A self-adaptive dimming method and system for high-power LED lamp beads based on intelligent driving

Abstract

The application discloses a kind of based on intelligent drive's high-power LED lamp bead self-adapting dimming method and system, belongs to lighting control technical field field, it includes the factor data of acquisition mode switching threshold and influence mode switching threshold, constructs historical data set and constructs prediction model;Periodically collect and construct real-time data set;According to current state value calculated from real-time data set;Real-time data set is input into prediction model to obtain current switching threshold;According to the size relationship of current state value and current mode switching threshold, switch current dimming mode;According to the corresponding control signal of current dimming mode and send to LED drive circuit, to control LED lamp bead.The application adopts neural network model to predict mode switching threshold, and according to the size relationship of current state value and current mode switching threshold to switch dimming mode, can realize the self-adapting dimming of high-power LED lamp bead, improve dimming range and precision, simultaneously according to environmental change self-adapting adjustment brightness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lighting control, and particularly to a high-power LED lamp bead adaptive dimming method and system based on intelligent driving. BACKGROUND

[0002] With the continuous development of LED technology, high-power LED lamp beads have been widely used in the field of lighting. However, traditional LED dimming methods often have limited dimming range, low dimming precision, and cannot adaptively adjust brightness according to environmental changes. These problems not only affect the lighting effect of LED lamps, but also may cause energy waste and user experience degradation.

[0003] Currently, most traditional LED dimming methods use fixed-frequency PWM (Pulse Width Modulation) dimming or analog dimming. PWM dimming adjusts the brightness of LEDs by changing the duty cycle of the LED driving signal, but fixed-frequency PWM dimming may produce obvious flicker during dimming, affecting user experience. Analog dimming adjusts brightness by changing the voltage or current of the LED driving circuit, but the dimming precision is relatively low, and it is difficult to achieve fine brightness adjustment.

[0004] Traditional LED dimming methods often cannot adaptively adjust according to environmental changes, which may lead to insufficient lighting or excessive lighting, which not only wastes energy, but also may cause discomfort to the user's eyes. SUMMARY

[0005] To solve the above problems, the present application provides a high-power LED lamp bead adaptive dimming method and system based on intelligent driving, which uses a neural network model to predict mode switching thresholds and switches dimming modes according to the size relationship between the current state value and the current mode switching threshold, which can realize adaptive dimming of high-power LED lamp beads, improve dimming range and precision, and adaptively adjust brightness according to environmental changes.

[0006] The above objectives can be achieved by the following solutions:

[0007] The application discloses a self-adaptive dimming method for high-power LED lamp beads based on intelligent driving, and the method comprises the following steps: collecting a mode switching threshold and factor data affecting the mode switching threshold, and constructing a historical data set; constructing and training a prediction model for predicting the mode switching threshold according to the historical data set; periodically collecting environmental light intensity data, working temperature data of the LED lamp beads and brightness demand data set by a user, and constructing a real-time data set; calculating a current state value according to the real-time data set; inputting the real-time data set into the prediction model to output a current switching threshold; switching a current dimming mode according to the size relationship between the current state value and the current mode switching threshold; generating a corresponding control signal according to the current dimming mode and sending the control signal to an LED driving circuit, so as to control the LED lamp beads; wherein the dimming mode comprises a high-frequency PWM dimming mode, a low-frequency PWM dimming mode, an analog dimming mode and a hybrid mode of the low-frequency PWM dimming and the analog dimming.

[0008] Optionally, the step of constructing and training the prediction model for predicting the mode switching value according to the historical data set comprises the following steps: establishing a neural network model comprising an input layer, at least one hidden layer and an output layer; setting three neurons in the input layer, which respectively receive environmental light intensity, working temperature of the LED lamp beads and brightness demand set by the user as input features; setting three neurons in the output layer, which are respectively used for outputting a first threshold value, a second threshold value and a third threshold value; setting an activation function, and constructing a prediction function of each neuron in the output layer according to the activation function; training the neural network model by using the historical data set to generate a stable prediction model; wherein the first threshold value is less than the second threshold value, and the second threshold value is less than the third threshold value.

[0009] Optionally, the step of calculating the current state value according to the real-time data set comprises the following steps: calculating the change rates of the environmental light intensity data, the working temperature data of the LED lamp beads and the brightness demand set by the user; judging whether the change rates are greater than a fourth threshold value; if yes, calculating the current state value according to the real-time data set; and if no, maintaining the current dimming mode.

[0010] Optionally, the step of calculating the current state value according to the real-time data set comprises the following steps: performing normalization processing on the environmental light intensity data, the working temperature data of the LED lamp beads and the brightness demand set by the user; setting a first weight of the environmental light intensity, a second weight of the working temperature of the LED lamp beads and a third weight of the brightness demand set by the user; calculating the current state value according to the first weight, the second weight, the third weight and the normalized data, and , ,

[0011] wherein, is a first weight, is a second weight, is a third weight, is normalized ambient light intensity data, is normalized LED light bead working temperature data, is normalized user set brightness demand data.

