Self-adaptive LED illumination regulation and control method based on multi-mode environment perception

Abstract

The invention relates to the technical field of electroluminescent light sources, and provides a self-adaptive LED illumination regulation and control method based on multi-modal environment perception, which comprises the following steps: acquiring multi-modal environment perception data, user feedback data, blue light risk level and user physiological rhythm data; performing space-time alignment on the multi-modal environment sensing data and the user feedback data, performing data enhancement, and extracting a fusion feature vector in an enhanced data set; performing dynamic weighting on the fused feature vector, generating an initial illumination parameter in combination with the weighted feature vector and the user physiological rhythm data, and performing energy-saving optimization on the initial illumination parameter to obtain a target illumination parameter; mapping the target illumination parameter into a pulse width modulation parameter, and determining the phase delay of the micro-lens array according to the personnel distribution characteristics in the fusion characteristics; driving constant-current circuit data according to the pulse width modulation parameters, performing micro-lens beam deflection according to the phase delay of the micro-lens array, and adjusting the optical filter according to the blue light risk level.

Description

Technical Field

[0001] This application relates to the technical field of electroluminescent light sources. Specifically, it relates to an adaptive LED lighting control method based on multi-modal environmental perception. Background Art

[0002] With the rapid development of Internet of Things and artificial intelligence technologies, intelligent lighting systems, as an important part of building energy management, are undergoing a profound transformation from simple control to intelligent regulation. Currently, in the field of LED (light-emitting diode) lighting, basic technologies such as ambient light sensing, occupancy detection, and timing control are mainly used to achieve energy-saving optimization.

[0003] However, existing intelligent lighting systems mostly rely on single or limited types of sensors (such as only light sensors or occupancy sensors), lacking the synchronous acquisition and fusion analysis of multi-dimensional data such as ambient light, personnel distribution, visual comfort, and natural environment, resulting in one-sided regulation strategies. In addition, traditional lighting strategies are mostly based on preset thresholds or fixed scenarios, unable to perform real-time dynamic optimization according to the time-varying characteristics of the environment (such as gradual change of natural light and change of personnel flow patterns), especially lacking the ability to quickly respond to sudden scene changes. Most systems ignore the quantitative evaluation and feedback adjustment of visual comfort during the energy-saving optimization process, and it is difficult to maximize energy efficiency while ensuring user comfort, especially in places such as offices and education that require high visual quality.

[0004] Therefore, how to improve the globality, real-time performance, and comfort of LED lighting control is an urgent problem to be solved currently. Summary of the Invention

[0005] In view of the above problems existing in the prior art, the purpose of this application is to propose an adaptive LED lighting control method based on multi-modal environmental perception, so as to at least solve the technical problem of how to improve the globality, real-time performance, and comfort of LED lighting control.

[0006] To achieve the above object and other related objects, this application provides an adaptive LED lighting control method based on multi-modal environmental perception, and the method includes: Obtain multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located; Perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fused feature vectors in the enhanced data set; Dynamically weight the fused feature vector, generate initial lighting parameters by combining the dynamically weighted feature vector and the user's physiological rhythm data, and perform energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; Map the target lighting parameters to pulse width modulation parameters, and determine the phase delay of the microlens array according to the personnel distribution characteristics in the fused feature vector; Drive the data of the constant current circuit according to the pulse width modulation parameters, deflect the microlens beam according to the phase delay of the microlens array, and adjust the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array.

[0007] In an embodiment of the present application, the acquisition of multi-modal environmental perception data, user feedback data, blue light risk level, and user physiological rhythm data in the environment where the target LED light source array is located includes: Obtain the multi-modal environmental perception data through a distributed sensor network pre-configured in the environment where the target LED light source array is located, where the multi-modal environmental perception data includes environmental spectral data, personnel distribution data, visual image data, and environmental parameter data; Obtain the user feedback data through a pre-configured user terminal; Perform blue light hazard analysis based on the environmental spectral data to obtain the blue light risk level; Obtain the user's heart rate change data and visual change data through a pre-configured photoplethysmogram sensor and an infrared pupil tracker, and generate the user's physiological rhythm data according to the heart rate change data and the visual change data.

[0008] In an embodiment of the present application, the spatio-temporal alignment processing of the multi-modal environmental perception data and the user feedback data to obtain an aligned data set, and when the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fused feature vector in the enhanced data set, including: Perform timestamp alignment processing on the multi-modal environmental perception data and the user feedback data to obtain an initial aligned data set; Map the initial aligned data set to a unified coordinate system through a pre-configured edge computing node and a perspective transformation matrix; Use a sliding window to perform cubic spline interpolation on the data in the unified coordinate system and synchronize it to a unified timestamp to obtain a target aligned data set; When the ambient illumination data in the target alignment dataset is lower than the preset illumination threshold, transmit the ambient spectral data and personnel distribution data to the visible light conversion channel and the heat map conversion channel in the cross-modal generative adversarial network to obtain enhanced visible light image data and enhanced personnel distribution heat map data; Transmit the enhanced dataset to a pre-configured feature extractor to obtain equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features, and fuse the equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features to obtain the fused feature vector.

[0009] In an embodiment of the present application, dynamically weighting the fused feature vector, generating an initial lighting parameter by combining the dynamically weighted feature vector and the user's physiological rhythm data, and performing energy-saving optimization on the initial lighting parameter through a pre-trained energy level model to obtain the target lighting parameter, including: According to the fused feature vector, determine the feature change rate, use the feature change rate as the importance, and dynamically weight the fused feature vector to obtain the dynamically weighted feature vector; According to the user's physiological rhythm data, determine the target lighting mode, and perform constraint optimization by combining the dynamically weighted feature vector and the target lighting mode to obtain the initial lighting parameter, where the initial lighting parameter includes a lamp brightness coefficient, a lamp color temperature coefficient, and a lamp dimming gain coefficient; Transmit the initial lighting parameter to the pre-trained energy level model to generate a light source array matrix and a ground state energy parameter, and perform energy-saving optimization on the lamp brightness coefficient in the initial lighting parameter according to the light source array matrix and the ground state energy parameter to obtain the target lighting parameter.

[0010] In an embodiment of the present application, mapping the target lighting parameter to a pulse width modulation parameter, and determining the phase delay of the microlens array according to the personnel distribution feature in the fused feature vector, including: According to the target lighting parameter, determine the duty cycle, frequency, and phase corresponding to the pulse width modulation, optimize the frequency, and compensate the phase to obtain the pulse width modulation parameter; According to the personnel distribution feature, determine the personnel's visual azimuth and elevation angles, and generate the phase delay of the microlens array according to the personnel's visual azimuth and elevation angles.

[0011] In an embodiment of the present application, driving the data of the constant current circuit according to the pulse width modulation parameter, deflecting the micro-lens light beam according to the phase delay of the micro-lens array, and adjusting the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array, including: According to the pulse width modulation parameter, perform feedback modulation on the voltage and current of the constant current circuit, and output an adjustable constant current source; According to the phase delay of the micro-lens array, determine the beam calibration angle, and based on the beam calibration angle, deflect the light beam of the micro-lens; When the blue light risk level is medium risk, activate the pre-configured amber LED channel, or when the blue light risk level is high risk, insert a blue light filter and perform brightness limiting.

