Intelligent lighting control method and system

Abstract

The invention discloses an intelligent illumination control method and system, and relates to the technical field of intelligent illumination control, and the method comprises the steps: starting three-dimensional space scanning through deploying a millimeter wave radar array at the top of a classroom, obtaining a coordinate mapping table, scanning an initial light environment through a spectrum radiometer, measuring the spectrum power distribution of a wave band, and obtaining a reference illumination parameter matrix; the wave radar array emits electromagnetic waves and receives reflection signals according to each monitoring area in the coordinate mapping table, extracts power spectrum density of beta waves and alpha waves, calculates a cognitive activity index and forms a dimensional feature vector; and according to the dimensional feature vector, through a nerve-spectrum mapping model, calculating a target color temperature adjustment amount, monitoring out-of-window wave band intensity by an ultraviolet sensor, and controlling a bias voltage of a quantum dot layer to adjust a light-emitting peak wavelength so as to obtain a spectrum parameter. The light-emitting peak wavelength is dynamically adjusted through the ultraviolet response type quantum dot bias, spectral parameter closed-loop control is formed, and multi-dimensional collaborative optimization of spectral characteristics is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent lighting control, and in particular to an intelligent lighting control method and system. Background Art

[0002] The application of intelligent lighting control technology in educational scenarios has become a research hotspot in recent years. Existing technologies mainly adopt lighting control strategies based on ambient light sensors and preset scene modes. Ambient brightness data is collected through light sensors, and the output of LED light sources is adjusted by combining a preset color temperature-illuminance curve. Some advanced systems introduce human body infrared sensing technology to achieve regional control, or match the needs of different teaching periods through timing strategies. In terms of spectral regulation, conventional methods usually use phosphor materials or filters with fixed formulations to achieve specific color temperature output. Some research-level systems achieve spectral regulation within a limited range through PWM dimming technology. These technologies provide basic solutions for educational lighting, can meet the basic functional requirements of classroom lighting, and have achieved certain results in terms of energy efficiency and visual comfort.

[0003] There is room for improvement in existing methods to achieve precise biological rhythm adaptation and dynamic environmental adaptation. The control strategy based on preset scene modes is difficult to respond to changes in students' cognitive states in real time, and it is impossible to establish a dynamic mapping relationship between lighting parameters and learners' physiological indicators. Traditional spectral regulation technologies are limited by the characteristics of fixed materials and are difficult to maintain precise rhythm stimulation effects under complex ambient light interference. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent lighting control method to solve the problems that it is difficult for existing technologies to achieve dynamic adaptation of cognitive state-lighting parameters and precise rhythm regulation under ambient light interference.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an intelligent lighting control method, which includes starting three-dimensional space scanning through a millimeter-wave radar array deployed on the top of a classroom to obtain a coordinate mapping table, and scanning the initial light environment with a spectroradiometer to measure the spectral power distribution of the measurement band to obtain a reference lighting parameter matrix; The wave radar array emits electromagnetic waves to each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral densities of β waves and α waves, calculates the cognitive activity index, and forms a three-dimensional feature vector; According to the three-dimensional feature vector, through a neural-spectral mapping model, calculate the target color temperature adjustment amount, and the ultraviolet sensor monitors the intensity of the band outside the window to control the bias voltage of the quantum dot layer to adjust the emission peak wavelength to obtain spectral parameters; Using a reinforcement learning algorithm based on spectral parameters, a reference illumination parameter matrix, and a three-dimensional feature vector, a control policy table is obtained. Monitor the millimeter-wave radar array, the spectral radiometer, and the quantum dot layer, and start the self-calibration program to obtain a repair log.

[0008] As a preferred embodiment of the intelligent lighting control method of the present invention, wherein: starting a three-dimensional space scan through a millimeter-wave radar array deployed on the top of the classroom to obtain a coordinate mapping table, including the following steps, The millimeter-wave radar module deployed on the top of the classroom performs a power-on self-check, forms a hardware initialization, and establishes a communication link. A frequency-modulated continuous wave signal is transmitted through the radar array to cover the monitoring area; Based on the electromagnetic wave signal reflected by the millimeter-wave radar, an original data stream is obtained; Perform a fast Fourier transform to generate a three-dimensional point cloud data set of distance-angle-velocity information. Identify the seat area through the DBSCAN clustering algorithm, generate the center coordinates of each seat area, and form a coordinate mapping table.

[0009] As a preferred embodiment of the intelligent lighting control method of the present invention, wherein: the spectral radiometer scans the initial light environment, measures the spectral power distribution of the band, and obtains a reference illumination parameter matrix, including the following steps, Capture the ambient light signal of the band through the spectral radiometer to obtain the original ADC readings; Deduct the background noise from the original ADC data to obtain the net light signal intensity, integrate the net light signal intensity matching function, and obtain the spectral power distribution curve; Based on the spectral power distribution, calculate the illuminance, correlated color temperature, color rendering index, and melatonin equivalent illuminance intensity to form a reference illumination parameter matrix.

