A learning method for intelligent adjustment of an illumination environment

Abstract

The application provides a learning intelligent adjustment method of an illumination environment, acquires an ambient light signal and an image signal of a learning area, determines a position of a learner based on the image signal and determines a target illumination area according to the position of the learner; identifies a motion type and an emotional state of the learner based on the image signal, determines an illumination mode containing a target illuminance value and a color temperature target value according to the motion type and the emotional state; determines an ambient stray light component of the target illumination area based on the ambient light signal, determines a compensation light intensity according to the target illuminance value and the ambient stray light component, and outputs a light intensity adjustment signal, so that the actual illuminance reaches the target illuminance value; outputs a color temperature adjustment signal according to the illumination mode to adjust the color temperature of the light source; and the above steps are cyclically executed. The method makes the light source contribution component reversely follow the ambient stray light component, and the total illuminance converges to the target illuminance value under the stray light fluctuation, and the light parameters are synchronously updated according to the motion type and the emotional state of the learner.

Description

Technical Field

[0001] This invention relates to the field of intelligent lighting, specifically to a method for intelligent adjustment of learning lighting environment. Background Technology

[0002] Existing learning lighting solutions typically use photosensors to detect the total ambient light intensity and then uniformly adjust the light brightness according to the detected value, or switch the lights on and off at preset times. These solutions use the total ambient light intensity as the sole basis for adjustment, failing to distinguish between the target learning surface and stray light entering through windows or screens. When stray light is included in the effective illumination, it causes misjudgment of illuminance, resulting in the target surface's illuminance deviating from the required value. Furthermore, these solutions do not consider the learner's behavior and emotions; illuminance and color temperature are fixed or only coarsely adjusted according to the total light intensity, failing to match the differentiated illuminance needs of different learning situations such as reading, writing, and resting, and even less able to adjust the color temperature according to the learner's emotional state. In addition, learning lighting with a single photosensor and coarse adjustment results in repeated oscillations in output when ambient light fluctuates, making it difficult to achieve stable illuminance convergence. Summary of the Invention

[0003] This invention proposes a method for intelligent adjustment of learning lighting environment, comprising: S1. Collect ambient light signals and image signals of the learning area, determine the learner's position based on the image signals, and determine the target illumination area based on the learner's position; S2. Based on image signal recognition, identify the learner's action type and emotional state, and determine the lighting pattern according to the action type and emotional state. The lighting pattern includes the target illuminance value and the target color temperature value. S3. Determine the ambient stray light component of the target illumination area based on the ambient light signal, determine the target illuminance value corresponding to the current learning situation according to the illumination mode, determine the compensation light intensity according to the target illuminance value and the ambient stray light component, and output the light intensity adjustment signal to drive the light source so that the actual illuminance of the target illumination area reaches the target illuminance value. Here, the learning situation refers to the combination of the currently recognized action type and the recognized emotional state. S4. Output a color temperature adjustment signal according to the lighting mode to adjust the color temperature of the light source; S1 to S4 are executed in a loop.

[0004] Furthermore, based on image signal recognition, the learner's action type and emotional state are identified, and the illumination pattern is determined according to the action type and emotional state. Specifically, this includes: inputting the image signal into a pre-trained behavior recognition model, which outputs the learner's action type; extracting the face region from the image signal, extracting emotional features from the face region, and identifying the learner's emotional state based on the emotional features; determining the target illuminance value in the illumination pattern based on the action type, and determining the target color temperature value in the illumination pattern based on the emotional state; determining whether the emotional state is a negative emotional state; when it is determined to be a negative emotional state, shifting the target color temperature value towards the warm color range and reducing the target illuminance value; and combining the determined, shifted, or reduced target illuminance value with the target color temperature value to form the illumination pattern.

[0005] Furthermore, between S2 and S3, the following steps are performed: based on the image signal used in S2, the following operations are performed: extracting the face region and head posture from the image signal; determining whether the face region extraction was successful; when successful extraction is determined, extracting three features from the face region and head posture: blink frequency, percentage of eye-closing time, and viewing distance; normalizing and weighting the three features to generate an eye load index; when extraction fails, extracting the head tilt angle from the head posture; replacing the blink frequency and percentage of eye-closing time with the head tilt angle and viewing distance, normalizing and weighting the sum to generate an eye load index; and storing the generated eye load index for verification processing before determining the compensation light intensity in S3.