[0012] In an embodiment of the present application, the method further includes: Obtain the comfort score of the user for the target LED light source array after adaptive lighting control, and the actual energy consumption data corresponding to the target LED light source array after adaptive lighting control; Update the target lighting parameters according to the comfort score and the actual energy consumption data.

[0013] In an embodiment of the present application, there is also provided an adaptive LED lighting control device based on multi-modal environment perception, and the device includes: A data acquisition module, configured to acquire multi-modal environment perception data, user feedback data, blue light risk level, and user physiological rhythm data in the environment where the target LED light source array is located; A feature fusion module, configured to perform spatio-temporal alignment processing on the multi-modal environment perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fusion feature vector in the enhanced data set; A lighting parameter generation module, configured to dynamically weight the fusion feature vector, generate initial lighting parameters by combining the dynamically weighted feature vector and the user physiological rhythm data, and perform energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; A parameter mapping module, configured to map the target lighting parameters to pulse width modulation parameters, and determine the phase delay of the micro-lens array according to the personnel distribution characteristics in the fusion feature vector; A control module is configured to drive data of a constant current circuit according to the pulse width modulation parameter, deflect a micro-lens light beam according to the phase delay of the micro-lens array, and adjust an optical filter of the target LED light source array according to the blue light risk level, so as to complete the adaptive illumination control of the target LED light source array.

[0014] In an embodiment of the present application, there is also provided a computer-readable storage medium, which includes a stored computer program, wherein when the computer program runs, it executes the above-mentioned adaptive LED illumination control method based on multi-modal environment perception.

[0015] In an embodiment of the present application, there is also provided an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above-mentioned adaptive LED illumination control method based on multi-modal environment perception through the computer program.

[0016] Advantages of the present invention: First, obtain multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located; then, perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fusion feature vectors in the enhanced data set; then, perform dynamic weighting on the fusion feature vectors, combine the dynamically weighted feature vectors and the user circadian rhythm data to generate initial lighting parameters, and perform energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; then, map the target lighting parameters to pulse width modulation parameters, and determine the phase delay of the microlens array according to the personnel distribution characteristics in the fusion feature vectors; finally, drive the data of the constant current circuit according to the pulse width modulation parameters, perform microlens beam deflection according to the microlens array phase delay, and adjust the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array. In this application, by obtaining multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data, a three-dimensional data system is constructed; when the environmental illuminance is insufficient, data enhancement through a cross-modal generative adversarial network fills problems such as blurred visual images and sparse personnel distribution data in low-illuminance scenarios, generating a more complete environmental representation. Extract fusion feature vectors including equivalent illuminance, personnel distribution, visual comfort, and environmental interference to realize the comprehensive modeling of global elements such as environmental light quality, personnel activity patterns, and visual health risks, enabling the control strategy to cover multiple objectives such as spatial distribution, spectral safety, and physiological adaptation. Dynamic weighting can adjust the feature weights in real time, prioritize responses to key changes, and avoid the lag response of fixed weights to complex scenarios. The energy level model regards the light source array as an energy system, and through optimizing the light source matrix and ground state energy parameters, balances the lighting requirements and energy consumption of each region globally, achieving global energy efficiency optimization of light supply on demand and avoiding energy waste caused by local over-illumination. Calculate the phase delay of the microlens array in real time according to the personnel distribution characteristics, and realize the dynamic deflection of the beam direction through an electronically controlled microlens, avoiding the lag of traditional mechanical adjustment. The real-time analysis of the blue light risk level triggers the dynamic switching of the optical filter, reducing the response time to sudden high-blue light scenarios. Dynamically identify the fatigue state or circadian rhythm phase through user circadian rhythm data and generate adapted initial lighting parameters. The microlens beam deflection accurately adjusts the optical path according to the personnel distribution characteristics, avoiding direct glare from the lamp and improving the comfort of LED lighting control BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings: Figure 1 is a schematic diagram of an application environment of an adaptive LED lighting control method based on multi-modal environment perception shown in an exemplary embodiment of the present application; Figure 2 is a schematic flowchart of an adaptive LED lighting control method based on multi-modal environment perception shown in an exemplary embodiment of the present application; Figure 3 is a schematic diagram of an adaptive LED lighting control device based on multi-modal environment perception shown in an exemplary embodiment of the present application; Figure 4 is a schematic diagram of the structure of an electronic device shown in an exemplary embodiment of the present application. Detailed implementation manners

[0018] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0019] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned accompanying drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0020] In an embodiment of the present application, an adaptive LED lighting control method based on multi-modal environment perception is provided. Optionally, as an alternative implementation manner, the above-mentioned adaptive LED lighting control method based on multi-modal environment perception can be but is not limited to being applied to an environment such as Figure 1 as shown. Figure 1It is a schematic diagram of the application environment of the adaptive LED lighting control method based on multi-modal environmental perception shown in an exemplary embodiment of the present application. Refer to Figure 1 , this implementation environment includes a target LED light source array 110, a sensor 120, and a control terminal 130. The target LED light source array 110, the sensor 120, and the control terminal 130 can communicate through a network, but are not limited to this. The control terminal 130 can perform operations on a database, for example, write data operations or read data operations. The above control terminal 130 can include, but is not limited to, a human-computer interaction screen, a processor, and a memory. The above human-computer interaction screen can be used to display the results of LED lighting control. The above processor can be used to respond to the above human-computer interaction operations, execute corresponding operations, or generate corresponding instructions, and send the generated instructions to the target LED light source array 110. The above memory is used to store relevant stored data.

[0021] As an optional method, data can be collected by the sensor 120, for example, multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located are collected and preprocessed.

[0022] As an optional method, the following steps in the adaptive LED lighting control method based on multi-modal environmental perception can be executed on the control terminal 130: Obtain multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located; Perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fused feature vectors in the enhanced data set; Dynamically weight the fused feature vectors, combine the dynamically weighted feature vectors and the user circadian rhythm data to generate initial lighting parameters, and perform energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; Map the target lighting parameters to pulse width modulation parameters, and determine the phase delay of the microlens array according to the personnel distribution characteristics in the fused feature vectors; Drive the data of the constant current circuit according to the pulse width modulation parameters, deflect the microlens beam according to the phase delay of the microlens array, and adjust the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array.

