order inertia controller and obtain the mathematical model of the controller; According to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index, a piecewise function is designed for the inertia coefficient in the controller mathematical model to obtain a dynamic inertia coefficient adjustment strategy; Design a photoelectric detection unit of the brightness feedback loop, compare the LED light source output brightness signal detected by the photoelectric detection unit with the target brightness signal to obtain a brightness error signal, and perform compensation processing on the brightness error signal to obtain a current adjustment signal; According to the dimming depth of the current working point and the sensitivity extreme point and sensitivity stable interval information in the piecewise linear dimming sensitivity index, the controller parameters are dynamically adjusted to obtain an adaptive control parameter table; The brightness feedback loop is used as the outer loop and the first-order inertia controller as the inner loop. The inner and outer loops are coupled through the current adjustment signal to build a cascade control structure. At the same time, the dynamic inertia coefficient adjustment strategy and adaptive control parameter table are introduced to dynamically adjust the parameters to obtain the brightness self-compensation control strategy.

[0030] Specifically, the transfer characteristics of the first-order inertia controller are defined to establish a mathematical model of the controller, and the dynamic change characteristics of the LED drive current are described by the model. In the LED drive system, the current adjustment has a certain inertia, that is, the change of the current input current will not immediately cause a stable response of the light output, but will undergo a dynamic transition process. Use the first-order inertia model to describe the LED drive current For input control signal The dynamic response is: ; in, Indicates the LED driving current over time changes, is the input control signal, represents the inertia coefficient, which determines the response speed of the system, and is the gain coefficient, which indicates the degree of influence of the input control signal on the LED current. This equation describes the dynamic characteristics of the LED drive current. In particular, at different dimming depths, the rate of change of the LED current is greatly affected by the inertia coefficient. Therefore, the inertia coefficient needs to be further optimized to ensure that the best brightness adjustment effect can be achieved under different dimming depth conditions. In order to optimize the response characteristics of the first-order inertia controller, the inertia coefficient is dynamically adjusted according to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index. Since there are significant differences in the dimming sensitivity of different dimming depth intervals, the inertia coefficient α cannot be a fixed value, but needs to be adaptively adjusted as the dimming depth changes. For example, in the low dimming depth interval, since the light output of the LED is more sensitive to current changes, the inertia coefficient should be appropriately increased to reduce the response speed of the system and prevent the current fluctuation from causing sudden changes in light intensity; while in the high dimming depth interval, since the response of the LED light intensity to the current is relatively stable, the inertia coefficient should be appropriately reduced to speed up the dimming response speed and improve the dynamic performance of the system. A piecewise function is used to define the adjustment strategy of the inertia coefficient. For example, in different dimming depth intervals, the sensitivity coefficient is used to adjust the inertia coefficient. As a reference, the inertia coefficient is nonlinearly mapped to obtain the dynamic inertia coefficient adjustment strategy. While implementing inertia control, a brightness feedback loop is designed to detect the light output of the LED in real time and ensure that the error between it and the target brightness is effectively compensated. The core component of the brightness feedback loop is the photoelectric detection unit, which consists of a photoelectric sensor and a signal conditioning circuit. It is used to monitor the brightness output of the LED in real time and convert it into a standardized electrical signal for subsequent error calculation and compensation control. The brightness signal detected by the photoelectric detection unit is converted into a standardized electrical signal. With the target brightness signal The comparison is performed to calculate the brightness error signal: ; The brightness error signal is used to adjust the LED drive current to ensure that the actual brightness output is consistent with the target brightness. In order to optimize dimming stability, the brightness error signal is compensated, such as by generating a current adjustment signal through PI or adaptive compensation strategy. , thus achieving precise control of LED brightness. In the brightness feedback loop, the influence of dimming depth on controller parameters is considered. Therefore, according to the dimming depth of the current working point and the sensitivity extreme point and sensitivity stable interval information in the piecewise linear dimming sensitivity index, the controller parameters are dynamically adjusted and an adaptive control parameter table is constructed. This parameter table is used to store the optimal control parameters under different dimming depths, including the inertia coefficient , Gain And the parameters of the compensation network and . During the operation of the LED, the corresponding control parameters are automatically found according to the current dimming depth, and real-time adjustments are made to ensure that the LED light output can stably respond to current changes. For example, near the extreme point of sensitivity, since the LED is most sensitive to current changes, the inertia coefficient is appropriately increased to suppress the fluctuation of light intensity, while in the stable sensitivity range, the inertia coefficient is reduced to improve the dynamic performance and energy efficiency of the dimming control. In order to achieve closed-loop control of the entire dimming system, the brightness feedback loop is used as the outer loop, and the first-order inertia controller is used as the inner loop. The inner and outer loops are coupled through the current adjustment signal, and a cascade control structure is constructed. In this structure, the outer loop is used to detect the brightness of the LED light output and calculate the brightness error signal, while the inner loop is responsible for adjusting the LED drive current to ensure that the brightness error is effectively compensated. In order to improve the intelligence of the system, a dynamic inertia coefficient adjustment strategy and an adaptive control parameter table are introduced in the cascade control structure, so that the system can automatically adjust the control parameters according to the characteristics of different dimming depths, thereby optimizing the dynamic response performance of the system. For example, in the low brightness range, a slower current adjustment rate is used to avoid flickering of light output, while in the high brightness range, a faster current adjustment rate is used to improve the response speed of the dimming system. During the dimming process, if the system detects that the brightness error signal continues to deviate from the target value, the control parameters are adaptively adjusted to optimize the duty cycle and frequency of the PWM signal, thereby further improving the stability and accuracy of the system.