[0010] As a preferred embodiment of the intelligent lighting control method of the present invention, wherein: the wave radar array emits electromagnetic waves according to each monitoring area of the coordinate mapping table, receives the reflected signal, extracts the power spectral density of the β wave and the α wave, calculates the cognitive activity index, and forms a three-dimensional feature vector, including the following steps, Read the coordinate mapping table through the meter-wave radar array to obtain the regional center coordinates and the coverage radius; Generate a directional beam based on the regional center coordinates and emit electromagnetic waves to obtain amplitude information, perform an STFT transform, and generate a three-dimensional spectrogram; Separate the power spectral density of the β wave and the α wave from the spectrogram and calculate the band power spectral density; Based on the calculation result of the band power spectral density, obtain the cognitive activity index; Process the cognitive activity index to obtain a three-dimensional feature vector.

[0011] As a preferred embodiment of the intelligent lighting control method of the present invention, the following steps are included: calculating the target color temperature adjustment amount through a neural-spectral mapping model based on the 3D feature vector, monitoring the intensity of the external window band by an ultraviolet sensor, controlling the bias voltage of the quantum dot layer to adjust the emission peak wavelength, and obtaining spectral parameters, Constructing a three-layer fully connected neural network based on the 3D feature vector, using the ReLU activation function and Dropout0.2 regularization, combining the mean square error loss with the circadian rhythm constraint term, and adding regularization to prevent overfitting; Training the three-layer fully connected neural network using the Adam optimizer to obtain the neural-spectral mapping model; Inputting the 3D feature vector into the neural-spectral mapping model to obtain the target color temperature adjustment amount; Monitoring the intensity of the external window band by the ultraviolet sensor and setting a band intensity threshold; When the band intensity is greater than the band intensity threshold, the control circuit applies a DC bias voltage to the quantum dot layer to control the emission peak wavelength; Combining the emission peak wavelength, the melatonin equivalent illuminance, and the target color temperature adjustment amount to obtain spectral parameters.

[0012] As a preferred embodiment of the intelligent lighting control method of the present invention, the following steps are included: obtaining a control strategy table using a reinforcement learning algorithm based on the spectral parameters, the reference lighting parameter matrix, and the 3D feature vector, Combining the spectral parameters, the reference lighting parameter matrix, and the 3D feature vector into a cognitive-spectral state vector; Constructing an Actor-Critic network architecture based on the cognitive-spectral state vector to obtain network parameters; Updating the network parameters using the PPO algorithm, mapping the updated network parameters into a JSON format control instruction table to form a control strategy table.

[0013] As a preferred embodiment of the intelligent lighting control method of the present invention, the following steps are included: monitoring the millimeter wave radar array, the spectral radiometer, and the quantum dot layer, and starting a self-calibration program to obtain a repair log, Generating a device state vector by monitoring the millimeter wave radar array, the spectral radiometer, and the quantum dot layer; Setting a device state vector threshold based on the device state vector, and triggering an exception flag when the device state vector is greater than the device state vector threshold; Executing a selected calibration protocol according to the exception flag, and calculating equivalent circuit parameters by fitting the Nyquist curve; Based on the equivalent circuit parameters, starting a self-calibration program to obtain a repair log.

[0014] In a second aspect, the present invention provides an intelligent lighting control system, including a reference light parameter matrix module, which starts three-dimensional space scanning through a millimeter-wave radar array deployed on the top of the classroom to obtain a coordinate mapping table, and a spectroradiometer scans the initial light environment to measure the spectral power distribution of the measurement band, thereby obtaining a reference light parameter matrix; A dimensional feature vector module, where the millimeter-wave radar array emits electromagnetic waves according to the coordinate mapping table for each monitoring area, receives the reflected signals, extracts the power spectral densities of β waves and α waves, calculates the cognitive activity index, and forms a dimensional feature vector; A spectral parameter module, which calculates the target color temperature adjustment amount through a neural-spectral mapping model according to the dimensional feature vector, and a ultraviolet sensor monitors the intensity of the external window band, controls the bias voltage of the quantum dot layer to adjust the emission peak wavelength, thereby obtaining spectral parameters; A repair log module, which uses a reinforcement learning algorithm according to the spectral parameters, the reference light parameter matrix, and the dimensional feature vector to obtain a control strategy table, monitors the millimeter-wave radar array, the spectroradiometer, and the quantum dot layer, and starts a self-calibration program to obtain a repair log.