[0023] In the above method, a three-dimensional data system is constructed by obtaining multi-modal environmental perception data, user feedback data, blue light risk levels, and user circadian rhythm data. When the environmental illuminance is insufficient, data augmentation is performed through a cross-modal generative adversarial network to address issues such as blurred visual images and sparse personnel distribution data in low-illuminance scenarios, generating a more complete environmental representation. A fused feature vector containing equivalent illuminance, personnel distribution, visual comfort, and environmental interference is extracted to comprehensively model global factors such as environmental light quality, personnel activity patterns, and visual health risks, enabling the control strategy to cover multiple objectives including spatial distribution, spectral safety, and physiological adaptation. Dynamic weighting can adjust the feature weights in real time, prioritizing responses to critical changes and avoiding the lagged response of fixed weights to complex scenarios. The energy level model treats the light source array as an energy system and optimizes the light source matrix and ground state energy parameters to balance the lighting requirements and energy consumption in each region globally, achieving global energy efficiency optimization for on-demand light supply and avoiding energy waste caused by local over-illumination. The phase delay of the microlens array is calculated in real time based on the personnel distribution characteristics, and the dynamic deflection of the light beam is achieved through an electronically controlled microlens, avoiding the lag of traditional mechanical adjustment. The real-time analysis of the blue light risk level triggers the dynamic switching of the optical filter, reducing the response time to sudden high-blue light scenarios. The fatigue state or circadian rhythm phase is dynamically identified through user circadian rhythm data, generating appropriate initial lighting parameters. The microlens beam deflection precisely adjusts the optical path according to the personnel distribution characteristics, avoiding direct glare from the luminaires and improving the comfort of LED lighting control. Optionally, in this embodiment, the above network may include, but is not limited to: a wireless network, where the wireless network includes: Bluetooth, WIFI, and other networks that enable wireless communication. The above control terminal 130 may be a server, and the server may be a single server or a server cluster composed of multiple servers. The above is only an example, and this embodiment does not make any limitations in this regard.

[0024] As an optional example, this embodiment does not limit the execution entity of the above adaptive LED lighting control method based on multi-modal environmental perception. Some or all of the steps of the above adaptive LED lighting control method based on multi-modal environmental perception may be executed on the control terminal 130.

[0025] In an embodiment of the present application, an adaptive LED lighting control method based on multi-modal environmental perception is provided. Figure 2 is a flowchart of the adaptive LED lighting control method based on multi-modal environmental perception shown in an exemplary embodiment of the present application. Refer to Figure 2 The adaptive LED lighting control method based on multi-modal environmental perception includes the steps described in S210 to S250 below: In step S210, multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located are obtained.

[0026] Among them, the target LED light source array refers to a combination of LED lamps deployed in a specific target area (such as an office, a classroom, a meeting room), which consists of multiple independently controllable LED light-emitting units. Each unit can adjust parameters such as brightness, color temperature, and beam direction to form a light source system with spatial distribution characteristics. As the control object, the target LED light source array realizes the refined control of lighting parameters (such as area directional lighting, dynamic spot adjustment) through integrated execution devices such as a micro-lens array, an optical filter, and a constant-current circuit.

[0027] Among them, the multi-modal environmental perception data is multi-dimensional environmental and personnel status data collected through a distributed sensor network, covering the following types: Environmental spectral data, obtained by a spectral sensor, includes the wavelength distribution, illuminance, color temperature, etc. of visible light / infrared light (such as the spectral curve of the D65 standard light source); Personnel distribution data, obtained by a millimeter-wave radar and a thermal imaging camera, generates heat map data such as personnel position coordinates, activity trajectories, and aggregation density; Visual image data, collected by a high-definition camera, is used to analyze environmental textures (such as desktop reflection, screen glare) and personnel postures (such as reading and writing angles); Environmental parameter data, including temperature, humidity, CO2 concentration (affecting personnel fatigue), and natural light incident angle (obtained through a solar tracking sensor).

[0028] Among them, the user feedback data is subjective evaluation data input by the user through a terminal device (such as a mobile phone application, a wall touch panel), and can include: comfort score, satisfaction with the brightness, color temperature, and glare of the current lighting (1-5 points); mode switching instruction, manually triggering preset scenarios such as "meeting mode" and "reading mode"; personalized parameters, preference parameters defined by the user (such as the color temperature value preferred by a certain employee).

[0029] Among them, the blue light risk level is a blue light hazard index calculated based on the environmental spectral data, and is divided into three levels: low risk, medium risk, and high risk according to the IEC62471 standard. The blue light risk level can be obtained by calculating a weighted function through spectral data fitting.

[0030] Among them, the user's circadian rhythm data is biometric data on the human body's response to light obtained through physiological sensors, which may include: heart rate variability, collected by a photoplethysmogram sensor, reflecting the stress response of the autonomic nervous system to light; pupil diameter change, measured by an infrared pupil tracker, to evaluate visual fatigue (abnormal pupil constriction indicates glare or inappropriate illuminance); circadian rhythm phase, the biological clock state inferred from heart rate variability and body temperature data (for example, being in the waking period at 10 am requires high-color-temperature lighting).

[0031] In step S220, perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fused feature vectors in the enhanced data set.

[0032] Among them, spatio-temporal alignment processing is a processing flow for unifying the time and space coordinates of multi-source heterogeneous data, which may include three steps: timestamp alignment, adding precise timestamps (accuracy ≤ 1 ms) to sensor data, and eliminating abnormal data with a time deviation > 100 ms through a sliding window (such as a 500-ms window); spatial coordinate mapping, using the perspective transformation matrix of an edge computing node to convert different sensor coordinate systems (such as the pixel coordinates of a camera and the polar coordinates of a radar) into a unified Cartesian coordinate system (with the light source array as the origin and the XYZ axes corresponding to the three dimensions of space); time synchronization interpolation, performing cubic spline interpolation on sparse data (such as spectral data once a minute) and synchronizing it to a high-frequency timestamp (such as a 20-ms interval) to generate an aligned data set with equal time intervals.

[0033] Among them, environmental illuminance data is the light intensity of the target area measured by an illuminance sensor (unit: lux), which is divided into: natural illuminance, the outdoor incident light intensity obtained through a shutter / skylight sensor; artificial illuminance, the real-time output illuminance of the LED light source array, fed back by a distributed illuminometer.

[0034] Among them, the cross-modal generative adversarial network is a deep learning model, which includes two sub-networks: a generator, which inputs spectral data at low illuminance (such as the intensity of wavelengths from 400 to 700 nm) and a personnel distribution heat map, and outputs an enhanced visible light image (RGB three channels) and a high-density personnel distribution heat map; a discriminator, which distinguishes real data from generated data and improves the authenticity of the generated data through adversarial training.

[0035] Among them, data augmentation is a process of noise reduction, completion, and enhancement for low-quality raw data (such as low-illumination images, sparse personnel trajectories), which specifically includes: spectral data augmentation to generate a full-spectrum distribution curve and fill in the missing short-wave / long-wave signals under low illumination; personnel distribution augmentation to generate a continuous heat map from sparse positioning points (such as signals from 3 positioning devices) and mark the probability areas of personnel activities.

[0036] Among them, the fused feature vector is a multi-dimensional feature vector extracted from the augmented dataset by a feature extractor, which includes: equivalent illuminance feature, which is a comprehensive illuminance uniformity index integrating natural light and artificial light; personnel distribution feature, including personnel position coordinates, visual azimuth angle (the angle between the line of sight and the light source), and activity type (static office / dynamic walking); visual comfort feature, including the screen reflection index and text contrast ratio (black / white pixel brightness ratio) based on image analysis; environmental interference feature, including window reflection intensity and the brightness of electronic device screens (to avoid the superimposed glare of illumination and screen light).

[0037] In step S230, dynamic weighting is performed on the fused feature vector, and the initial lighting parameters are generated by combining the dynamically weighted feature vector and the user's physiological rhythm data. Then, the initial lighting parameters are energy-saving optimized through a pre-trained energy level model to obtain the target lighting parameters.