[0006] Furthermore, determining the ambient stray light component of the target illumination area based on the ambient light signal specifically includes: acquiring the illuminance values ​​collected by each illuminance collection point arranged within the learning area; decomposing the illuminance value of each illuminance collection point into a light source contribution component and an ambient stray light component according to the illuminance superposition relationship; reading the illuminance collection points covered by the target illumination area; and using the ambient stray light component of the covered illuminance collection points as the ambient stray light component of the target illumination area.

[0007] Furthermore, the compensation light intensity is determined based on the target illuminance value and the ambient stray light component. Specifically, this includes: reading the target illuminance value and the ambient stray light component of the target illuminated area; calculating the actual illumination contribution required by the light source for the target illuminated area by subtracting the ambient stray light component from the target illuminance value based on the illuminance superposition relationship, wherein the actual illumination contribution decreases as the ambient stray light component increases and increases as the ambient stray light component decreases; and calculating the compensation light intensity based on the required actual illumination contribution and the current output of the light source, so that the total illuminance after superimposing the actual illumination contribution of the light source and the ambient stray light component in the target illuminated area converges to the target illuminance value.

[0008] Furthermore, the output of the light intensity adjustment signal for driving the light source specifically includes: acquiring the actual illuminance of the target illumination area; calculating the deviation between the actual illuminance and the target illuminance value; generating a duty cycle adjustable switching drive signal as the light intensity adjustment signal; and performing closed-loop iterative update of the duty cycle based on the calculated deviation to adjust the conduction current of the light source.

[0009] Furthermore, after determining the target illuminance value corresponding to the current learning situation based on the illumination mode in S3, and before determining the compensation light intensity based on the target illuminance value and the ambient stray light component, the following operations are performed: reading the lower illuminance threshold and the upper illuminance threshold corresponding to the currently identified action type; determining whether the target illuminance value is lower than the lower illuminance threshold, and if it is determined to be lower, setting the target illuminance value to the lower illuminance threshold; reading the stored eye strain index; determining whether the eye strain index exceeds the preset strain threshold, and if it is determined to exceed the threshold, raising the target illuminance value to the upper illuminance threshold and outputting a prompt signal to the prompting device held by the user; and using the set or raised target illuminance value as the input for determining the compensation light intensity in S3.

[0010] Furthermore, the color temperature adjustment signal is output according to the illumination mode to adjust the color temperature of the light source. Specifically, this includes: reading the target color temperature value in the illumination mode; generating a color temperature adjustment signal based on the target color temperature value; distributing the color temperature adjustment signal to multiple constant current channels; driving the corresponding multi-color light-emitting units to emit light through each constant current channel, mixing the light to obtain the color temperature corresponding to the target color temperature value; and performing gamma correction on the constant current channels to make the obtained color temperature continuously adjustable.

[0011] Furthermore, before executing S1 in each cycle, the following operations are performed: the presence status of the learner in the learning area is monitored by infrared detection to obtain a presence status signal; based on the presence status signal, it is determined whether the learner has been continuously undetected for more than a preset time; when it is determined that the learner has been continuously undetected for more than a preset time, the light source is turned off and the execution of S1 to S4 is stopped, and the presence status of the learner is continuously monitored by infrared detection during the stop period; when the learner is detected again, the execution of S1 to S4 is resumed.

[0012] Furthermore, after S4 of each cycle is completed, the following operations are performed: record the change in the light intensity adjustment signal output by S3 in that cycle over time, and accumulate the learning period duration; determine whether the learning period has ended; when it is determined that the learning period has ended, generate usage record data based on the recorded changes in the light intensity adjustment signal and the accumulated learning period duration, store the usage record data in the cloud server, update the mapping relationship between action type and emotional state and lighting mode based on the usage record data stored in the cloud server, and distribute the updated mapping relationship; use the distributed mapping relationship as the basis for determining the lighting mode based on action type and emotional state in the next learning period.