[0012] Optionally, the inputting the real-time data set into the prediction model to output a current switching threshold value includes: normalizing ambient light intensity data, LED light bead working temperature data, and user set brightness demand data; inputting the normalized data into the prediction model to obtain a current first threshold value, a current second threshold value, and a current third threshold value, wherein the current first threshold value is determined by , ,

[0013] wherein, is an activation function, is a transpose of a weight vector corresponding to the current first threshold value, is a bias corresponding to the current first threshold value, the current second threshold value is determined by , ,

[0014] wherein, is a transpose of a weight vector corresponding to the current second threshold value, is a bias corresponding to the current second threshold value, the current third threshold value is determined by , ,

[0015] wherein, is a transpose of a weight vector corresponding to the current third threshold value, is a bias corresponding to the current third threshold value; wherein the current switching threshold value includes the current first threshold value, the current second threshold value, and the current third threshold value.

[0016] Optionally, the switching of the current dimming mode according to the size relationship between the current state value and the current mode switching threshold value includes: determining whether the current state value is greater than the current first threshold value; if not, switching the analog dimming mode; if yes, determining whether the current state value is greater than the current second threshold value; if not, switching the low-frequency PWM dimming mode; if yes, determining whether the current state value is greater than the current third threshold value; if not, switching the hybrid mode of the low-frequency PWM dimming plus the analog dimming; if yes, switching the high-frequency PWM dimming mode.

[0017] Optionally, the switching the hybrid mode of the low-frequency PWM dimming plus analog dimming comprises: calculating a proportion of analog dimming according to the current state value, the current second threshold value and the current third threshold value, and switching the hybrid mode of the low-frequency PWM dimming plus analog dimming for the proportion of analog dimming , there is ;

[0018] switching the hybrid mode of the low-frequency PWM dimming plus analog dimming according to the proportion of analog dimming.

[0019] Optionally, the switching the current dimming mode according to the size relationship between the current state value and the current mode switching threshold value further comprises: adding the real-time data set, the current state value and the current mode switching threshold value to the historical data set to obtain a new historical data set; and training the prediction model according to the new historical data set, so as to optimize the transpose of the weight vector corresponding to the first threshold value, the bias of the first threshold value, the transpose of the weight vector corresponding to the second threshold value, the bias of the second threshold value, the transpose of the weight vector corresponding to the third threshold value and the bias of the third threshold value.

[0020] Optionally, the generating a corresponding control signal according to the current dimming mode and sending the control signal to the LED driving circuit comprises: judging whether the category of the current dimming mode is PWM wave dimming; if yes, generating a duty cycle value according to the current state value; generating a PWM waveform control signal according to the duty cycle value and sending the PWM waveform control signal to the LED driving circuit; and if no, generating a voltage and current control signal according to the current state value and sending the voltage and current control signal to the LED driving circuit; wherein if the current dimming mode is the hybrid mode of the low-frequency PWM dimming plus analog dimming, the PWM waveform control signal and the voltage and current control signal are generated at the same time and sent to the LED driving circuit.

[0021] Based on the same inventive concept, the application also provides a high-power LED lamp bead self-adaptive dimming system based on intelligent driving, which comprises: a historical data set construction module, which is used to collect mode switching threshold and factor data affecting the mode switching threshold, and construct a historical data set; a prediction model construction module, which is used to construct and train a prediction model for predicting the mode switching threshold according to the historical data set; a real-time data set acquisition module, which is used to periodically acquire environmental light intensity data, LED lamp bead working temperature data and user-set brightness demand data, and construct a real-time data set; a state value calculation module, which is used to calculate a current state value according to the real-time data set; a switching threshold calculation module, which is used to input the real-time data set into the prediction model, and output a current switching threshold; a dimming mode switching module, which is used to switch a current dimming mode according to the size relationship between the current state value and the current mode switching threshold; the dimming mode switching module is also used to generate a corresponding control signal according to the current dimming mode and send it to an LED driving circuit; and the LED driving circuit, which is used to control the LED lamp bead according to the corresponding control signal.

[0022] Compared with the prior art, the application has the following advantages:

[0023] 1. The application can intelligently switch the dimming mode and realize self-adaptive dimming by real-time acquisition of environmental light intensity, LED lamp bead working temperature and user-set brightness demand data, and combination of the mode switching threshold predicted by the neural network model.

[0024] 2. The application covers multiple dimming modes such as high-frequency PWM dimming mode, low-frequency PWM dimming mode, analog dimming mode and hybrid mode of low-frequency PWM dimming and analog dimming, and can select a suitable dimming mode according to actual needs, thereby improving the dimming range and precision.

[0025] 3. The application dynamically adjusts the switching threshold of the dimming mode according to environmental, LED lamp bead working temperature and user-set brightness demand, can adjust the brightness according to actual needs, avoids unnecessary energy consumption, helps energy saving and environmental protection, and provides more comfortable and energy-saving lighting experience.