[0031] In a specific embodiment, the execution step uses the brightness feedback loop as the outer loop and the first-order inertia controller as the inner loop, realizes the coupling of the inner and outer loops through the current adjustment signal, constructs a cascade control structure, and introduces a dynamic inertia coefficient adjustment strategy and an adaptive control parameter table to dynamically adjust the parameters. The process of obtaining the brightness self-compensation control strategy can specifically include the following steps: A signal flow topology structure is established through a brightness feedback loop and a first-order inertia controller, a brightness command signal is introduced into the brightness feedback loop, and a current command signal is introduced into the first-order inertia controller to obtain a cascade control structure; A gain regulator is designed according to the brightness error signal, and the gain value of the gain regulator is dynamically adjusted according to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index to obtain an adaptive gain adjustment mechanism; A feedforward-feedback structure is established between the current adjustment signal and the actual current signal of the LED driver, and the dynamic inertia coefficient adjustment strategy is introduced into the feedback link to obtain a current control loop with dynamic inertia characteristics; According to the control characteristics at the extreme sensitivity point in the adaptive control parameter table, a transition strategy is designed for the sensitivity mutation interval, and a cross-interval disturbance-free switching mechanism is obtained by using a sliding average control parameter change; A state observer is added to the coupling node of the brightness feedback loop and the first-order inertia controller to monitor the corresponding relationship between the current command and the actual brightness in real time, and a self-correction coupling mechanism is obtained. The cascade control structure, adaptive gain adjustment mechanism, current control loop with dynamic inertia characteristics, cross-interval disturbance-free switching mechanism and self-correction coupling mechanism are integrated, and stability analysis and boundary condition limitation are performed to obtain the brightness self-compensation control strategy.