[0038] Among them, dynamic weighting is a mechanism for adjusting the weights of fused features in real time according to the feature change rate, which may include the following steps: calculating the feature change rate, such as the displacement speed of personnel coordinates and the change gradient of illuminance; weight assignment, assigning a high weight (such as 0.8) to features with high change rates (such as personnel movement) and a low weight (such as 0.2) to stable features (such as the background illuminance of a fixed workstation).

[0039] Among them, the energy level model is a mathematical model based on energy optimization, which regards the light source array as a quantum system, and each LED unit corresponds to an energy level. The goal is to minimize the ground state energy (total energy consumption) of the system while meeting the lighting requirements.

[0040] Among them, energy-saving optimization is a process of reducing energy consumption on the premise of ensuring lighting comfort, which is achieved through the energy level model: regional differential dimming, reducing the LED brightness in unoccupied areas to 10% and maintaining 80% in occupied areas; spectral energy distribution, preferentially activating high-efficiency LED chips when high color temperature is required; dynamic duty cycle adjustment, optimizing the PWM (pulse width modulation) parameters to reduce the current peak value without affecting vision (such as reducing the duty cycle from 90% to 75% at a frequency of 20 kHz).

[0041] In step S240, the target lighting parameters are mapped to pulse width modulation parameters, and the phase delay of the microlens array is determined according to the personnel distribution feature in the fused feature vector.

[0042] Among them, the mapping is to convert abstract lighting parameters (brightness coefficient, color temperature coefficient) into pulse width modulation signal parameters that can drive hardware, including: duty cycle, high-level time / cycle, which determines the average current of the LED (for example, a brightness coefficient of 0.8 corresponds to a 80% duty cycle); frequency, PWM frequency (such as 20 kHz to avoid human eye flicker perception), which needs to be optimized to reduce electromagnetic interference; phase, the phase difference of the PWM signals of multi-channel LEDs (such as the phase synchronization of red, green, and blue LEDs to avoid color distortion).

[0043] Among them, the personnel distribution characteristics are parameters in the fusion feature vector that characterize the personnel position and visual requirements, and can include two-dimensional coordinates, visual azimuth angle, and activity type label.

[0044] Among them, the micro-lens array phase delay is the phase modulation amount (unit: radian) of each lens unit in the micro-lens array. By electronically controlling the refractive index of the lens, the deflection of the light beam direction is achieved.

[0045] In step S250, drive the data of the constant current circuit according to the pulse width modulation parameters, perform micro-lens beam deflection according to the micro-lens array phase delay, and adjust the optical filter of the target LED light source array according to the blue light risk level, so as to complete the adaptive lighting control of the target LED light source array.

[0046] Among them, the constant current circuit is a drive circuit that provides a stable current for the LED, and adjusts the output current through PWM parameters, including: a feedback modulation module that real-time collects the LED voltage / current and adjusts the duty cycle through an algorithm to ensure the constant current accuracy (such as ±1%); multi-channel drive, which supports independent control of the current of each LED unit to achieve distributed dimming of the light source array.

[0047] Among them, the micro-lens beam deflection is to change the wavefront phase of the light emitted by the LED by adjusting the micro-lens phase delay, so as to achieve dynamic adjustment of the light beam direction. For example: when a person is located at a 30° azimuth on the right side of the light source array, a positive phase delay is applied to the right-side micro-lens to deflect the light beam 15° to the right and focus on the person's working surface; in a multi-person scenario, multiple focused light spots are generated (such as each work station corresponding to an independent light beam) to avoid energy consumption waste of global lighting.

[0048] Among them, the optical filter is a switchable light filtering device installed in front of the LED light source for adjusting the spectral output, including: a blue light filter, which inserts a 450 nm cut-off filter in case of high risk to block short-wave blue light (400 - 450 nm); an amber compensation channel, which activates an amber LED (peak wavelength 590 nm) in case of medium risk to neutralize the cold tone of blue light, improve visual comfort and reduce hazards at the same time; a dynamic diffraction grating, which adjusts the diffraction angle in real time according to spectral data to optimize color purity.

[0049] Exemplarily, the environment where the target LED light source array is located is a smart office. First, data collection is carried out: the ceiling millimeter-wave radar detects that 3 people are located at workstations A (x = 2m, y = 3m) and B (x = 5m, y = 4m), and 1 person is moving in the meeting room (identified by the thermal imaging camera); the spectral sensor detects that the current natural illuminance is 200 lux (lower than the preset threshold of 300 lux), and the blue light proportion is 35% (medium risk); the photoplethysmogram sensor and the infrared pupil tracker show that the heart rate variability of a certain employee has decreased (indicating fatigue), and the pupil diameter has decreased (possibly due to screen reflection and glare); the user feedbacks "serious screen reflection" through the mobile phone (user feedback data). The radar coordinates (polar coordinates) are converted into a Cartesian coordinate system, and the spectral data at 200 ms intervals are synchronized to 50 ms time stamps through cubic spline interpolation. Due to insufficient illuminance, the low-illuminance image is input into a cross-modal generative adversarial network to generate a clear image of the desktop reflection area (marking the keyboard area reflection value of 80%), and the personnel distribution heat map is enhanced (showing that two people are sitting still at workstation A and one person gets up frequently at workstation B). The equivalent illuminance, personnel azimuth, and visual comfort features are extracted. Since the personnel at workstation A are stationary (low change rate, weight 0.3) and the personnel at workstation B are moving frequently (high change rate, weight 0.7), the lighting requirements of workstation B are preferentially responded to. The initial parameters are a brightness coefficient of 0.8 (target illuminance 300 lux) and a color temperature of 4500K. Through the energy level model calculation, 20% of the LED units are turned off at workstation A, and a directional beam is turned on at workstation B, reducing the total energy consumption by 12%. The brightness coefficient of 0.8 is converted into an 80% duty cycle (frequency 25 kHz, phase synchronization to avoid flicker). According to the personnel azimuth of -30° at workstation B, the micro-lens phase delay matrix is calculated to deflect the beam 15° to the left and focus on the desktop keyboard area. Due to the medium risk of blue light, the amber LED channel is activated (outputting light with a wavelength of 590 nm), and the color temperature is adjusted to 4000K (reducing the blue light proportion), while maintaining the color rendering index Ra = 92.

[0050] By adopting the above embodiments provided in the present application, a three-dimensional data system is constructed by obtaining multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data; when the environmental illuminance is insufficient, cross-modal generative adversarial network data augmentation is used to fill in problems such as blurred visual images and sparse personnel distribution data in low-illuminance scenarios, and a more complete environmental representation is generated. Extract the fusion feature vector including equivalent illuminance, personnel distribution, visual comfort, and environmental interference to realize the comprehensive modeling of global elements such as environmental light quality, personnel activity patterns, and visual health risks, so that the control strategy covers multiple objectives such as spatial distribution, spectral safety, and physiological adaptation. Dynamic weighting can adjust the feature weights in real time, prioritize the response to key changes, and avoid the lag response of fixed weights to complex scenarios. The energy level model regards the light source array as an energy system, and by optimizing the light source matrix and ground state energy parameters, it balances the lighting requirements and energy consumption of each region globally, realizes the global energy efficiency optimization of light supply on demand, and avoids energy waste caused by local over-illumination. Calculate the phase delay of the microlens array in real time according to the personnel distribution characteristics, and realize the dynamic deflection of the light beam direction through the electronically controlled microlens, avoiding the lag of traditional mechanical adjustment. The real-time analysis of the blue light risk level triggers the dynamic switching of the optical filter, reducing the response time to sudden high-blue light scenarios. Dynamically identify the fatigue state or circadian rhythm phase through the user circadian rhythm data, and generate appropriate initial lighting parameters. The microlens beam deflection accurately adjusts the optical path according to the personnel distribution characteristics, avoids direct glare of the lamp, and improves the comfort of LED lighting control.