[0026] Other features and advantages of the application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the application. The objects and other advantages of the application will be realized and attained by the structure particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0028] Figure 1 is a flowchart of a high-power LED lamp bead adaptive dimming method based on intelligent driving according to an embodiment of the present application.

[0029] Figure 2 is an execution flowchart of a high-power LED lamp bead adaptive dimming method based on intelligent driving according to an embodiment of the present application.

[0030] Figure 3 is a structural diagram of a high-power LED lamp bead adaptive dimming system based on intelligent driving according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0032] With reference to Figure 1 , one embodiment of the present application proposes a high-power LED lamp bead adaptive dimming method based on intelligent driving, which adopts a neural network model to predict a mode switching threshold value, and switches a dimming mode according to the size relationship between a current state value and the current mode switching threshold value, so as to realize adaptive dimming of high-power LED lamp beads, improve the dimming range and precision, and simultaneously adaptively adjust the brightness according to environmental changes.

[0033] The method of the embodiment specifically includes:

[0034] Collecting mode switching threshold values and factor data affecting the mode switching threshold values, and constructing a historical data set;

[0035] Exemplarily, it is assumed that in a certain period of time, different ambient light intensities (such as sunny, cloudy, night, etc.), different working temperatures of LED beads (such as low temperature, normal temperature, high temperature, etc.), and mode switching threshold values set by the user under various brightness requirements are recorded; for example, when the user suddenly increases the brightness requirement from low to high in sunny and normal working temperature of LED beads, the recorded mode switching threshold value can be a specific brightness value or a brightness change rate; arranging these data and corresponding mode switching threshold values into a table or database form constitutes a historical data set.

[0036] According to the historical data set, a prediction model for predicting the mode switching threshold value is constructed and trained;

[0037] Exemplarily, a neural network model is constructed according to the historical data set using a neural network algorithm in machine learning; the input layer of the model receives three characteristic values of ambient light intensity, working temperature of LED beads, and brightness requirement set by the user, and the output layer outputs the predicted mode switching threshold value; through training, the model can learn the complex relationship between these characteristic values and the mode switching threshold value; after training is completed, the model can predict the corresponding mode switching threshold value according to new input data.

[0038] Periodically collect ambient light intensity data, LED bead working temperature data, and user-set brightness requirement data to construct a real-time data set;

[0039] Exemplarily, every certain period of time (such as every second or every few minutes), the current ambient light intensity, LED bead working temperature, and brightness requirement data set by the user through a remote control or a smart device are collected through sensors; these data constitute a real-time data set for subsequent calculation and prediction.

[0040] According to the real-time data set, a current state value is calculated;

[0041] The real-time data set is input into the prediction model, and the current switching threshold value is output;

[0042] According to the size relationship between the current state value and the current mode switching threshold value, the current dimming mode is switched;

[0043] According to the current dimming mode, a corresponding control signal is generated and sent to the LED driving circuit, thereby controlling the LED beads;

[0044] The dimming mode includes a high-frequency PWM dimming mode, a low-frequency PWM dimming mode, an analog dimming mode, and a hybrid mode of low-frequency PWM dimming and analog dimming.

[0045] Optionally, the constructing and training of the prediction model for predicting the mode switching value based on the historical data set comprises:

[0046] a neural network model is established, which comprises an input layer, at least one hidden layer and an output layer;

[0047] An example is provided, in which a neural network model is established, which comprises an input layer, at least one hidden layer and an output layer. This model will be used to receive input features (i.e. ambient light intensity, operating temperature of LED beads and user-set brightness requirement) and output predicted mode switching thresholds.

[0048] Three neurons are arranged in the input layer, which respectively receive ambient light intensity, operating temperature of LED beads and user-set brightness requirement as input features;

[0049] An example is provided, in which three neurons are arranged in the input layer, each of which corresponds to an input feature. Thus, when the system collects ambient light intensity, operating temperature of LED beads and user-set brightness requirement data, these data can be transmitted to the neural network model as input features.

[0050] Three neurons are arranged in the output layer, which are respectively used to output first, second and third thresholds;

[0051] An example is provided, in which three neurons are arranged in the output layer, which are respectively used to output first, second and third thresholds. These three thresholds will be used to switch the dimming mode according to the current state value. It should be noted that the first threshold is less than the second threshold, and the second threshold is less than the third threshold, so as to ensure that the dimming mode switching logic is reasonable.

[0052] An activation function is arranged, and a prediction function of each neuron in the output layer is constructed according to the activation function;

[0053] An example is provided, in which an activation function is arranged in the output layer in order to enable the neural network model to process nonlinear problems. The selection of the activation function can be determined according to specific problems and data characteristics. Then, a prediction function of each neuron in the output layer is constructed according to the activation function, which will be used to calculate the predicted mode switching threshold.