[0013] This invention determines the learner's position based on image signals and then identifies the target illumination area based on that position. This narrows the illuminance determination from the entire learning area to the actual learning surface where the learner is located, effectively mitigating interference from external lighting conditions. It also identifies the learner's action type and emotional state based on image signals and determines the illumination pattern, causing the illuminance and color temperature values ​​to update synchronously with the learning behavior and emotional state, significantly improving the matching degree between the lighting environment and the learning situation. The invention determines the compensation light intensity based on the target illuminance value and ambient stray light components, causing the light source contribution component to follow the changes in the ambient stray light components, thus converging the total illuminance to the target illuminance value under ambient light fluctuations. Finally, it outputs a color temperature adjustment signal based on the illumination pattern to adjust the color temperature of the light source, causing the color temperature to adjust according to the emotional state, thereby improving the visual comfort of the learning lighting environment. The above steps S1 to S4 are executed cyclically, causing each step to continuously repeat, ensuring that the lighting environment continuously follows changes in the learning state. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a method for intelligent adjustment of learning lighting environment proposed in this invention. Detailed Implementation

[0015] refer to Figure 1This invention proposes an intelligent adjustment method for learning lighting environment. Addressing the problems of existing learning lighting methods that only detect total ambient light intensity, cannot distinguish between the target learning area and stray light entering through windows or screens, and do not adjust illuminance and color temperature according to the learner's actions and emotions, this invention decomposes the total illuminance within the target lighting area into two separately calculable mathematical components: a light source contribution component and an ambient stray light component. This prevents stray light from being included in the effective lighting and causing illuminance misjudgment. The compensation light intensity is determined in real-time based on the difference between the target illuminance value and the ambient stray light component, causing the light source contribution component to follow the ambient stray light component in the opposite direction, so that the superimposed total illuminance converges to the target illuminance value despite stray light fluctuations. Furthermore, the image signal simultaneously identifies the learner's action type and emotional state; the action type determines the illuminance, and the emotional state determines the color temperature, ensuring that the light parameters simultaneously match the eye use state and emotional state. The acquisition, determination, identification, and adjustment stages are executed cyclically, forming a closed loop of perception, decision-making, and execution.

[0016] The aforementioned method for intelligent adjustment of lighting environment specifically includes: S0. Before executing S1 in each cycle, the following human-sensor control operation is performed: The presence status of learners within the learning area is monitored by infrared detection to obtain a presence status signal. Based on the presence status signal, it is determined whether the learner has been continuously undetected for more than a preset time. When it is determined that the learner has been continuously undetected for more than the preset time, the light source is turned off and the execution of S1 to S4 is stopped. During the stop period, the presence status of learners is continuously monitored by infrared detection. When the learner is detected again, the execution of S1 to S4 is resumed.

[0017] Among them, the human-sensing control operation refers to the pre-gating operation that continuously monitors the learner's presence status through infrared detection and controls the opening and closing of the light source and the cyclic execution status based on the monitoring results; infrared detection refers to the presence status acquisition method composed of a pyroelectric human infrared sensor. This sensor is based on the pyroelectric effect, with Fresnel lenses focusing the 8 to 14 micrometer far-infrared light emitted by the human body. It is equipped with dual-element detection elements and differential comparison circuits to suppress false triggering caused by ambient temperature fluctuations, has a 120-degree wide-angle coverage, and a detection delay of no more than 0.5 seconds; the presence status signal refers to the level signal output by the sensor that characterizes whether there is learner activity in the learning area. It comes from the continuous monitoring of the learning area by the pyroelectric human infrared sensor; the preset duration refers to the human body departure judgment threshold set by the user in advance. It comes from the user's setting input, and its value is calibrated based on the typical class break and temporary departure duration. If the value is too short, the light will be turned off accidentally due to temporary obstruction. If the value is too long, the energy-saving effect during unoccupied periods will be weakened.

[0018] Specifically, since the energy used to keep the light source on for an extended period of time without detecting a learner constitutes wasted energy, the light source is turned off and the processes of acquisition, determination, identification, and adjustment are stopped. Since the infrared detector continues to operate during the shutdown period, the status signal immediately flips when the learner returns, and the cycle restarts from the acquisition stage. Whether the illuminance meets the standard is determined by the recovered acquisition, identification, and dimming stages. The gating system is only responsible for determining whether the learner is present and is independent of the dimming mechanism.

[0019] S1. Collect ambient light signals and image signals of the learning area, determine the learner's position based on the image signals, and determine the target illumination area based on the learner's position.