[0015] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the intelligent lighting control method described in the first aspect of the present invention is implemented.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the intelligent lighting control method described in the first aspect of the present invention is implemented.

[0017] The beneficial effects of the present invention are as follows: Through non-contact millimeter-wave detection technology, a dimensional feature vector reflecting the cognitive state of students is dynamically generated based on the power spectral densities of β / α waves, realizing the direct coupling of physiological signals and lighting control, enabling the lighting adjustment to shift from passive environmental adaptation to active neural feedback regulation, achieving the beneficial effect of precisely matching the brain wave rhythm of learners with the light environment. By adopting a three-layer neural network model integrating biological rhythm constraints, the cognitive feature vector is mapped to the target color temperature adjustment amount, and the emission peak wavelength is dynamically adjusted through the ultraviolet-responsive quantum dot bias voltage to form a closed-loop control of spectral parameters, realizing the multi-dimensional collaborative optimization of spectral characteristics, and achieving the beneficial effect of maintaining the best visual-physiological double comfort under complex ambient light interference. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0019] Figure 1 It is a flowchart of a smart lighting control method.

[0020] Figure 2 It is a schematic diagram of a smart lighting control system.

[0021] Figure 3 It is a flowchart of a coordinate mapping table.

[0022] Figure 4 It is a flowchart of spectral parameters. Detailed implementation manners

[0023] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific implementation manners of the present invention with reference to the drawings in the specification.

[0024] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0025] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0026] Referring to Figures 1 to 4 , which is an embodiment of the present invention. This embodiment provides a smart lighting control method, including the following steps:

[0027] S1. Start three-dimensional space scanning through the millimeter-wave radar array deployed on the top of the classroom to obtain a coordinate mapping table.

[0028] S1.1. The millimeter-wave radar module deployed on the top of the classroom powers on for self-check, forms hardware initialization and establishes a communication link, and emits a frequency-modulated continuous wave signal through the radar array to cover the monitoring area.

[0029] Further, after the millimeter-wave radar module deployed on the top of the classroom completes power-on startup, it executes a hardware self-check process to verify the electrical performance of the antenna array, RF front-end, and signal processing unit. After confirmation, it establishes a communication link with the central controller through the SPI bus. The initialized millimeter-wave radar array emits a frequency-modulated continuous wave signal in the range of 30 - 64 GHz, using a linear frequency modulation method, with the beam width controlled within the range of ±10°, covering a monitoring area with a radius of 6 meters.

[0030] S1.2. Obtain the original data stream based on the electromagnetic wave signal reflected by the millimeter-wave radar.

[0031] Further, after the millimeter-wave radar array receives the frequency-modulated continuous wave signal reflected by the human body, it down-converts the RF signal to the baseband through a mixer and outputs an I / Q analog signal containing in-phase and quadrature components. The I / Q signal is converted into a digital signal through a 24-bit ADC, and the sampling rate is set to the example value of 10 MHz, forming an original data stream containing distance, speed, and angle information.

[0032] S1.3. Perform a fast Fourier transform to generate a three-dimensional point cloud dataset of distance-angle-velocity information, identify the seat area through the DBSCAN clustering algorithm, generate the central coordinates of each seat area, and form a coordinate mapping table.

[0033] Further, perform a fast Fourier transform on the complex time-domain signal sequence collected by the millimeter-wave radar array. Use a 2048-point FFT to calculate the distance dimension information, extract the velocity dimension component through Doppler processing, and calculate the angle dimension data by combining the phase differences of multiple antennas to generate a three-dimensional point cloud dataset containing distance-angle-velocity information; input the three-dimensional point cloud dataset into the DBSCAN clustering algorithm, set the neighborhood radius to 0.5 meters and the minimum number of samples to 5, and identify the density-connected point cloud clusters; calculate the geometric center coordinates of each point cloud cluster, determine the central positions of 16 seat areas with an accuracy of 0.1 meters, and output a coordinate mapping table containing region ID, three-dimensional coordinates, and coverage radius.

[0034] S2. The spectroradiometer scans the initial light environment, measures the spectral power distribution of the measurement band, and obtains the reference illumination parameter matrix.

[0035] S2.1. Capture the ambient light signal of the band through the spectroradiometer to obtain the original ADC readings.

[0036] Further, after the spectroradiometer is started, the photodiode array captures the ambient light signal in the band of 380 - 780 nm, converts the light signal into an analog electrical signal, adjusts the amplitude of the analog electrical signal through a programmable gain amplifier, and then inputs it into a 24-bit analog-to-digital converter for sampling. The sampling rate is set to the example value of 100 Hz; the digital signal output by the analog-to-digital converter is the original ADC reading.