[0054] The neural network model is trained using the historical data set to generate a stable prediction model;

[0055] The first threshold is less than the second threshold, and the second threshold is less than the third threshold.

[0056] Exemplarily, a historical data set (i.e. mode switching threshold and factors affecting mode switching threshold data in the past period of time) is used to train the neural network model; through continuous iteration and adjustment of model parameters, a stable prediction model can be obtained, which can accurately predict the mode switching threshold according to the input features.

[0057] Exemplarily, assume a historical data set containing environmental light intensity, LED lamp bead working temperature and user set brightness requirement data at different time periods in the past week, and corresponding mode switching threshold; these data can be used to train the neural network model; in the training process, the model learns the complex relationship between the input features and the mode switching threshold, and learns how to predict the future mode switching threshold according to new input data.

[0058] Optionally, as shown in Figure 2 The current state value calculated according to the real-time data set comprises:

[0059] Calculate the rate of change of environmental light intensity data, LED lamp bead working temperature data and user set brightness requirement data;

[0060] Specifically, first, real-time collection of environmental light intensity, LED lamp bead working temperature and user set brightness requirement data; these data are the basis for the system to judge whether to adjust the dimming mode; next, calculate the rate of change of these data; the rate of change can reflect the fluctuation of data in a short period of time; for example, environmental light intensity may change rapidly due to weather changes, while user set brightness requirement may change slowly due to user activity or preference.

[0061] Judge whether the rate of change is greater than a fourth threshold;

[0062] Specifically, a fourth threshold is set to judge whether the data change is significant; this threshold can be adjusted according to actual conditions to ensure that the system can respond sensitively to environmental changes, while avoiding too frequent dimming mode switching.

[0063] If yes, calculate the current state value according to the real-time data set;

[0064] If not, maintain the current dimming mode.

[0065] Specifically, if any of the change rate of ambient light intensity, the working temperature of LED lamp beads or the user-set brightness requirement data is greater than the fourth threshold value, it is considered that the current environment or user requirement has changed significantly, and the current state value needs to be calculated and the dimming mode needs to be considered to be switched; if the change rate of all data is less than or equal to the fourth threshold value, the system considers that the current environment is relatively stable, and the user requirement has not changed significantly, so the current dimming mode is maintained unchanged; if the change rate of data is greater than the fourth threshold value, the current state value is calculated according to the real-time data set; this state value can be a value that comprehensively reflects the current overall state of the ambient light intensity, the working temperature of LED lamp beads and the user-set brightness requirement; the calculated current state value is compared with the current switching threshold value output by the previously trained prediction model; according to the comparison result, it is determined whether the dimming mode needs to be switched, and the corresponding control signal is generated and sent to the LED driving circuit.

[0066] Exemplarily, it is assumed that at a certain time point, it is detected that the ambient light intensity suddenly becomes dark, and the change rate is much greater than the fourth threshold value; at the same time, the working temperature of the LED lamp beads slightly rises, but the change rate is still within the threshold value range; the user-set brightness requirement does not change; in this case, it is considered that the significant change of the ambient light intensity is the main one, so the current state value is calculated and compared with the switching threshold value output by the prediction model; if the current state value is lower than a certain threshold value (such as the first threshold value), the brighter dimming mode can be switched to adapt to the change of the ambient light; and if the current state value is still within the threshold value range, the current dimming mode is maintained unchanged; by introducing the judgment logic of the change rate, the environmental change can be more intelligently responded, unnecessary dimming mode switching can be avoided, and a more comfortable and energy-saving lighting experience can be provided.

[0067] Optionally, the calculation of the current state value according to the real-time data set comprises:

[0068] normalizing the ambient light intensity data, the working temperature data of the LED lamp beads and the user-set brightness requirement data;

[0069] Specifically, first, the ambient light intensity, the working temperature of the LED lamp bead and the brightness requirement set by the user are collected in real time; these data are usually expressed in different units and ranges, for example, the ambient light intensity can be expressed in lux (lx), the working temperature of the LED lamp bead can be expressed in Celsius (°C), and the brightness requirement set by the user can be expressed in percentage or specific brightness value; in order to enable these data to be compared and calculated on the same scale, they are normalized; normalization is a process of scaling data to a specific range (usually 0 to 1); for example, for the ambient light intensity data, the system can scale it to the range of 0 (complete darkness) to 1 (maximum light intensity); similarly, the working temperature of the LED lamp bead and the brightness requirement set by the user are also scaled to the corresponding ranges.

[0070] The first weight of the ambient light intensity, the second weight of the working temperature of the LED lamp bead and the third weight of the brightness requirement set by the user are set;

[0071] Specifically, the first weight (W1) of the ambient light intensity, the second weight (W2) of the working temperature of the LED lamp bead and the third weight (W3) of the brightness requirement set by the user are set; these weights reflect the degree of influence of each factor on the current state value; for example, if the ambient light intensity is considered the most important factor, a higher value can be assigned to W1; on the contrary, if the brightness requirement set by the user is considered secondary, a lower value can be assigned to W3.