[0032] Specifically, an efficient signal flow topology is constructed, in which the brightness feedback loop and the first-order inertia controller work together to achieve accurate brightness adjustment and current control. In this structure, the brightness command signal is introduced into the brightness feedback loop. The task of the brightness feedback loop is to monitor the output brightness of the LED light source and compare it with the target brightness to generate a brightness error signal. At the same time, the current command signal is introduced into the first-order inertia controller, which is used to adjust the current output of the LED according to the current command signal. Through the signal flow topology, the brightness feedback loop and the first-order inertia controller cooperate with each other to achieve the coupling control of the inner and outer loops, thereby ensuring the stability and accuracy of the brightness adjustment process and obtaining a cascade control structure. The gain regulator is designed according to the brightness error signal, and the gain value of the gain regulator is dynamically adjusted according to the interval sensitivity coefficient in the segmented linear dimming sensitivity index. For example, in the interval with high dimming sensitivity, the gain should be reduced accordingly to reduce the over-adjustment phenomenon of the system, thereby avoiding excessive light intensity fluctuations; while in the interval with low dimming sensitivity, the gain should be increased to improve the response speed of the system. Through the adaptive gain adjustment mechanism, the system dynamically adjusts the gain according to the real-time brightness error signal to ensure a smooth transition of the brightness adjustment process and reduce unnecessary errors. When implementing current regulation, a feedforward-feedback structure is established between the current adjustment signal and the actual current signal of the LED driver. Feedforward control is used to calculate the required current value according to the target brightness in advance, while feedback control is corrected according to the actual measured current value to ensure accurate current adjustment. In this structure, the dynamic inertia coefficient adjustment strategy plays an important role, because LED light output usually has inertia, and current changes do not immediately cause light intensity changes. Therefore, dynamically adjusting the inertia coefficient α helps the control system to adaptively adjust the speed of current response according to different dimming depths. By introducing the dynamic adjustment strategy of the inertia coefficient, it is ensured that the current changes smoothly in the low sensitivity range to avoid excessive fluctuations in light intensity, while in the high sensitivity range, the inertia coefficient will increase, slowing down the current change speed, thereby preventing drastic fluctuations in light intensity, forming a current control loop with dynamic inertia characteristics. A transition strategy is designed for the range with large sensitivity changes to smooth the transition of the system between different dimming depths. For example, in the sensitivity mutation interval, the relationship between current and brightness will change significantly. If the control parameters are switched directly, the system output will be discontinuous and flickering will occur. In order to avoid this situation, a cross-interval disturbance-free switching mechanism is designed. The sliding average method is used to smoothly adjust the control parameters to ensure that the control parameters of the system will not change suddenly in the sensitivity mutation interval. In this way, a smooth transition of the control parameters is achieved, unnecessary control errors are avoided, and the stability and responsiveness of the system are enhanced. In order to improve the reliability and adaptability of the system, a state observer is added to the coupling node of the brightness feedback loop and the first-order inertia controller.The state observer can monitor the correspondence between the current command and the actual brightness in real time, and continuously correct the control parameters during the operation of the system to adapt to environmental changes or other interference factors. The self-correction coupling mechanism helps the system maintain high accuracy and stability in long-term operation and reduce the impact of external factors on the dimming system. The brightness feedback loop, first-order inertia controller, adaptive gain adjustment mechanism, dynamic inertia coefficient adjustment strategy, cross-interval disturbance-free switching mechanism and self-correction coupling mechanism are integrated to construct a brightness self-compensation control strategy. By performing stability analysis and boundary condition limitation, the stability of the system under different working conditions is ensured to avoid system instability or mis-dimming caused by improper parameter adjustment. And by optimizing and integrating each link of the system, it is possible to reduce the energy consumption of LEDs and improve the energy efficiency of the system while maintaining high-precision dimming control.

[0033] In a specific embodiment, the process of executing step 500 may specifically include the following steps: The transition time parameter and brightness change function are set for the change process between the target brightness value and the current brightness value, and the LED response characteristic prediction model is input to obtain the brightness change prediction sequence; Based on the brightness change prediction sequence and the brightness self-compensation control strategy, a comprehensive objective function is constructed, and a trajectory smoothing algorithm is used to perform B-spline curve fitting on the brightness change prediction sequence, and the control point density and tension coefficient are set to smooth the brightness change trajectory to obtain a smooth brightness control trajectory. Perform differential operation on the smooth brightness control trajectory, extract the velocity and acceleration components, and perform constraint screening on the velocity and acceleration components to limit the brightness change rate and brightness change acceleration, and obtain a brightness change trajectory that meets dynamic constraints; Calculate the adaptive PID control parameters of the PID controller according to the brightness change trajectory that meets the dynamic constraints; The adaptive PID control parameters are applied to the tracking control of the brightness change trajectory that meets the dynamic constraints, the real-time control values ​​of PWM frequency and duty cycle are generated, and dynamic limiting and dead zone control are performed to generate a PWM dimming control signal.