[0037] S2.2. Deduct the background noise from the original ADC data to obtain the net optical signal intensity, and integrate the matching function of the net optical signal intensity to obtain the spectral power distribution curve.

[0038] Furthermore, after the original ADC readings are input into the digital signal processor, first read the pre-stored dark current reference value, which is measured under lightless conditions. Subtract the dark current reference value from the original ADC readings to obtain the net optical signal intensity after deducting the background noise. Apply the CIE1931 standard color matching function to the net optical signal intensity for weighted integration, calculate the spectral power distribution values at 5nm intervals for each sample in the range of 380 - 780nm. After the integration result is normalized, a standardized spectral power distribution curve is formed.

[0039] S2.3. Based on the spectral power distribution, calculate the illuminance, correlated color temperature, color rendering index, and melatonin equivalent irradiance, and form a reference lighting parameter matrix.

[0040] Furthermore, apply the photometry formula to the standardized spectral power distribution curve to calculate the illuminance value, use the CIE15:2004 standard method to obtain the correlated color temperature, calculate the color rendering index through the test color sample method, and obtain the melatonin equivalent irradiance according to the CIE S 026 standard. Combine the four parameters of illuminance, correlated color temperature, color rendering index, and melatonin equivalent irradiance in a fixed format to form a reference lighting parameter matrix.

[0041] S3. The meter-wave radar array emits electromagnetic waves to each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral densities of β waves and α waves, calculates the cognitive activity index, and forms a three-dimensional feature vector.

[0042] S3.1. Read the coordinate mapping table through the meter-wave radar array to obtain the regional center coordinates and coverage radius.

[0043] Furthermore, the millimeter-wave radar array reads the pre-generated coordinate mapping table from the storage unit. The coordinate mapping table contains the region IDs, three-dimensional center coordinates, and coverage radius information of 16 monitoring areas. The center coordinates of each monitoring area are recorded with an accuracy of 0.1 meter for each sample, and the coverage radius is fixed at 0.5 meter for each sample. The millimeter-wave radar array determines the spatial positions and scanning ranges of each monitoring area according to the center coordinates and coverage radius recorded in the coordinate mapping table.

[0044] S3.2. Generate a directional beam based on the regional center coordinates, emit electromagnetic waves to obtain amplitude information, and perform STFT transformation to generate a three-dimensional spectrogram.

[0045] Further, based on the central coordinates of the 16 monitoring areas recorded in the coordinate mapping table, the millimeter-wave radar array calculates the antenna phase weights corresponding to each area through a beamforming algorithm, generates a directional beam, and transmits a 60 GHz frequency-modulated continuous wave signal. The beam width is controlled within the example range of ±10°, ensuring that the signal covers the specified monitoring area. After receiving the reflected signal, it is down-converted by a mixer to obtain a baseband signal, and I / Q data containing amplitude and phase information is collected. A 256-point short-time Fourier transform is performed on the I / Q data, with a Hamming window used as the window function and a sliding step of 64 points for the time window, generating a three-dimensional spectrogram containing frequency-time-amplitude information.

[0046] S3.3. Separate the power spectral densities of β waves and α waves from the spectrogram and calculate the band power spectral density.

[0047] Further, perform frequency-domain analysis on the generated three-dimensional spectrogram, identify the characteristic frequency bands of β waves (14 - 30 Hz) and α waves (8 - 13 Hz) in the range of 8 - 30 Hz, use band-pass filters to extract the signals of β waves and α waves respectively, set the filter order to the example value of 8th order, and control the passband ripple within the example value of less than 0.5 dB; calculate the power spectral densities of the signals in each frequency band, convert the energy of the time-domain signal into frequency-domain power representation through Parseval's theorem, obtain the power spectral densities of β waves and α waves, perform logarithmic transformation and normalization processing on the power spectral densities of β waves and α waves, and output the standardized power spectral density values of β waves and α waves.

[0048] Specifically, the expression is ; where is the band power spectral density, is the power spectral density of the wave, is the power spectral density of the wave, and

[0049] S3.4. Obtain the cognitive activity index based on the calculation result of the band power spectral density.

[0050] Further, perform band-pass filtering on the EEG signals collected by the millimeter-wave radar, separate the signals in three frequency bands, calculate the power spectral density of each frequency band using Parseval's theorem, and the numerical range is usually from -1 to 2. A positive value indicates a cognitively active state, and the cognitive activity index is obtained.

[0051] S3.5. Process the cognitive activity index to obtain a d-dimensional feature vector.