[0072] According to the first weight, the second weight, the third weight and the normalized data, the current state value is calculated; for the current state value S, there are

[0073] In the formula, W1 is the first weight, W2 is the second weight, W3 is the third weight, S1 is the normalized ambient light intensity data, S2 is the normalized working temperature of the LED lamp bead data, and S3 is the normalized brightness requirement set by the user.

[0074] Specifically, according to the normalized data and the set weights, the system calculates the current state value.

[0075] ​​​​​​​​​​​​​Exemplarily, assume that the real-time data collected by the system are as follows: the intensity of ambient light is 500 lux (normalized to 0.5), the working temperature of the LED lamp bead is 40 degrees Celsius (normalized to 0.6, assuming that the temperature range is 0 to 100 degrees Celsius), and the brightness requirement set by the user is 70% (normalized to 0.7); meanwhile, the weights are set as follows: = 0.5 (the importance of the intensity of ambient light is medium), = 0.3 (the importance of the working temperature of the LED lamp bead is lower), = 0.2 (the importance of the brightness requirement set by the user is the lowest); according to these data and weights, the system can calculate the current state value = 0.5x0.5+0.3x0.6+0.2x0.7=0.25+0.18+0.14=0.57; this current state value reflects the current overall state, which can be compared with the switching threshold value output by the prediction model to determine whether the dimming mode needs to be switched.

[0076] Optionally, the inputting the real-time data set into the prediction model and outputting a current switching threshold value comprises:

[0077] normalizing the intensity of ambient light data, the working temperature data of the LED lamp bead, and the brightness requirement set by the user;

[0078] Specifically, the intensity of ambient light, the working temperature of the LED lamp bead, and the brightness requirement set by the user are collected in real time, so that these data can be compared and calculated on the same scale, and they are normalized.

[0079] The normalized data are input into the prediction model, thereby obtaining a current first threshold value, a current second threshold value, and a current third threshold value; for the current first threshold value , there is ,

[0080] wherein, is an activation function, is the transpose of the weight vector corresponding to the current first threshold value, is the bias of the current first threshold value,

[0081] for the current second threshold value , there is ,

[0082] wherein, is the transpose of the weight vector corresponding to the current second threshold value, is the bias corresponding to the current second threshold value,

[0083] For the current third threshold ,have ,

[0084] In the formula, This is the transpose of the weight vector corresponding to the current third threshold. This is the bias corresponding to the current third threshold;

[0085] The current switching threshold includes the current first threshold, the current second threshold, and the current third threshold.

[0086] For example, the normalized data As input features, they are fed into a pre-trained neural network prediction model; the prediction model contains three output neurons, each corresponding to the current first threshold (…). ), the current second threshold ( ) and the current third threshold ( For each output neuron, the model uses the transpose of the corresponding weight vector. , , ) and bias ( , , The output value is calculated using an activation function; the calculation process uses an activation function. (For example, it could be the sigmoid function, ReLU function, etc.), which transforms the linear combination of input features into a non-linear output; after the prediction model completes its calculation, it outputs the current first threshold. ), the current second threshold ( ) and the current third threshold ( ).

[0087] Optionally, such as Figure 2 As shown, switching the current dimming mode based on the relationship between the current state value and the current mode switching threshold includes:

[0088] Determine whether the current state value is greater than the current first threshold; if not, switch the analog dimming mode.

[0089] Specifically, the current state value The dimming mode should be compared to four thresholds to determine which dimming mode should be switched to; if < For applications requiring precise brightness control, such as reading lights and artistic lighting, analog dimming offers a smooth brightness transition and stable luminous effect. Therefore, switching to analog dimming mode allows for adjusting brightness by changing the current of the LED.

[0090] If yes, determine whether the current state value is greater than the current second threshold value; if no, switch to the low-frequency PWM dimming mode;

[0091] Specifically, if < < the current environment is considered moderate or the user needs moderate brightness, and thus the low-frequency PWM (Pulse Width Modulation) dimming mode is switched to; low-frequency PWM dimming adjusts brightness by changing the switching duty cycle of LEDs, which is suitable for scenes that are not very sensitive to brightness changes.

[0092] If yes, determine whether the current state value is greater than the current third threshold value; if no, switch to the hybrid mode of low-frequency PWM dimming plus analog dimming.

[0093] Specifically, if < < the current environment is considered relatively bright or the user needs relatively high brightness, but it has not reached the level requiring high-frequency PWM dimming, and thus the hybrid mode of low-frequency PWM dimming plus analog dimming is switched to; this mode combines the advantages of the previous two modes, and can improve dimming efficiency and accuracy to a certain extent.

[0094] If yes, switch to the high-frequency PWM dimming mode.

[0095] Specifically, if < the current environment is considered very bright or the user needs very high brightness, and thus the high-frequency PWM dimming mode is switched to; high-frequency PWM dimming reduces the flicker phenomenon visible to the human eye by increasing the switching frequency, which is suitable for scenes that require high brightness and are sensitive to flicker.