[0034] Specifically, a transition time parameter is set for the change process between the target brightness value and the current brightness value, and the parameter defines the time required from the current brightness value to the target brightness value. The setting of the transition time parameter depends on the response speed requirement of the system and the user's demand for dimming smoothness. A brightness change function is defined for the brightness change, which is based on the time definition and controls the LED light intensity to smoothly transition from the current brightness value to the target brightness value. The brightness change function is input into the LED response characteristic prediction model. The model predicts the current response of the LED under a given dimming depth based on the pre-trained current-light intensity mapping relationship, and obtains a brightness change prediction sequence to describe the change process of the LED from the current brightness to the target brightness. Based on the predicted brightness change sequence and the brightness self-compensation control strategy, a comprehensive objective function is constructed. The purpose of this objective function is to balance the accuracy, response time and stability of brightness adjustment. For example, the objective function comprehensively considers multiple factors such as minimization of brightness error, optimization of response time and improvement of energy efficiency ratio. By performing trajectory smoothing on the brightness change prediction sequence, the B-spline curve fitting method is used to smooth the brightness change trajectory. B-spline curve can generate a smooth curve without introducing too much complexity, and adjust the shape of the curve by setting the control point density and tension coefficient. In this method, the control point density determines the smoothness of the curve. The denser the control points, the higher the fitting accuracy of the curve, while the tension coefficient controls the smoothness of the curve and the sharpness of the change. In this way, B-spline curve fitting can obtain a smooth brightness control trajectory, which can not only ensure that the brightness gradually transitions from the current value to the target value, but also avoid visual discomfort or light intensity fluctuations caused by too fast adjustment. Differentiation operation is performed on the brightness change trajectory to extract the velocity and acceleration components of the trajectory. The brightness change rate (velocity component) and the brightness change acceleration are two important dynamic parameters of the control system. The velocity component reflects the rate of brightness adjustment, while the acceleration component reflects the change of the brightness change rate. Too high a brightness change rate causes LED flickering or light intensity fluctuations, and too low a brightness change rate affects the response speed of the system. Constraint conditions are screened for the velocity and acceleration components, and the brightness change rate and brightness change acceleration are limited to ensure that the dimming process meets the dynamic response constraints. For example, by setting the maximum brightness change rate and the maximum acceleration limit, the system is prevented from generating excessive light intensity fluctuations or excessive current shocks at high brightness change rates, thereby ensuring stable operation of the system. After screening, the obtained brightness change trajectory not only meets the accuracy requirements, but also complies with the dynamic constraints of the system, and can effectively guide the subsequent optimization of the control strategy. The adaptive PID control parameters of the PID controller are calculated based on the brightness change trajectory that meets the dynamic constraints. The purpose of the PID controller is to adjust the output of the system by controlling the proportional (P), integral (I) and differential (D) parameters so that the brightness error gradually decreases and eventually tends to zero.The PID parameters are dynamically adjusted through an adaptive mechanism to provide the best control effect in different dimming depth intervals. According to the brightness change trajectory that meets the dynamic constraints, the model predictive control or feedback-based optimization algorithm is used to calculate the optimal PID parameters. For example, when the brightness change rate is large, the differential gain is appropriately increased to improve the system's response speed to the brightness change; while in the interval where the brightness changes slowly, the proportional gain is reduced to avoid over-adjustment of the system. Through adaptive PID control parameter adjustment, it is possible to ensure that the system can operate stably and efficiently at any dimming depth. The adaptive PID control parameters are applied to the tracking control of the brightness change trajectory that meets the dynamic constraints. The controller generates real-time control values ​​of PWM frequency and duty cycle based on the current brightness change trajectory, thereby achieving precise LED dimming. The frequency and duty cycle of the PWM signal directly determine the light output of the LED. In this process, dynamic limiting and dead zone control are performed to avoid excessive fluctuations in the PWM signal. The role of dynamic limiting control is to limit the frequency and duty cycle of the PWM signal within a reasonable range to prevent the control signal from exceeding the working range of the LED driver, resulting in overheating or overload. Dead-zone control is used to prevent frequent switching of PWM signals within a small range of variation, which helps reduce the switching loss of the LED drive circuit and improve the energy efficiency of the system. A smooth and efficient PWM dimming control signal is generated to accurately adjust the brightness output of the LED.