[0052] Furthermore, based on the β-wave power spectral density and the α-wave power spectral density, the cognitive activity index is applied. During the calculation process, a moving average filter is performed on the β-wave power spectral density and the α-wave power spectral density. The time window is set to an example value of 30 seconds to eliminate instantaneous fluctuation interference. The calculation result is mapped to the 0-1 interval through the Sigmoid function to obtain the standardized cognitive activity index. Finally, the cognitive activity index of each of the 16 monitoring regions is output to form a 16-dimensional feature vector.

[0053] S4. Calculate the target color temperature adjustment amount according to the 16-dimensional feature vector. The ultraviolet sensor monitors the intensity of the outdoor band, and controls the bias voltage of the quantum dot layer to adjust the emission peak wavelength to obtain spectral parameters.

[0054] S4.1. Construct a three-layer fully connected neural network based on the 16-dimensional feature vector. Adopt the ReLU activation function and Dropout0.2 regularization. By combining the mean squared error loss and the biological rhythm constraint term, regularization is added to prevent overfitting.

[0055] Furthermore, the 16-dimensional feature vector is input into the three-layer fully connected neural network architecture. The first layer performs a linear transformation to expand the input dimension to 64 dimensions and processes it through the ReLU activation function. The second layer compresses the 64-dimensional features to 32 dimensions and applies ReLU again. The third layer maps the 32-dimensional features to a 1-dimensional output. During the training process, 20% of the neuron outputs in each layer are randomly set to zero to achieve Dropout0.2 regularization. Through the construction of the loss function, where the MSE term calculates the prediction error, the second term enforces the biological rhythm constraint, and the third term implements L2 weight decay. The Adam optimizer is used to iteratively update the network parameters with a fixed learning rate of 0.001 until the validation set loss does not decrease for 20 consecutive rounds, at which point the training is terminated.

[0056] S4.2. Use the Adam optimizer to train the three-layer fully connected neural network to obtain the neural-spectral mapping model.

[0057] Furthermore, the Adam optimizer initializes the first-order moment estimate and the second-order moment estimate as zero vectors. In each iteration, it calculates the gradient of the loss function with respect to the parameters of the three-layer fully connected neural network, updates the first-order moment estimate and the second-order moment estimate, performs bias correction on the parameters and then executes parameter updates. The learning rate divided by the square root of the corrected second-order moment estimate is used as the adaptive learning rate. During the training process, the mean squared error and the biological rhythm constraint term on the validation set are monitored. When the validation loss does not decrease for 20 consecutive rounds (example value), the training is terminated. Finally, the parameters of the three-layer fully connected neural network with the best performance on the validation set are saved to form the neural-spectral mapping model.

[0058] S4.3. Input the 16-dimensional feature vector into the neural-spectral mapping model to obtain the target color temperature adjustment amount.

[0059] Furthermore, after the 3D feature vector is input into the neural-spectral mapping model, the first fully connected layer performs a linear transformation to map the input to a 64-dimensional space and applies the ReLU activation function. The second fully connected layer compresses the 64-dimensional features to 32 dimensions and processes them again through the ReLU activation function. The third fully connected layer converts the 32-dimensional features into a one-dimensional output. After applying the hyperbolic tangent function to the output value to constrain it to the interval [-1, 1] and multiplying it by the example value of 500, the target color temperature adjustment amount is obtained, and its numerical range is limited between -500K and +-500K. Finally, the target color temperature adjustment amount output is added to the current color temperature as the color temperature control command for the lamp.

[0060] Specifically, the expression is ; Where is the target color temperature adjustment amount, is the th weight, is the th activation value, is the bias.

[0061] S4.3. Monitor the intensity of the ultraviolet band outside the window through an ultraviolet sensor and set the band intensity threshold.

[0062] Furthermore, the ultraviolet sensor continuously monitors the ambient ultraviolet radiation intensity in the 290-400nm band at an example sampling rate of 10Hz, and the output analog voltage signal is converted into a digital quantity by an ADC to set the band intensity threshold.

[0063] S4.4. When the band intensity is greater than the band intensity threshold, the control circuit applies a DC bias voltage to the quantum dot layer to control the emission peak wavelength.

[0064] Furthermore, when the band intensity is greater than the band intensity threshold, the control circuit outputs a DC bias voltage of an example value of 3.2V to the CdSe / ZnS quantum dot layer through a digital-to-analog converter. The applied bias voltage changes the energy band structure of the quantum dots, causing the emission peak wavelength to shift from the default 620nm to 525nm. During the wavelength adjustment process, the current change in the quantum dot layer is monitored in real time to ensure that the working current is stable within the example range of 20±1mA. The finally output spectral parameter records the peak wavelength offset to control the emission peak wavelength.

[0065] S4.5. Combine the emission peak wavelength, melatonin equivalent light intensity, and target color temperature adjustment amount to obtain spectral parameters.