[0020] Among them, the ambient light signal refers to the electrical or digital signal representing the visible light flux received per unit area at each illuminance acquisition point in the learning area, which originates from the acquisition output of the illuminance acquisition component; the image signal refers to the image data obtained by the image acquisition unit from continuous shooting of the learning area, which originates from the shooting output of the image acquisition unit; the learner's position refers to the learner's spatial coordinates obtained by feature extraction from the image signal; and the target illumination area refers to the learning surface range directly affected by illumination adjustment with the learner's position as the center, which originates from the mapping of the learner's position.

[0021] Specifically, illuminance is defined as the luminous flux of visible light received per unit area, measured in lux. When an object is uniformly illuminated, the luminous flux received over one square meter is 1 lumen, and the illuminance of that object is 1 lux. The luminous flux over a surface element is... At that time, the illuminance of the surface element is:

[0022] in, Illuminance, measured in lux; The luminous flux received by a surface element, measured in lumens; The area of ​​the illuminated surface is expressed in square meters.

[0023] Specifically, the lighting equipment arrangement and external light sources have different effects on light intensity at different locations, and the indoor light intensity distribution is uneven. Therefore, at least two illuminance sampling points are set up at different locations in the learning area, and the sampling results are recorded as follows: to .

[0024] in, The illuminance value at the first illuminance sampling point, in lux; The illuminance value at the second illuminance sampling point, in lux; For the first Illuminance value at the illuminance collection point, in lux; This represents the total number of illuminance collection points.

[0025] The illuminance acquisition component can be a ready-made illuminance meter, which provides an interface to output the acquisition results to the communication module, and then the wireless communication unit transmits them to the intelligent control module. The acquisition results are transmitted once every fifteen minutes.

[0026] The communication module refers to the interface circuit responsible for data transmission, originating from the output interface of the illuminance acquisition unit. The wireless communication unit refers to the transmission unit that wirelessly forwards the acquired data. The intelligent control module refers to the processing unit that performs determination, recognition, and adjustment calculations based on the acquired and image data, originating from the data input of each sensing unit and the image acquisition unit. The illuminance acquisition unit can also use a digital illuminance sensor instead of a commercial illuminance meter to reduce costs. This sensor is based on the photoelectric effect principle, incorporates a photodiode array, and generates an environmental parameter dataset after sliding mean filtering. The image signal is obtained by the image acquisition unit. The behavior recognition module extracts the learner's posture and hand movement features from the image based on computer vision, outputs the learner's position, and locks the target illumination area based on the learner's position. The behavior recognition module is the computational unit that performs posture and movement feature extraction and position output, originating from the image output of the image acquisition unit.

[0027] S2. Based on image signal recognition, identify the learner's action type and emotional state, and determine the lighting pattern according to the action type and emotional state. The lighting pattern includes the target illuminance value and the target color temperature value.

[0028] Among them, action type refers to the result output by the behavior recognition model that represents the learner's current learning behavior category, which comes from the classification output of the behavior recognition model on the image signal; emotional state refers to the emotion category result determined by the emotional features of the face region, which comes from the output of the emotion feature classifier; illumination mode refers to the combination of light parameters consisting of target illuminance value and target color temperature value; target illuminance value refers to the illuminance value of the target surface pointed to by the illumination mode; and target color temperature value refers to the color temperature value of the light source pointed to by the illumination mode.

[0029] Specifically, the behavior recognition model is trained using deep learning. Computer vision's object detection and behavior classification capabilities extract learner posture and hand gesture features from image signals, outputting the learner's action type. Action types include at least reading, writing, resting, and projected viewing. The initial source of the target illuminance value is a preset mode: different lighting modes and corresponding illuminance standard values ​​are set for different usage scenarios in the learning area. This illuminance standard value and the target illuminance value determined by the action type are cross-checked, with action type taking priority. Specifically, illuminance requirements vary with the sophistication of the visual task, so action type serves as the primary control dimension for illuminance; because the spectral energy distribution of warm light is conducive to emotional relaxation, emotional state serves as the dimension for color temperature regulation, thus adapting the lighting mode to both visual and emotional states.

[0030] Furthermore, based on image signal recognition, the learner's action type and emotional state are identified, and the illumination pattern is determined according to the action type and emotional state. Specifically, this includes: inputting the image signal into a pre-trained behavior recognition model, which outputs the learner's action type; extracting the face region from the image signal, extracting emotional features from the face region, and identifying the learner's emotional state based on the emotional features; determining the target illuminance value in the illumination pattern based on the action type, and determining the target color temperature value in the illumination pattern based on the emotional state; determining whether the emotional state is a negative emotional state; when it is determined to be a negative emotional state, shifting the target color temperature value towards the warm color range and reducing the target illuminance value; and combining the determined, shifted, or reduced target illuminance value with the target color temperature value to form the illumination pattern.