[0035] In the present embodiment, a self-learning optimization method for an intelligent dimming system of an LED driver is also included. The self-learning optimization method for an intelligent dimming system of an LED driver includes: establishing a real-time data acquisition channel for the LED driver during actual operation, recording parameters such as current waveforms, light intensity changes, temperature changes, and power supply fluctuations in multiple dimming cycles, and obtaining an operation status data set; performing error calculation on the operation status data set and the prediction results of the LED response characteristic prediction model, setting a multi-dimensional error threshold vector, marking the time points and parameters that exceed the threshold, and obtaining a model deviation map; performing cluster analysis on the model deviation map, using a density peak clustering algorithm to identify deviation patterns, and summarizing similar deviation features into typical scenarios to obtain deviation feature classification results; weighting the original training data according to the deviation feature classification results, increasing the sampling density in high deviation areas, constructing an incremental training data set, and online fine-tuning the LED response characteristic prediction model to obtain an adaptive Enhanced prediction model; recalculate the piecewise linear dimming sensitivity index based on the adaptively enhanced prediction model, adaptively adjust the sensitivity interval boundary and piecewise linear parameters to obtain the optimized sensitivity index; update the first-order inertia controller parameters and brightness feedback loop parameters according to the optimized sensitivity index, adjust the threshold conditions and piecewise functions in the dynamic inertia coefficient adjustment strategy to obtain the optimized control parameter set; simulate and verify the optimized control parameter set under specific working conditions, calculate the system stability margin and dynamic response index, perform parameter sensitivity analysis, screen out the optimal parameter combination, and obtain the verified optimal control parameters; update the verified optimal control parameters to the LED driver control system, record the external environmental conditions and load characteristics at the same time, establish an environment-parameter association table for fast parameter switching under similar conditions, and form a complete self-learning closed-loop optimization system of "data acquisition-model update-strategy optimization-control execution-performance evaluation".

[0036] The above describes the intelligent dimming control method based on the LED driver in the embodiment of the present application. The following describes the intelligent dimming control system 10 based on the LED driver in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of an intelligent dimming control system 10 based on an LED driver includes: The acquisition module 11 is used to collect current and light intensity data of the LED driver at different dimming depth values ​​to obtain a response characteristic data set; A training module 12 is used to input the response characteristic data set into the convolution block attention module and the convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; A calculation module 13 is used to calculate the ratio of the light intensity change rate to the current change rate based on the LED response characteristic prediction model, and obtain a piecewise linear dimming sensitivity index by fitting a piecewise linear function; A construction module 14 is used to construct a first-order inertia controller and a brightness feedback loop according to a piecewise linear dimming sensitivity index, and generate a brightness self-compensation control strategy; The generating module 15 is used to predict the brightness change response according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and to generate a PWM dimming control signal according to the trajectory smoothing algorithm and the PID controller.

[0037] Through the cooperation of the above components, by constructing the LED response characteristic prediction model based on CBAM-CNN and the piecewise linear dimming sensitivity index, the high-precision prediction and control of LED brightness change is achieved, especially in the low dimming depth area. The cascade control structure of the first-order inertia controller and the brightness feedback loop is adopted to solve the problem that the current control and brightness adjustment are independent of each other and may conflict in the traditional method, and the current stability is significantly improved. Based on the dynamic inertia coefficient adjustment strategy and the cross-interval disturbance-free switching mechanism, the LED flickering phenomenon is effectively eliminated. At the same time, by accurately controlling the current transient impact, combining adaptive PID control and trajectory smoothing algorithm, the current stress of the LED during the dimming process is reduced, and the service life of the LED light source is extended. In addition, this method uses the comprehensive objective function of multi-objective optimization to ensure the brightness adjustment accuracy while taking into account the optimization of energy efficiency ratio, and does not require additional dimming signal processing circuits, which simplifies the system structure and reduces the hardware cost and complexity. Through the closed-loop self-optimization system of data acquisition-model update-strategy optimization-control execution, adaptive dimming for different LED device characteristics and environmental conditions is achieved, and the versatility and adaptability of the system are improved. The brightness change trajectory design based on B-spline curve fitting and satisfying dynamic constraints ensures a smooth transition of the dimming process. Through real-time monitoring by self-correcting coupling mechanism and state observer, the adaptability of the LED driver to external interference and parameter changes is enhanced, and the stability and reliability of the LED driver are improved.

[0038] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0039] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0040] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent dimming control method based on LED driver, characterized in that: include: The current and light intensity data of the LED driver are collected at different dimming depth values ​​to obtain a response characteristic data set; Inputting the response characteristic data set into the convolutional block attention module and the convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; The ratio of the light intensity change rate to the current change rate is calculated based on the LED response characteristic prediction model, and a piecewise linear dimming sensitivity index is obtained by piecewise linear function fitting; Constructing a first-order inertia controller and a brightness feedback loop according to the piecewise linear dimming sensitivity index, and generating a brightness self-compensation control strategy; The brightness change response is predicted according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and a PWM dimming control signal is generated according to a trajectory smoothing algorithm and a PID controller.