[0066] Furthermore, input the three parameters of the peak wavelength of the quantum dot layer emission, the melatonin equivalent illumination intensity, and the target color temperature adjustment amount into the spectral synthesis algorithm. Combine them to generate a Gaussian spectral power distribution curve within the peak wavelength range of 380 - 780 nm. Set the full width at half maximum of the curve to the example value of 20 nm. Adjust the relative intensity of the spectral curve in the 460 - 480 nm band according to the melatonin equivalent illumination intensity. Finally, output the spectral parameters in JSON format including the peak wavelength, the target color temperature value, the melatonin equivalent illumination intensity value, and the normalized spectral power distribution curve.

[0067] S5. Use the reinforcement learning algorithm based on the spectral parameters, the reference illumination parameter matrix, and the n-dimensional eigenvector to obtain the control policy table.

[0068] S5.1. Combine the spectral parameters, the reference illumination parameter matrix, and the n-dimensional eigenvector into a cognitive-spectral state vector.

[0069] Furthermore, vertically splice the three parameters of the peak wavelength, the target color temperature, and the melatonin equivalent illumination intensity in the spectral parameters, the four parameters of illuminance, correlated color temperature, color rendering index, and melatonin equivalent illumination intensity in the reference illumination parameter matrix, and the n-dimensional eigenvector. Perform Min-Max normalization processing on the spliced n-dimensional eigenvector to linearly transform the values of each dimension to the interval [0, 1], where min and max are respectively the minimum and maximum values of each dimension on the training set, to obtain the normalized cognitive-spectral state vector.

[0070] S5.2. Construct an Actor-Critic network architecture based on the cognitive-spectral state vector to obtain the network parameters.

[0071] Furthermore, input the cognitive-spectral state vector into the Actor-Critic network architecture. The Actor network consists of an input layer (23 nodes), a first hidden layer (64 nodes with ReLU activation), a second hidden layer (64 nodes with ReLU activation), and an output layer (3 nodes with tanh activation), and outputs three control parameters of the pulse width modulation change amount / quantum dot bias change amount / color temperature adjustment change amount. The Critic network consists of a state input branch (23 nodes), an action input branch (3 nodes), a first hidden layer (128 nodes with ReLU activation), a second hidden layer (128 nodes with ReLU activation), and an output layer (1 node with linear activation), and outputs the state-action value Q value. The network parameters are initialized using the Xavier method. The weights of the output layer of the Actor network are initialized to a uniform distribution with an example value of 0.003, and the bias of the last layer of the Critic network is initialized to an example value of 0.01. Finally, save the initialized network weight matrix and bias vector as the network parameters.

[0072] S5.3. Update the network parameters using the PPO algorithm, map the updated network parameters to a JSON format control instruction table, and form a control strategy table.

[0073] Furthermore, when updating the Actor-Critic network parameters using the PPO algorithm, first calculate the probability ratio of the new and old policies, combine with the generalized advantage estimation, calculate the gradient through the loss function and update the parameters. The updated Actor network parameters are used to generate control instructions. Classify the three parameters of the pulse width modulation change amount, the quantum dot bias voltage change amount, and the color temperature adjustment change amount by region ID, convert them into JSON format key-value pairs, and finally output a JSON document containing all region control instructions as the control strategy table.

[0074] S6. Monitor the millimeter wave radar array, the spectroradiometer, and the quantum dot layer, start the self-calibration program, and obtain the repair log.

[0075] S6.1. Generate a device status vector by monitoring the millimeter wave radar array, the spectroradiometer, and the quantum dot layer.

[0076] Furthermore, the millimeter wave radar array outputs the signal-to-noise ratio value, the spectroradiometer records the current spectral drift value, and the quantum dot layer monitoring module feeds back the working current volatility. Concatenate the three parameters of the signal-to-noise ratio value, the spectral drift value, and the working current volatility in a fixed order to form the device status vector.

[0077] S6.2. Set the device status vector threshold based on the device status vector. When the device status vector is greater than the device status vector threshold, trigger the abnormal flag bit.

[0078] Furthermore, compare the device status vector with the preset threshold vector element by element. When the signal-to-noise ratio value is lower than 15 dB, or the spectral drift value exceeds 0.005, or the working current volatility is greater than 5%, set the corresponding dimension abnormal flag bit to 1. The three abnormal flag bits generate a global abnormal flag bit through an OR logical operation, otherwise keep the global abnormal flag bit = 0; the global abnormal flag bit triggers an interrupt signal through the GPIO interface.

[0079] S6.3. Execute the selected calibration protocol according to the abnormal flag bit, and calculate the equivalent circuit parameters by fitting the Nyquist curve.