[0051] In an embodiment of the present application, the obtaining of the multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located includes: Obtain the multi-modal environmental perception data through a distributed sensor network pre-configured in the environment where the target LED light source array is located, where the multi-modal environmental perception data includes environmental spectral data, personnel distribution data, visual image data, and environmental parameter data; Obtain the user feedback data through a pre-configured user terminal; Perform blue light hazard analysis according to the environmental spectral data to obtain the blue light risk level; Obtain the heart rate change data and visual change data of the user through a pre-configured photoplethysmogram sensor and infrared pupil tracker, and generate the user circadian rhythm data according to the heart rate change data and visual change data.

[0052] Among them, the distributed sensor network is a wireless / wired network composed of multiple types of sensors deployed in the target area (such as offices, classrooms), with the capabilities of self-organizing network and distributed computing, and realizes the collaborative acquisition of multi-modal data. The sensor nodes can be spectral sensors, millimeter-wave radars, high-definition cameras, and environmental parameter sensors.

[0053] Among them, the environmental spectral data is the data of the light radiation wavelength distribution in the target area obtained by the spectral sensor; the personnel distribution data is the data characterizing the personnel position, movement trajectory, and aggregation state in the target area, obtained through multi-sensor fusion: the visual image data is the visual information of the target area collected by the high-definition camera and is used for analyzing visual comfort and environmental characteristics after preprocessing; the environmental parameter data is the other physical environmental data except the light environment, which affects the personnel comfort and lighting strategy.

[0054] Among them, the blue light hazard analysis is to evaluate the potential hazard of blue light to the human eye according to the environmental spectral data, and divide the risk levels according to the IEC62471 standard: low risk, no intervention required; medium risk, activate the amber compensation channel; high risk, insert a blue light filter and limit the brightness ≤ 300 lux.

[0055] For example, taking the environment where the target LED light source array is located as a smart classroom, 5 millimeter-wave radar nodes (covering the entire classroom), 2 spectral sensors (1 each at the podium / back row), and 3 high-definition cameras (monitoring students' posture from multiple angles) are deployed on the ceiling. Students wear bracelets with integrated sensors, and infrared pupil trackers are installed on desks (hidden above the monitor). Environmental parameter sensors are distributed in the four corners of the classroom to collect temperature, humidity, and CO2 concentration in real time. The spectral sensor on the podium detects that blue light accounts for 32% (medium risk) and the color temperature is 5500K (high, which may cause student fatigue). The millimeter-wave radar shows that among the 50 students, 10 are interacting in front of the podium and 40 are sitting at the desks. The camera recognizes that 30% of the students are hunched over (poor sitting posture) and 15% of the students rub their eyes frequently (possibly due to glare). The photoelectric volumetric pulse wave sensor detects that the average heart rate change data has dropped by 20% (compared to the baseline before class), and pupil tracking shows that the average pupil diameter is 2.2mm (2.8mm lower than the baseline, indicating visual fatigue). If the blue light analysis is medium risk, the amber LED channel compensation can be triggered to reduce the color temperature to 4500K and the proportion of blue light to 25%. Through Fourier analysis of heart rate change data, it is determined that the current fatigue period is at 2 pm (rhythm phase 14h), and the fatigue index FI=0.7 (high fatigue), which indicates that the illumination uniformity needs to be improved and blue light stimulation needs to be reduced. Combined with the distribution of personnel (students in front of the podium need to be illuminated), visual images (increase vertical illumination in the area of students with poor sitting posture), and physiological data (reduce blue light during fatigue), the target lighting parameters are generated: the illumination of the podium area is 500lux (color temperature 4500K), the illumination of the desk area is 400lux (color temperature 4000K), and the microlens beam is deflected 10° toward the student's face (to reduce screen reflection).

[0056] In this embodiment, five types of data are collected through a distributed network to avoid the one-sidedness of the control strategy (such as turning on all the lights only because of the presence of personnel, ignoring the fatigue caused by spectral discomfort). Distributed sensor nodes realize grid-level data collection, support regional differentiated control (such as high illumination in the front row of the classroom and low illumination in the back row), and improve the uniformity of lighting. Based on the blue light hazard analysis algorithm of the IEC standard, real-time graded response (medium risk compensation, high risk blocking) is implemented to reduce the frequency of high-risk scenes and reduce the measured retinal blue light exposure. Combined with the environmental spectrum and physiological data, the filter device (such as amber channel / blue light filter) is automatically switched. In the computer room and other blue light high emission scenes, the color rendering index is maintained while the proportion of blue light energy is controlled within a certain range. Through the capacitive product pulse wave sensor and infrared pupil tracker, the accuracy of detecting visual fatigue is improved, and the "relaxation mode between classes" can be automatically triggered after 45 minutes of continuous eye use. According to the circadian rhythm phase analyzed by the heart rate change data, the lighting parameters are automatically adjusted to improve the subjective comfort score of personnel compared with the fixed mode.

[0057] In one embodiment of the present application, for the spatio-temporal alignment processing of the multi-modal environmental perception data and user feedback data to obtain an aligned data set, when the environmental illumination data in the aligned data set is lower than a preset illumination threshold, the aligned data set is transmitted to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and the fused feature vectors in the enhanced data set are extracted, including: Perform timestamp alignment processing on the multi-modal environmental perception data and user feedback data to obtain an initial aligned data set; Map the initial aligned data set to a unified coordinate system through a pre-configured edge computing node and a perspective transformation matrix; Adopt a sliding window to perform cubic spline interpolation on the data in the unified coordinate system and synchronize it to a unified timestamp to obtain a target aligned data set; When the environmental illumination data in the target aligned data set is lower than the preset illumination threshold, transmit the environmental spectral data and personnel distribution data to the visible light conversion channel and the heat map conversion channel in the cross-modal generative adversarial network to obtain enhanced visible light image data and enhanced personnel distribution heat map data; Transmit the enhanced data set to a pre-configured feature extractor to obtain equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features, and fuse the equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features to obtain the fused feature vectors.

[0058] Among them, the timestamp alignment processing is a process of uniformly calibrating the timestamps of multi-source sensor data to solve the differences in sampling frequencies / delays of different sensors. The edge computing node is a low-latency computing device deployed in the target area, responsible for local data processing and partial model inference, reducing the dependence on the cloud.

[0059] Among them, the perspective transformation matrix is a 3×3 projection matrix used to convert different sensor coordinate systems into a unified Cartesian coordinate system. At least 4 groups of corresponding points (such as fixed marks at the corners of the classroom podium and door corners) are collected through the checkerboard calibration method (camera) or known fiducial points (radar), and the matrix parameters are solved.

[0060] Among them, mapping the initial aligned data set to a unified coordinate system is a process of converting the spatial coordinates of multi-source data into the same reference system using the perspective transformation matrix. The sliding window is a dynamic data interception window for time synchronization. For example, the window size is T (such as 500 ms) and the step size is 100 ms, which can ensure data continuity.