2. The intelligent dimming control method based on LED driver according to claim 1, characterized in that: The current and light intensity data of the LED driver are collected at different dimming depth values ​​to obtain a response characteristic data set, including: Connect the current sampling circuit, the light intensity detection circuit and the data storage module through the LED driver, and set a value sequence for the dimming depth value to obtain multiple dimming depth values; Performing transient response measurement on each dimming depth value to obtain time domain response raw data, and extracting current rise time, fall time, steady-state value and light intensity change parameters from the time domain response raw data to obtain a characteristic parameter set; The frequency range and duty cycle range are set for the PWM modulation mode to obtain a PWM parameter test matrix, and a mapping relationship between dimming parameters and system response characteristics is established according to the characteristic parameter set and the PWM parameter test matrix to obtain a response characteristic data set.

3. The intelligent dimming control method based on LED driver according to claim 1, characterized in that: The step of inputting the response characteristic data set into a convolutional block attention module and a convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model includes: Performing data normalization, outlier filtering, and time series alignment processing on the response characteristic data set to obtain training input features and output labels; Create a parallel structure of maximum pooling and average pooling and a multi-layer perceptron processing unit in the channel attention submodule, and create a maximum-average pooling fusion layer and a convolution kernel feature extraction unit in the spatial attention submodule; The channel attention submodule and the spatial attention submodule are combined in series to obtain a convolutional block attention module, and four convolutional blocks of the main structure in the convolutional neural network are constructed at the same time, each convolutional block includes two 3×3 convolutional layers, a batch normalization layer and a ReLU activation function; Two fully connected layers are set in the convolutional neural network, and the two fully connected layers are respectively connected to the current prediction output terminal and the light intensity prediction output terminal. At the same time, the convolution block attention module is embedded after the convolution block, and the recognition ability of the target time period and target features in the dimming process is enhanced through the attention mechanism, so as to obtain the CBAM-CNN combined architecture; The training input features and output labels are input into the CBAM-CNN combined architecture for training using a mean square error loss function and an Adam optimizer to obtain an LED response characteristic prediction model.

4. The intelligent dimming control method based on LED driver according to claim 1, characterized in that: The calculating the ratio of the light intensity change rate to the current change rate based on the LED response characteristic prediction model, and obtaining the piecewise linear dimming sensitivity index by piecewise linear function fitting, includes: The LED response characteristic prediction model is used to predict the dimming depth point to obtain an LED current response curve and a light intensity response curve; Calculate the current change rate between adjacent points according to the LED current response curve, calculate the light intensity change rate between adjacent points according to the light intensity response curve, and obtain a current change rate data set and a light intensity change rate data set for each dimming depth point; Performing a point-to-point division operation on the light intensity change rate data set and the current change rate data set to obtain a dimming sensitivity curve; According to the slope variation characteristics of the dimming sensitivity curve, the dimming depth range is divided into a plurality of intervals to obtain a plurality of sensitivity segmented intervals; Using a piecewise linear function to fit each of the sensitivity segment intervals to obtain a piecewise linear function parameter set; The interval sensitivity coefficient, sensitivity change rate, sensitivity extreme point and sensitivity stability interval are calculated according to the piecewise linear function parameter set to obtain the piecewise linear dimming sensitivity index.

5. The intelligent dimming control method based on LED driver according to claim 1, characterized in that: The first-order inertia controller and the brightness feedback loop are constructed according to the piecewise linear dimming sensitivity index, and a brightness self-compensation control strategy is generated, including: Define the transfer function of the first-order inertia controller and obtain the mathematical model of the controller; According to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index, a piecewise function is designed for the inertia coefficient in the controller mathematical model to obtain a dynamic inertia coefficient adjustment strategy; Designing a photoelectric detection unit of a brightness feedback loop, comparing the LED light source output brightness signal detected by the photoelectric detection unit with the target brightness signal to obtain a brightness error signal, and performing compensation processing on the brightness error signal to obtain a current adjustment signal; According to the dimming depth of the current working point and the sensitivity extreme point and sensitivity stability interval information in the piecewise linear dimming sensitivity index, the controller parameters are dynamically adjusted to obtain an adaptive control parameter table; The brightness feedback loop is used as the outer loop, the first-order inertia controller is used as the inner loop, the inner and outer loops are coupled through the current adjustment signal, a cascade control structure is constructed, and the dynamic inertia coefficient adjustment strategy and the adaptive control parameter table are introduced to dynamically adjust the parameters to obtain the brightness self-compensation control strategy.