[0096] Exemplarily, assume that the current state value is 0.6, and the threshold values obtained by training of the system are =0.5, =0.7, =0.9; first compare with , find =0.6> =0.5, so the condition for switching to the analog dimming mode is not met; then compare with , find =0.6< =0.7, so the condition for switching to the low-frequency PWM dimming mode is met; since has not exceeded Therefore, there is no need to consider switching to the hybrid mode or the high-frequency PWM dimming mode; thus, eventually switching to the low-frequency PWM dimming mode; this process is automatic, without manual intervention by the user; in this way, the intelligent LED lighting system can automatically adjust the dimming mode according to real-time environmental conditions and user needs, providing the best lighting effect and user experience.

[0097] Optionally, the switching of the low-frequency PWM dimming plus analog dimming hybrid mode comprises:

[0098] calculating a proportion of analog dimming according to the current state value, the current second threshold value and the current third threshold value, for the proportion of analog dimming , there is

[0099] ;

[0100] Specifically, according to the given formula, the proportion of analog dimming can be calculated, the current state value is linked with the two threshold values and , and the proportion of analog dimming in the hybrid mode is determined by calculating the position of relative to and .

[0101] Exemplarily, assuming that the current state value is 0.75, the second threshold value is 0.7, and the third threshold value is 0.8; substituting these values into the formula, we get =(0.75-0.7) / (0.8-0.7)x100%=50%; this means that in the hybrid mode, analog dimming should account for 50% of the proportion, while low-frequency PWM dimming accounts for the remaining 50%.

[0102] Switching of the low-frequency PWM dimming plus analog dimming hybrid mode according to the proportion of analog dimming.

[0103] Exemplarily, once the proportion of analog dimming is calculated, the system can switch the hybrid dimming mode according to this proportion; this usually involves adjusting the parameters of the LED driving circuit to output both a low-frequency PWM signal and an analog dimming signal at the same time; in actual implementation, a digital-to-analog converter (DAC) can be used to generate the analog dimming signal, and a PWM controller can be used to generate the low-frequency PWM signal; then, the two signals can be combined together to produce the desired hybrid dimming effect.

[0104] Optionally, switching the current dimming mode based on the relationship between the current state value and the current mode switching threshold further includes:

[0105] The real-time dataset, the current state value, and the current mode switching threshold are added to the historical dataset to obtain a new historical dataset;

[0106] Specifically, data such as ambient light intensity, LED chip operating temperature, and user-defined brightness requirements are collected in real time to form a real-time dataset; based on the real-time dataset and the prediction model, the current state value is calculated. , set the current state value Compared with the three preset thresholds (first threshold) Second threshold Third threshold The comparison is performed, and based on the comparison result, the corresponding dimming mode is switched (such as analog dimming, low-frequency PWM dimming, hybrid dimming, or high-frequency PWM dimming). After the dimming mode switching is completed, the real-time dataset and current state value are compared. and the current mode switching threshold (i.e. , , Add the historical dataset; in this way, the historical dataset contains real-time data, status values ​​and switching thresholds at multiple time points, as well as the corresponding dimming mode switching results.

[0107] The prediction model is trained based on the new historical dataset, thereby optimizing the transpose of the weight vector corresponding to the first threshold, the bias of the first threshold, the transpose of the weight vector corresponding to the second threshold, the bias of the second threshold, the transpose of the weight vector corresponding to the third threshold, and the bias of the third threshold.

[0108] Specifically, the prediction model is retrained using a new historical dataset. During the training process, parameters in the prediction model can be adjusted, such as the transpose of the weight vector corresponding to the first threshold, the bias of the first threshold, the transpose of the weight vector corresponding to the second threshold, the bias of the second threshold, the transpose of the weight vector corresponding to the third threshold, and the bias of the third threshold. The purpose of adjusting these parameters is to enable the prediction model to more accurately reflect changes in the actual environment and user needs, thereby more accurately calculating the current state value and making more reasonable dimming mode switching decisions.

[0109] For example, suppose that at a certain time point t, the collected real-time dataset is {ambient light intensity: 500 lx, LED operating temperature: 40℃, user brightness requirement: 70%}; based on the prediction model, the current state value is calculated. It is 0.65; Compared with the threshold, it was found that Falling on the second threshold (assuming 0.6) and the third threshold value (assuming 0.7), thus switching to the hybrid dimming mode; the real-time data set, the state value and the second threshold value and the third threshold value The historical data set is added; at a certain time point in the future, the prediction model is retrained using the new historical data set, and the model parameters are adjusted to optimize performance; in this way, the intelligent LED lighting system can continuously learn and adapt to changes in the environment and changes in user needs, thereby providing more accurate and efficient dimming services.