[0080] Furthermore, when the global abnormal flag bit is 1, select the calibration protocol according to the abnormal type. For the millimeter wave radar abnormality, perform antenna array phase calibration; for the spectroradiometer abnormality, perform built-in standard light source calibration; for the quantum dot layer abnormality, perform electrochemical impedance spectroscopy test. Apply a 0.1 - 100 Hz swept AC signal to the quantum dot layer, collect complex impedance data, fit the Nyquist curve by the least squares method, and calculate the equivalent circuit parameters; Specifically, the expression is ; wherein, is the equivalent circuit parameter at frequency ; is the parallel resistance, is the series resistance, is the parallel capacitance, is the frequency index.

[0081] S6.4. Start the self - calibration program based on the equivalent circuit parameters to obtain the repair log.

[0082] Furthermore, based on the equivalent circuit parameters, adjust the compensation network parameters of the quantum dot layer driving circuit, update the series resistance value to 45 Ω, correct the parallel capacitance value to 2.3 μF, re - measure the working current volatility. If the volatility drops below the example value of 2%, it is determined that the calibration is successful, and a repair log in JSON format including the timestamp, device type, anomaly type, parameter comparison before and after calibration, and calibration status is generated.

[0083] This embodiment also provides an intelligent lighting control system, including: a reference lighting parameter matrix module, which starts three - dimensional space scanning through a millimeter - wave radar array deployed on the top of the classroom to obtain a coordinate mapping table, and a spectral radiometer scans the initial light environment to measure the spectral power distribution of the measurement band to obtain the reference lighting parameter matrix; A one - dimensional feature vector module. The millimeter - wave radar array emits electromagnetic waves to each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral density of the β - wave and α - wave, calculates the cognitive activity index, and forms a one - dimensional feature vector; A spectral parameter module. According to the one - dimensional feature vector through a neural - spectral mapping model, calculates the target color temperature adjustment amount, and the ultraviolet sensor monitors the intensity of the window - outside band, controls the bias voltage of the quantum dot layer to adjust the emission peak wavelength, and obtains the spectral parameters; A repair log module. According to the spectral parameters, the reference lighting parameter matrix and the one - dimensional feature vector, uses a reinforcement learning algorithm to obtain a control strategy table, monitors the millimeter - wave radar array, the spectral radiometer and the quantum dot layer, starts the self - calibration program, and obtains the repair log.

[0084] This embodiment also provides a computer device applicable to the intelligent lighting control method, including: a memory and a processor; the memory is used to store computer - executable instructions, and the processor is used to execute the computer - executable instructions to implement the intelligent lighting control method proposed in the above - mentioned embodiment.

[0085] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, carrier networks, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads provided on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0086] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the intelligent lighting control method proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disks, or optical discs.

[0087] In summary, through non-contact millimeter-wave detection technology, the present invention dynamically generates a multi-dimensional feature vector reflecting the cognitive state of students based on the power spectral density of β / α waves, realizing the direct coupling of physiological signals and lighting control, enabling the lighting adjustment to shift from passive environmental adaptation to active neural feedback regulation, achieving the beneficial effect of precisely matching the brain wave rhythm of learners with the light environment. By using a three-layer neural network model with biological rhythm constraints, the cognitive feature vector is mapped to the target color temperature adjustment amount, and the peak emission wavelength is dynamically adjusted through the bias voltage of ultraviolet-responsive quantum dots to form a closed-loop control of spectral parameters, realizing the multi-dimensional collaborative optimization of spectral characteristics, and achieving the beneficial effect of maintaining the best visual-physiological double comfort under complex environmental light interference.

[0088] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A smart lighting control method, characterized in that: including Start three-dimensional space scanning by deploying a millimeter-wave radar array on the top of the classroom to obtain a coordinate mapping table. The spectral radiometer scans the initial light environment and measures the spectral power distribution of the measurement band to obtain a reference illumination parameter matrix; The wave radar array emits electromagnetic waves to each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral densities of beta waves and alpha waves, calculates the cognitive activity index, and forms a dimensional feature vector; According to the dimensional feature vector, through the neural-spectral mapping model, calculate the target color temperature adjustment amount. The ultraviolet sensor monitors the intensity of the outdoor band and controls the bias voltage of the quantum dot layer to adjust the emission peak wavelength to obtain spectral parameters; According to the spectral parameters, the reference illumination parameter matrix, and the dimensional feature vector, use the reinforcement learning algorithm to obtain a control strategy table, monitor the millimeter-wave radar array, the spectral radiometer, and the quantum dot layer, and start the self-calibration program to obtain a repair log.