6. The intelligent dimming control method based on LED driver according to claim 5, characterized in that: The brightness feedback loop is used as an outer loop, the first-order inertia controller is used as an inner loop, the inner and outer loops are coupled through the current adjustment signal, a cascade control structure is constructed, and the dynamic inertia coefficient adjustment strategy and the adaptive control parameter table are introduced to dynamically adjust parameters to obtain the brightness self-compensation control strategy, including: A signal flow topology structure is established through the brightness feedback loop and the first-order inertia controller, a brightness command signal is introduced into the brightness feedback loop, and a current command signal is introduced into the first-order inertia controller to obtain a cascade control structure; Designing a gain regulator according to the brightness error signal, and dynamically adjusting the gain value of the gain regulator according to the interval sensitivity coefficient in the piecewise linear dimming sensitivity index to obtain an adaptive gain adjustment mechanism; A feedforward-feedback structure is established between the current adjustment signal and the actual current signal of the LED driver, and the dynamic inertia coefficient adjustment strategy is introduced into the feedback link to obtain a current control loop with dynamic inertia characteristics; According to the control characteristics at the extreme sensitivity point in the adaptive control parameter table, a transition strategy is designed for the sensitivity mutation interval, and a sliding average control parameter change is adopted to obtain a cross-interval disturbance-free switching mechanism; A state observer is added to the coupling node of the brightness feedback loop and the first-order inertia controller to monitor the corresponding relationship between the current command and the actual brightness in real time, thereby obtaining a self-correcting coupling mechanism; The cascade control structure, the adaptive gain adjustment mechanism, the current control loop with dynamic inertia characteristics, the cross-interval disturbance-free switching mechanism and the self-correction coupling mechanism are integrated, and stability analysis and boundary condition limitation are performed to obtain a brightness self-compensation control strategy.

7. The intelligent dimming control method based on LED driver according to claim 1, characterized in that: The method of predicting the brightness change response according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and generating a PWM dimming control signal according to a trajectory smoothing algorithm and a PID controller includes: Setting a transition time parameter and a brightness change function for the change process between the target brightness value and the current brightness value, and inputting the LED response characteristic prediction model to obtain a brightness change prediction sequence; Based on the brightness change prediction sequence and the brightness self-compensation control strategy, a comprehensive objective function is constructed, and a trajectory smoothing algorithm is used to perform B-spline curve fitting on the brightness change prediction sequence, and control point density and tension coefficient are set to smooth the brightness change trajectory to obtain a smooth brightness control trajectory; Performing differential operation on the smooth brightness control trajectory, extracting velocity and acceleration components, and performing constraint condition screening on the velocity and acceleration components, limiting the brightness change rate and brightness change acceleration, and obtaining a brightness change trajectory that meets dynamic constraints; Calculate the adaptive PID control parameters of the PID controller according to the brightness change trajectory that satisfies the dynamic constraints; The adaptive PID control parameters are applied to the tracking control of the brightness change trajectory that meets the dynamic constraints, the real-time control values ​​of PWM frequency and duty cycle are generated, and dynamic limiting and dead zone control are performed to generate a PWM dimming control signal.

8. An intelligent dimming control system based on LED driver, characterized in that: Used to execute the intelligent dimming control method based on the LED driver according to any one of claims 1 to 7, the intelligent dimming control system based on the LED driver comprises: The acquisition module is used to collect the current and light intensity data of the LED driver at different dimming depth values ​​to obtain a response characteristic data set; A training module, used for inputting the response characteristic data set into a convolutional block attention module and a convolutional neural network combined architecture for training to obtain an LED response characteristic prediction model; A calculation module, used to calculate the ratio of the light intensity change rate to the current change rate based on the LED response characteristic prediction model, and obtain a piecewise linear dimming sensitivity index by fitting a piecewise linear function; A construction module, used to construct a first-order inertia controller and a brightness feedback loop according to the piecewise linear dimming sensitivity index, and generate a brightness self-compensation control strategy; A generation module is used to predict the brightness change response according to the LED response characteristic prediction model and the brightness self-compensation control strategy, and to generate a PWM dimming control signal according to a trajectory smoothing algorithm and a PID controller.

Images