[0110] Optionally, the generating a corresponding control signal according to the current dimming mode and sending it to the LED drive circuit comprises:

[0111] determining whether the category of the current dimming mode is PWM wave dimming;

[0112] If yes, generating a duty cycle value according to the current state value;

[0113] generating a PWM waveform control signal according to the duty cycle value and sending it to the LED drive circuit;

[0114] For example, assuming that the current dimming mode is PWM wave dimming and the current state value has been calculated according to the real-time data set and the prediction model, it is determined that the category of the current dimming mode is PWM wave dimming; according to the current state value , a duty cycle value is calculated; this duty cycle value determines the proportion of time that the LED lamp beads are on in a PWM cycle; when the duty cycle value is large, the LED lamp beads are on for a longer time in each cycle, so the average brightness is higher; conversely, when the duty cycle value is small, the LED lamp beads are on for a shorter time in each cycle, so the average brightness is lower; for example, if is 0.5, the duty cycle value can be 50%, meaning that the LED lamp beads are on for half the time in each cycle, so the average brightness of the LED will be half the maximum brightness; a PWM waveform control signal is generated according to the duty cycle value, and this signal is a periodic high-low level signal, the length of the high level being determined by the duty cycle value; the generated PWM waveform control signal is sent to the LED drive circuit, and the drive circuit controls the lighting and extinguishing of the LED lamp beads according to this signal, thereby achieving the dimming effect.

[0115] If no, generating a voltage and current control signal according to the current state value and sending it to the LED drive circuit;

[0116] For example, assuming that the current dimming mode is analog dimming and the current state value ; judging that the current dimming mode is analog dimming according to the current state value , a voltage or current control value V / I is calculated; the value determines the voltage or current flowing through the LED lamp bead, thereby controlling the brightness of the LED; a voltage and current control signal is generated according to the voltage / current control value V / I; the signal can be an analog voltage signal or a digital signal (converted into an analog voltage by a DAC), used to control the output voltage or current of the LED driving circuit; the generated voltage and current control signal is sent to the LED driving circuit, and the driving circuit adjusts the voltage or current of the LED lamp bead according to the signal, thereby realizing the dimming effect.

[0117] If the current dimming mode is a hybrid mode of low-frequency PWM dimming and analog dimming, the PWM waveform control signal and the voltage and current control signal are generated and sent to the LED driving circuit at the same time.

[0118] Exemplarily, assuming that the current dimming mode is a hybrid mode of low-frequency PWM dimming and analog dimming, and the current state value has also been calculated ; judging that the current dimming mode is a hybrid mode; according to the current state value , a duty cycle value and a voltage / current control value V / I are calculated respectively, and the two values together determine the brightness of the LED in the hybrid dimming mode; a PWM waveform control signal is generated according to the duty cycle value, and a voltage and current control signal is generated according to the voltage / current control value V / I; the two control signals are sent to the LED driving circuit at the same time, and the driving circuit can include two independent control channels, one for receiving the PWM waveform control signal and the other for receiving the voltage and current control signal; the two signals act together to realize the hybrid effect of low-frequency PWM dimming and analog dimming.

[0119] Based on the same inventive concept, as shown in Figure 3 , the application also provides a high-power LED lamp bead adaptive dimming system based on intelligent driving, which comprises:

[0120] A historical data set construction module is configured to collect mode switching threshold and factor data affecting the mode switching threshold, and construct a historical data set.

[0121] A prediction model construction module is configured to construct and train a prediction model for predicting the mode switching threshold according to the historical data set.

[0122] A real-time data set acquisition module is configured to periodically collect ambient light intensity data, LED lamp bead working temperature data and user-set brightness demand data, and construct a real-time data set.

[0123] A state value calculation module is configured to calculate a current state value according to the real-time data set;

[0124] A switching threshold calculation module is configured to input the real-time data set into the prediction model and output a current switching threshold;

[0125] A dimming mode switching module is configured to switch a current dimming mode according to a size relationship between the current state value and the current mode switching threshold, and to generate a corresponding control signal according to the current dimming mode and send the control signal to an LED driving circuit;

[0126] The LED driving circuit is configured to control the LED lamp beads according to the corresponding control signal.

[0127] It should be noted that the electrical connection between the above-mentioned various units does not necessarily represent the connection between the lines, and the indirect connection mode can be applied to the embodiments of the present application as long as the purpose of the present application is achieved. The above is only an exemplary embodiment of the present application, and cannot limit the scope of the present application.

[0128] That is, any equivalent changes and modifications made according to the teachings of the present application are still within the scope of the present application. Other embodiments of the present application will be readily apparent to those skilled in the art upon considering the specification and practice of the true principles disclosed herein. The present application is intended to cover any variations, uses, or adaptive changes of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not disclosed in the present application.