2. The intelligent lighting control method according to claim 1, wherein: Start three-dimensional space scanning by deploying a millimeter-wave radar array on the top of the classroom to obtain a coordinate mapping table, including the following steps The millimeter-wave radar module deployed on the top of the classroom performs power-on self-check, forms hardware initialization and establishes a communication link, and emits a frequency-modulated continuous wave signal through the radar array to cover the monitoring area; Based on the electromagnetic wave signals reflected by the millimeter-wave radar, obtain the original data stream; Perform fast Fourier transform to generate a three-dimensional point cloud dataset of distance-angle-velocity information, identify the seat area through the DBSCAN clustering algorithm, generate the central coordinates of each seat area, and form a coordinate mapping table.

3. The intelligent lighting control method according to claim 2, wherein: The spectral radiometer scans the initial light environment and measures the spectral power distribution of the measurement band to obtain a reference illumination parameter matrix including the following steps Capture the ambient light signal of the band through the spectral radiometer to obtain the original ADC readings; Deduct the background noise from the original ADC data to obtain the net light signal intensity, integrate the net light signal intensity matching function, and obtain the spectral power distribution curve; Based on the spectral power distribution, calculate the illuminance, correlated color temperature, color rendering index, and melatonin equivalent illuminance intensity to form a reference illumination parameter matrix.

4. The intelligent lighting control method according to claim 3, wherein: The wave radar array emits electromagnetic waves to each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral densities of beta waves and alpha waves, calculates the cognitive activity index, and forms a dimensional feature vector [[ID= ​ ​ ​ ​ ​ 5. The intelligent lighting control method according to claim 4, wherein: ​ Construct a three-layer fully connected neural network based on the dimensional feature vector, adopt the ReLU activation function and Dropout 0.2 regularization, combine the mean square error loss with the biological rhythm constraint term, and add regularization to prevent overfitting; Use the Adam optimizer to train the three-layer fully connected neural network to obtain a neural-spectral mapping model; Input the dimensional feature vector into the neural-spectral mapping model to obtain the target color temperature adjustment amount; Monitor the intensity of the window band through an ultraviolet sensor and set a band intensity threshold; When the band intensity is greater than the band intensity threshold, the control circuit applies a DC bias voltage to the quantum dot layer to control the emission peak wavelength; Combine the emission peak wavelength, the equivalent melatonin illumination intensity, and the target color temperature adjustment amount to obtain spectral parameters; 6. The intelligent lighting control method according to claim 5, wherein: Use the reinforcement learning algorithm according to the spectral parameters, the reference illumination parameter matrix, and the dimensional feature vector to obtain a control strategy table, including the following steps Merge the spectral parameters, the reference illumination parameter matrix, and the dimensional feature vector into a cognitive-spectral state vector; Construct an Actor-Critic network architecture based on the cognitive-spectral state vector to obtain network parameters; Use the PPO algorithm to update the network parameters, and map the updated network parameters to a JSON format control instruction table to form a control strategy table; 7. The intelligent lighting control method according to claim 6, characterized in that: Monitor the millimeter wave radar array, the spectral radiometer, and the quantum dot layer, and start the self-calibration program to obtain a repair log, including the following steps Generate a device state vector by monitoring the millimeter wave radar array, the spectral radiometer, and the quantum dot layer; Set a device state vector threshold based on the device state vector. When the device state vector is greater than the device state vector threshold, trigger an exception flag bit; Execute the selected calibration protocol according to the exception flag bit, and calculate the equivalent circuit parameters by fitting the Nyquist curve; Based on the equivalent circuit parameters, start the self-calibration program to obtain a repair log; 8. An intelligent lighting control system, based on the intelligent lighting control method according to any one of claims 1 to 7, characterized in that: Including A reference illumination parameter matrix module that starts a three-dimensional space scan through a millimeter wave radar array deployed on the top of the classroom to obtain a coordinate mapping table, and the spectral radiometer scans the initial light environment to measure the spectral power distribution of the band to obtain a reference illumination parameter matrix; A dimensional feature vector module. The wave radar array emits electromagnetic waves in each monitoring area according to the coordinate mapping table, receives the reflected signals, extracts the power spectral densities of the β wave and the α wave, calculates the cognitive activity index, and forms a dimensional feature vector; A spectral parameter module that calculates the target color temperature adjustment amount through a neural-spectral mapping model according to the dimensional feature vector. The ultraviolet sensor monitors the intensity of the window band, and controls the bias voltage of the quantum dot layer to adjust the emission peak wavelength to obtain spectral parameters; A repair log module that uses the reinforcement learning algorithm according to the spectral parameters, the reference illumination parameter matrix, and the dimensional feature vector to obtain a control strategy table, monitors the millimeter wave radar array, the spectral radiometer, and the quantum dot layer, and starts the self-calibration program to obtain a repair log; 9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent lighting control method according to any one of claims 1 to 7; 10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent lighting control method according to any one of claims 1 to 7;

Images