[0061] Among them, cubic spline interpolation is a method for continuous fitting of sparse data (such as CO2 concentration data once per minute) within a sliding window, approximating the original data points through a piecewise cubic polynomial function \(S(t)\). Compared with linear interpolation, cubic spline interpolation preserves the curvature characteristics of the data (such as the process of gradual change in illuminance) and reduces the interpolation error.

[0062] Among them, the visible light conversion channel is a sub-network in the cross-modal generative adversarial network responsible for converting low-illuminance spectral data into enhanced visible light images; the heat map conversion channel is a sub-network in the cross-modal generative adversarial network responsible for converting sparse personnel distribution data into high-density heat maps. Among them, the enhanced visible light image data is high-definition and color-accurate image data output by the visible light conversion channel, which is used for visual comfort analysis. The enhanced personnel distribution heat map data is a continuous probability matrix output by the heat map conversion channel, representing the probability distribution of the presence of personnel in the area.

[0063] Among them, the feature extractor is a multi-modal feature extraction model based on deep learning. It inputs an enhanced data set (visible light images, heat maps, spectral data, etc.) and outputs a multi-dimensional feature vector.

[0064] In this embodiment, through timestamp remapping, the time deviation of multi-sensor data is reduced to ensure the time consistency of the same event in different data. After the perspective transformation matrix is calibrated, the coordinate error of multi-sensors is reduced, supporting centimeter-level area regulation (such as the beam accurately covering the personnel seats and avoiding irradiating unoccupied areas). After the cross-modal generative adversarial network is enhanced, the signal-to-noise ratio of the visible light image is improved, and the accuracy of personnel recognition is improved; the probability error of the heat map is reduced, and the accuracy of detecting personnel aggregation areas is improved. The time resolution of sparse data is improved by enhancing cubic spline interpolation, avoiding regulation lag caused by data loss.

[0065] In an embodiment of the present application, dynamically weighting the fusion feature vector, generating an initial lighting parameter by combining the dynamically weighted feature vector and the user's physiological rhythm data, and performing energy-saving optimization on the initial lighting parameter through a pre-trained energy level model to obtain a target lighting parameter, including: According to the fusion feature vector, determining the feature change rate, using the feature change rate as the importance, dynamically weighting the fusion feature vector to obtain the dynamically weighted feature vector; According to the user's physiological rhythm data, determining the target lighting mode, and performing constraint optimization by combining the dynamically weighted feature vector and the target lighting mode to obtain the initial lighting parameter, where the initial lighting parameter includes a lamp brightness coefficient, a lamp color temperature coefficient, and a lamp dimming gain coefficient; Transmit the initial lighting parameters to the pre-trained energy level model to generate a light source array matrix and ground state energy parameters, and perform energy-saving optimization on the lamp brightness coefficients in the initial lighting parameters according to the light source array matrix and the ground state energy parameters to obtain the target lighting parameters.

[0066] Among them, the feature change rate is the degree of change of each dimension feature in the fused feature vector over time series, reflecting the dynamics and importance of the feature. The target lighting mode is a differentiated lighting strategy determined according to user physiological rhythm data (such as fatigue index, circadian phase), and can include a predefined mode set: wake-up mode (7-12 am), high color temperature (5000-6500K), high illuminance (400-500 lux) to enhance attention; fatigue relief mode (2-4 pm), medium color temperature (4000-4500K), medium illuminance (300-400 lux) to reduce blue light stimulation; relaxation mode (after 18:00 at night), low color temperature (2700-3500K), low illuminance (150-200 lux) to promote melatonin secretion.

[0067] Among them, constraint optimization is to solve the optimal initial parameters that meet the lighting requirements and physical limitations by combining the dynamic weighted feature vector under the target lighting mode. The constraint conditions can be: illuminance constraint, target area illuminance ≥ 300 lux; blue light constraint, blue light risk level ≤ medium risk; hardware constraint, LED brightness coefficient ∈ [0, 1], color temperature coefficient ∈ [2700K, 6500K].

[0068] Among them, the light source array matrix represents the working state of each unit in the target LED light source array, and the matrix elements directly control the PWM duty cycle of the LED drive circuit, supporting regional dimming. The ground state energy parameter is an energy consumption optimization parameter based on the quantum mechanics model. Regarding the light source array as a quantum system, the ground state energy represents the lowest energy consumption state of the system.

[0069] In this embodiment, through feature change rate weighting, the system improves the response speed to key factors with rapid changes (such as a sudden increase in the fatigue index, rapid movement of personnel), and reduces the response time. In the scenario where a person suddenly gets up and leaves the work station, dynamic weighting can quickly detect a sudden increase in the coordinate change rate, immediately reduce the brightness coefficient of that area, and improve the energy-saving efficiency. Dynamically adjusting the lighting mode according to the ground state energy parameters shortens the phase error between the artificial light cycle and the human biological clock phase. Long-term use can reduce the health risks caused by circadian rhythm disorders.

[0070] In an embodiment of the present application, the mapping of the target lighting parameters to pulse width modulation parameters and determining the micro-lens array phase delay according to the personnel distribution characteristics in the fused feature vector includes: Determine the duty cycle, frequency, and phase corresponding to pulse width modulation according to the target lighting parameters, optimize the frequency, and compensate the phase to obtain the pulse width modulation parameters; Determine the visual azimuth angle and elevation angle of the personnel according to the personnel distribution characteristics, and generate the phase delay of the microlens array according to the visual azimuth angle and elevation angle of the personnel.

[0071] Among them, determining the PWM parameters according to the target lighting parameters controls the average current of the LED through a high-frequency switching signal to achieve brightness / color temperature adjustment. When determining the PWM parameters according to the target lighting parameters, the brightness coefficient B and color temperature coefficient C can be extracted from the target lighting parameters and converted into the duty cycle of each LED channel. The frequency is initially set according to hardware characteristics (such as the optimal operating frequency of the LED drive circuit) and electromagnetic compatibility (EMC) requirements. Calculate the phase difference of multiple channels (such as the RGB channels have a unified phase of 0° to avoid color fluctuations caused by timing misalignment).

[0072] Among them, the azimuth angle is the angle between the projection of the personnel's line of sight on the horizontal plane and the normal line of the light source array (the vertical downward direction) (range: 0° - 360°, positive clockwise). The elevation angle is the angle between the personnel's line of sight and the horizontal plane (positive upward, range: -90° - 90°), which reflects whether the line of sight is upward (such as looking at the ceiling) or downward (such as looking at the desktop).

[0073] Among them, generating the phase delay of the microlens array is composed of N×M independently electronically controllable liquid crystal microlenses. Each lens adjusts the phase delay through voltage, changes the wavefront direction of the outgoing light, and realizes beam deflection.

[0074] In this embodiment, through the precise mapping of PWM parameters and the dynamic beam control of microlenses, an efficient conversion from abstract lighting parameters to physical execution is achieved. This solution provides technical support for on-demand light distribution in scenarios with high visual quality requirements such as reading and office through orientation control driven by visual angles.