Claims

1. A self-adaptive dimming method for high-power LED lamp beads based on intelligent driving, characterized in that The method comprises: collecting mode switching threshold and factors affecting mode switching threshold data, and constructing a historical data set; constructing and training a prediction model for predicting mode switching threshold according to the historical data set; periodically collecting ambient light intensity data, LED lamp bead working temperature data, and user-set brightness demand data, and constructing a real-time data set; calculating a current state value according to the real-time data set; inputting the real-time data set into the prediction model to output a current mode switching threshold; switching a current dimming mode according to the size relationship between the current state value and the current mode switching threshold; generating a corresponding control signal according to the current dimming mode and sending it to an LED driving circuit to control the LED lamp bead; wherein the dimming mode includes a high-frequency PWM dimming mode, a low-frequency PWM dimming mode, an analog dimming mode, and a hybrid mode of low-frequency PWM dimming plus analog dimming; constructing and training a prediction model for predicting mode switching threshold according to the historical data set comprises: establishing a neural network model including an input layer, at least one hidden layer, and an output layer; setting three neurons in the input layer to receive ambient light intensity, LED lamp bead working temperature, and user-set brightness demand as input features, respectively; setting three neurons in the output layer to output a first threshold value, a second threshold value, and a third threshold value, respectively; setting an activation function and constructing a prediction function for each neuron in the output layer according to the activation function; training the neural network model using the historical data set to generate a stable prediction model; wherein the first threshold value is less than the second threshold value, and the second threshold value is less than the third threshold value; calculating a current state value according to the real-time data set comprises: normalizing ambient light intensity data, LED lamp bead working temperature data, and user-set brightness demand data; setting a first weight for ambient light intensity, a second weight for LED lamp bead working temperature, and a third weight for user-set brightness demand; According to the first weight, the second weight, the third weight and the normalized data, the current state value is calculated, and the current state value , has ; in the formula, is the first weight, is the second weight, is the third weight, is the normalized ambient light intensity data, is the normalized LED lamp bead working temperature data, is the normalized user set brightness requirement data. inputting the real-time data set into the prediction model to output a current mode switching threshold comprises: normalizing ambient light intensity data, LED lamp bead working temperature data, and user-set brightness demand data; The normalized data is input into the prediction model, so as to obtain a current first threshold value, a current second threshold value and a current third threshold value, for the current first threshold value , there is ; in the formula, is an activation function, is a transpose of a weight vector corresponding to the current first threshold value, is a bias of the current first threshold value, for the current second threshold value , there is ; in the formula, is a transpose of a weight vector corresponding to the current second threshold value, is a bias corresponding to the current second threshold value, for the current third threshold value , have ; in which is a bias corresponding to the current third threshold value; wherein the current mode switching threshold includes the current first threshold value, the current second threshold value, and the current third threshold value; switching a current dimming mode according to the size relationship between the current state value and the current mode switching threshold comprises: determining whether the current state value is greater than the current first threshold value; if not, switching to the analog dimming mode; if yes, determining whether the current state value is greater than the current second threshold value; if not, switching to the low-frequency PWM dimming mode; if yes, determining whether the current state value is greater than the current third threshold value; if not, switching to the hybrid mode of low-frequency PWM dimming plus analog dimming; if yes, switching to the high-frequency PWM dimming mode. switching the hybrid mode of low frequency PWM dimming plus analog dimming includes: calculating a proportion of analog dimming according to the current state value, the current second threshold value and the current third threshold value, for the proportion of analog dimming , there is ; switching the hybrid mode of low frequency PWM dimming plus analog dimming according to the proportion of analog dimming; The switching of the current dimming mode according to the size relationship between the current state value and the current mode switching threshold further comprises: adding the real-time data set, the current state value and the current mode switching threshold into the historical data set to obtain a new historical data set; training the prediction model according to the new historical data set, thereby optimizing the transpose of the weight vector corresponding to the current first threshold, the bias of the current first threshold, the transpose of the weight vector corresponding to the current second threshold, the bias of the current second threshold, the transpose of the weight vector corresponding to the current third threshold, and the bias of the current third threshold; The generation of the corresponding control signal according to the current dimming mode and the sending of the control signal to the LED driving circuit comprises: judging whether the category of the current dimming mode is PWM wave dimming; if yes, generating a duty cycle value according to the current state value; generating a PWM waveform control signal according to the duty cycle value and sending the PWM waveform control signal to the LED driving circuit; if no, generating a voltage and current control signal according to the current state value and sending the voltage and current control signal to the LED driving circuit; wherein if the current dimming mode is a hybrid mode of low-frequency PWM dimming and analog dimming, the PWM waveform control signal and the voltage and current control signal are generated simultaneously and sent to the LED driving circuit. 2.The self-adaptive dimming method of high-power LED lamp bead based on intelligent driving according to claim 1, characterized in that The calculation of the current state value according to the real-time data set comprises: calculating the change rates of the ambient light intensity data, the working temperature data of the LED lamp beads and the brightness demand data set by the user; judging whether the change rates are greater than a fourth threshold; if yes, calculating the current state value according to the real-time data set; if no, maintaining the current dimming mode.

Images