[0075] In an embodiment of the present application, driving the data of the constant current circuit according to the pulse width modulation parameters, deflecting the microlens beam according to the phase delay of the microlens array, and adjusting the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array, including: According to the pulse width modulation parameters, perform feedback modulation on the voltage and current of the constant current circuit and output an adjustable constant current source; According to the phase delay of the microlens array, determine the beam calibration angle, and based on the beam calibration angle, deflect the microlens beam; When the blue light risk level is medium risk, activate the pre-configured amber LED channel, or when the blue light risk level is high risk, insert a blue light filter and perform brightness limiting.

[0076] Among them, the beam calibration angle is the deviation correction amount between the actual deflection angle of the microlens and the theoretically calculated value. The amber LED channel is an integrated amber LED with a peak wavelength of 590 nm. The blue light filter is made of borosilicate glass as the substrate, with more than 100 dielectric films evaporated on the surface and a cut-off wavelength of 450 nm.

[0077] Among them, brightness limiting is to set a brightness upper limit in a high-risk scenario to avoid a sudden drop in illuminance after inserting a blue light filter.

[0078] In this embodiment, through the hardware cooperation mechanism of precise control by constant current drive, dynamic adjustment of beam calibration, and response to blue light risk grading, the efficient conversion and safe control from electrical signals to optical signals are achieved, the order of magnitude of current stability is improved, the engineering realization of beam pointing accuracy is achieved, and the active grading prevention and control of blue light hazards are realized, providing a stable, precise, and safe lighting solution for high-precision and high-reliability scenarios.

[0079] In an embodiment of the present application, the method further includes: Obtain the comfort score of the user for the target LED light source array after adaptive lighting regulation, and the actual energy consumption data corresponding to the target LED light source array after adaptive lighting regulation; Update the target lighting parameters according to the comfort score and the actual energy consumption data.

[0080] Among them, the comfort score is the subjective evaluation data of the user on the result of adaptive lighting regulation, collected through preset multi-dimensional indicators. The actual energy consumption data is the real-time energy consumption parameter of the light source array after adaptive regulation, collected in real time through a hardware sensor.

[0081] In this embodiment, through the closed-loop feedback mechanism of comfort score, energy consumption data, and parameter update, self-learning and dynamic optimization are carried out, realizing the explicit utilization of subjective experience, the continuous optimization of energy efficiency performance, and the long-term adaptation to complex scenarios, breaking through the limitations of traditional intelligent lighting preset strategies and passive execution, establishing a complete intelligent closed-loop, and realizing the personalized, efficient, and sustainable operation of the lighting system.

[0082] In an embodiment of the present application, an adaptive LED lighting regulation device based on multi-modal environment perception is also provided. Figure 3 It is a schematic diagram of an adaptive LED lighting regulation device based on multi-modal environment perception shown in an exemplary embodiment of the present application. Refer to Figure 3 This device includes: A data acquisition module 301, configured to acquire multi-modal environmental perception data, user feedback data, blue light risk levels, and user circadian rhythm data in the environment where the target LED light source array is located; A feature fusion module 302, configured to perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract fusion feature vectors from the enhanced data set; An illumination parameter generation module 303, configured to perform dynamic weighting on the fusion feature vectors, generate initial illumination parameters by combining the dynamically weighted feature vectors and the user circadian rhythm data, and perform energy-saving optimization on the initial illumination parameters through a pre-trained energy level model to obtain target illumination parameters; A parameter mapping module 304, configured to map the target illumination parameters to pulse width modulation parameters, and determine the phase delay of the microlens array according to the personnel distribution characteristics in the fusion feature vectors; A regulation module 305, configured to drive the data of a constant current circuit according to the pulse width modulation parameters, perform microlens beam deflection according to the phase delay of the microlens array, and adjust the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive illumination regulation of the target LED light source array.

[0083] In the embodiment of the present application, the adaptive LED lighting control device based on multi-modal environmental perception constructs a three-dimensional data system by obtaining multi-modal environmental perception data, user feedback data, blue light risk levels, and user physiological rhythm data. When the environmental illuminance is insufficient, cross-modal generative adversarial network data augmentation is used to fill in problems such as blurred visual images and sparse personnel distribution data in low-illuminance scenarios, generating a more complete environmental representation. Extracting a fusion feature vector containing equivalent illuminance, personnel distribution, visual comfort, and environmental interference enables comprehensive modeling of global elements such as environmental light quality, personnel activity patterns, and visual health risks, making the control strategy cover multiple objectives such as spatial distribution, spectral safety, and physiological adaptation. Dynamic weighting can adjust the feature weights in real time, giving priority to responding to key changes and avoiding the lag response of fixed weights to complex scenarios. The energy level model regards the light source array as an energy system. By optimizing the light source matrix and ground state energy parameters, it balances the lighting requirements and energy consumption of each region globally, achieving global energy efficiency optimization of light supply on demand and avoiding energy waste caused by local over-illumination. According to the personnel distribution characteristics, the phase delay of the microlens array is calculated in real time, and the dynamic deflection of the light beam direction is realized through the electro-controlled microlens, avoiding the lag of traditional mechanical adjustment. The real-time analysis of the blue light risk level triggers the dynamic switching of the optical filter, reducing the response time to sudden high-blue light scenarios. The fatigue state or circadian rhythm phase is dynamically identified through user physiological rhythm data, generating appropriate initial lighting parameters. The microlens beam deflection accurately adjusts the optical path according to the personnel distribution characteristics, avoiding direct glare from the lamps and improving the comfort of LED lighting control.

[0084] For the specific embodiments of the adaptive LED lighting control device based on multi-modal environmental perception in the present application, reference can be made to the examples shown in the above-mentioned adaptive LED lighting control method based on multi-modal environmental perception, and details are not described herein again.

[0085] In an embodiment of the present application, an electronic device for implementing the above-mentioned adaptive LED lighting control method based on multi-modal environmental perception is also provided. The electronic device includes a memory and a processor. A computer program is stored in the memory, and the processor is configured to execute the above-mentioned adaptive LED lighting control method based on multi-modal environmental perception through the computer program.

[0086] See Figure 4 , Figure 4It is a schematic structural diagram of an electronic device shown in an exemplary embodiment of the present application. The computer system 400 includes a Central Processing Unit (CPU) 401, which can perform various appropriate actions and processes according to a program stored in a Read-Only Memory (ROM) 402 or a program loaded from a storage section 408 into a Random Access Memory (RAM) 403, such as executing the method described in the above embodiment. In the RAM 403, various programs and data required for system operation are also stored. The CPU 401, ROM 402, and RAM 403 are connected to each other via a bus 404. An Input / Output (I / O) interface 405 is also connected to the bus 404.

[0087] The following components are connected to the I / O interface 405: an input section 406 including, such as, a keyboard, a mouse, etc.; an output section 407 including, such as, a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc. and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as required. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as required so that a computer program read from it can be installed into the storage section 408 as required.

[0088] Particularly, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 409, and / or installed from the removable medium 411. When the computer program is executed by a Central Processing Unit (CPU) 401, various functions defined in the system of the present application are executed.

[0089] It should be noted that the computer-readable medium shown in the embodiments of the present application may be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0090] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. Among them, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the above module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0091] The units involved in the embodiments described in this application can be implemented in software or in hardware, and the described units can also be provided in a processor. Among them, the names of these units do not, in some cases, constitute a limitation on the units themselves.

[0092] On the other hand, this application also provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program. When the computer program runs, it executes the above-mentioned adaptive LED lighting control method based on multi-modal environment perception. The computer-readable storage medium can be included in the electronic device described in the above embodiments, or can exist alone without being assembled into the electronic device.

[0093] On the other hand, this application also provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the adaptive LED lighting control method based on multi-modal environment perception provided in the above various embodiments.

[0094] The above embodiments are only used to exemplarily illustrate the principles and effects of the present invention, rather than to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes completed by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. An adaptive LED lighting control method based on multi-modal environmental perception, characterized in that, The method includes: Obtaining multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located; Performing spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmitting the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extracting the fused feature vectors in the enhanced data set; Dynamically weighting the fused feature vectors, generating initial lighting parameters by combining the dynamically weighted feature vectors and the user circadian rhythm data, and performing energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; Mapping the target lighting parameters to pulse width modulation parameters, and determining the phase delay of the microlens array according to the personnel distribution characteristics in the fused feature vectors; Driving the data of the constant current circuit according to the pulse width modulation parameters, deflecting the microlens beam according to the phase delay of the microlens array, and adjusting the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array.

2. The adaptive LED lighting control method based on multi-modal environmental perception according to claim 1, wherein The obtaining of the multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located includes: Obtaining the multi-modal environmental perception data through a distributed sensor network pre-configured in the environment where the target LED light source array is located, where the multi-modal environmental perception data includes environmental spectrum data, personnel distribution data, visual image data, and environmental parameter data; Obtaining the user feedback data through a pre-configured user terminal; Performing blue light hazard analysis according to the environmental spectrum data to obtain the blue light risk level; Obtaining the heart rate change data and visual change data of the user through a pre-configured photoplethysmogram sensor and an infrared pupil tracker, and generating the user circadian rhythm data according to the heart rate change data and visual change data.

3. The adaptive LED lighting control method based on multi-modal environmental perception according to claim 2, wherein, The performing of spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmitting the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extracting the fused feature vectors in the enhanced data set includes: Performing timestamp alignment processing on the multi-modal environmental perception data and user feedback data to obtain an initial aligned data set; Mapping the initial aligned data set to a unified coordinate system through a pre-configured edge computing node and a perspective transformation matrix; Using a sliding window to perform cubic spline interpolation on the data in the unified coordinate system and synchronize it to a unified timestamp to obtain a target aligned data set; When the ambient illumination data in the target alignment dataset is lower than the preset illumination threshold, the ambient spectral data and the personnel distribution data are transmitted to the visible light conversion channel and the heat map conversion channel in the cross-modal generative adversarial network to obtain enhanced visible light image data and enhanced personnel distribution heat map data; The enhanced dataset is transmitted to a pre-configured feature extractor to obtain equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features, and the equivalent illumination features, personnel distribution features, visual comfort features, and environmental interference features are fused to obtain the fused feature vector.

4. The adaptive LED lighting control method based on multi-modal environment perception according to claim 1, wherein, The dynamic weighting of the fused feature vector, the generation of initial lighting parameters by combining the dynamically weighted feature vector and the user's physiological rhythm data, and the energy-saving optimization of the initial lighting parameters by a pre-trained energy level model to obtain target lighting parameters include: According to the fused feature vector, determining the feature change rate, using the feature change rate as the importance, and dynamically weighting the fused feature vector to obtain the dynamically weighted feature vector; According to the user's physiological rhythm data, determining the target lighting mode, and performing constraint optimization by combining the dynamically weighted feature vector and the target lighting mode to obtain the initial lighting parameters, where the initial lighting parameters include a lamp brightness coefficient, a lamp color temperature coefficient, and a lamp dimming gain coefficient; The initial lighting parameters are transmitted to the pre-trained energy level model to generate a light source array matrix and a ground state energy parameter, and the lamp brightness coefficient in the initial lighting parameters is energy-saving optimized according to the light source array matrix and the ground state energy parameter to obtain the target lighting parameters.

5. The adaptive LED lighting control method based on multi-modal environmental perception according to claim 1, characterized in that, The mapping of the target lighting parameters to pulse width modulation parameters and the determination of the micro-lens array phase delay according to the personnel distribution features in the fused feature vector include: According to the target lighting parameters, determining the duty cycle, frequency, and phase corresponding to the pulse width modulation, optimizing the frequency, and compensating the phase to obtain the pulse width modulation parameters; According to the personnel distribution features, determining the personnel visual azimuth angle and elevation angle, and generating the micro-lens array phase delay according to the personnel visual azimuth angle and elevation angle.

6. The adaptive LED lighting control method based on multi-modal environment perception according to claim 1, wherein Driving the data of the constant current circuit according to the pulse width modulation parameters, deflecting the micro-lens beam according to the micro-lens array phase delay, and adjusting the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting control of the target LED light source array, including: According to the pulse width modulation parameters, performing feedback modulation on the voltage and current of the constant current circuit to output an adjustable constant current source; According to the micro-lens array phase delay, determining the beam calibration angle, and deflecting the micro-lens beam based on the beam calibration angle; When the blue light risk level is medium risk, activating a pre-configured amber LED channel, or when the blue light risk level is high risk, inserting a blue light filter and performing brightness limiting.

7. The adaptive LED lighting control method based on multi-modal environment perception according to claim 1, characterized in that, The method further includes: Obtain the comfort score of the user for the target LED light source array after adaptive lighting regulation, and the actual energy consumption data corresponding to the target LED light source array after adaptive lighting regulation; Update the target lighting parameters according to the comfort score and the actual energy consumption data.

8. An adaptive LED lighting control device based on multi-modal environmental perception, characterized in that, The device includes: A data acquisition module, configured to acquire multi-modal environmental perception data, user feedback data, blue light risk level, and user circadian rhythm data in the environment where the target LED light source array is located; A feature fusion module, configured to perform spatio-temporal alignment processing on the multi-modal environmental perception data and user feedback data to obtain an aligned data set. When the environmental illuminance data in the aligned data set is lower than a preset illuminance threshold, transmit the aligned data set to a pre-constructed cross-modal generative adversarial network for data enhancement to obtain an enhanced data set, and extract the fused feature vectors in the enhanced data set; A lighting parameter generation module, configured to perform dynamic weighting on the fused feature vectors, generate initial lighting parameters by combining the dynamically weighted feature vectors and the user circadian rhythm data, and perform energy-saving optimization on the initial lighting parameters through a pre-trained energy level model to obtain target lighting parameters; A parameter mapping module, configured to map the target lighting parameters to pulse width modulation parameters, and determine the phase delay of the microlens array according to the personnel distribution characteristics in the fused feature vectors; A regulation module, configured to drive the data of a constant current circuit according to the pulse width modulation parameters, perform microlens beam deflection according to the phase delay of the microlens array, and adjust the optical filter of the target LED light source array according to the blue light risk level to complete the adaptive lighting regulation of the target LED light source array.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein the computer program, when running, executes the multi-modal environment perception-based adaptive LED lighting regulation method according to any one of claims 1 to 7.

10. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to execute the multi-modal environment perception-based adaptive LED lighting regulation method according to any one of claims 1 to 7 through the computer program.

Images