[0031] The behavior recognition model, after training, classifies posture and hand gesture features. Facial features are extracted after facial key point localization. These features are input into a classifier to obtain the emotional state. Negative emotional states refer to categories such as fatigue and irritability, which represent emotional burden. The detectable features are the combination of the drooping degree of facial key points and the opening and closing degree of the eyelids. When the emotional state is negative, such as fatigue or irritability, the target color temperature value shifts towards the warmer color range, and the target illuminance value is reduced. This is because the spectral energy distribution of warm light is conducive to relaxation. The magnitude of the shift and reduction is set according to the confidence level of the emotional features. The combination of the determined, shifted, or reduced target illuminance value with the target color temperature value constitutes the lighting pattern.

[0032] Furthermore, the following is also included between S2 and S3: Based on the image signal used in S2, the following operations are performed: extract the face region and head posture from the image signal; determine whether the face region extraction was successful; if the extraction was successful, extract three features from the face region and head posture: blink frequency, percentage of time spent with eyes closed, and viewing distance; normalize the three features and sum them by weight to generate an eye strain index; if the extraction failed, extract the head tilt angle from the head posture; replace the blink frequency and percentage of time spent with eyes closed with the head tilt angle and viewing distance, normalize and sum them by weight to generate an eye strain index; save the generated eye strain index for the verification process in S3 before determining the compensation light intensity.

[0033] Among them, the eye load index is a dimensionless quantity that characterizes the degree of visual fatigue of learners, obtained by normalizing and weighting the sum of blink frequency, the proportion of eye closure time, and viewing distance. It is derived from the weighted operation of the three features. Blink frequency refers to the count of eyelid closure events per unit time. The proportion of eye closure time refers to the proportion of eyelid closure frames to the total number of frames. Viewing distance refers to the distance between the learner's eyes and the learning surface. Head tilt angle refers to the tilt angle of the head relative to the vertical direction in head posture, which is derived from the estimated output of head posture.

[0034] Specifically, the decrease in blinking frequency and the increase in the proportion of eye closure time are explicit physiological signals of continuous ciliary muscle tension. The shortened viewing distance increases the accommodative load. Therefore, the weighted sum of the three features can quantify the eye load. The typical case of failure in face region extraction is when the learner lowers their head or turns their face. In this case, the head tilt angle and viewing distance are used as substitutes. The increased head tilt angle and the shortened viewing distance have the same signal meaning. The substitute path ensures that the eye load index can still be calculated continuously when the face is not visible, without data interruption. The retained eye load index is read by the verification process before the light intensity is determined. The data streams of the two stages are connected end to end.

[0035] S3. Determine the ambient stray light component of the target illumination area based on the ambient light signal, determine the target illuminance value corresponding to the current learning situation according to the illumination mode, determine the compensation light intensity based on the target illuminance value and the ambient stray light component, and output the light intensity adjustment signal to drive the light source so that the actual illuminance of the target illumination area reaches the target illuminance value. Here, the learning situation refers to the combination of the currently recognized action type and the recognized emotional state.

[0036] Among them, the ambient stray light component refers to the illuminance portion contributed by light sources other than the light source within the target illumination area, which is derived from the difference calculation between the illuminance value at the illuminance acquisition point and the component contributed by the light source; the compensation light intensity refers to the light intensity that needs to be adjusted for the light source to offset the fluctuation of the ambient stray light component; the light intensity adjustment signal refers to the drive signal output by the intelligent control module to the brightness adjustable light module, which is derived from the calculation output of the intelligent control module; the brightness adjustable light module refers to the light-emitting component that is driven by the light intensity adjustment signal and changes the output light intensity.

[0037] Specifically, the illuminance value at each illuminance collection point is the sum of the contribution from indoor light sources and the contribution from outdoor light passing through the window, satisfying the illuminance superposition relationship:

[0038] in, The illuminance value at this illuminance collection point, in lux; The light source contribution component to this point, expressed in lux; The ambient stray light component contributing to this point, in lux.

[0039] The contribution component of the light source is calculated based on the light source parameters:

[0040] in, Number of light sources; Luminous flux of a single light source, expressed in lumens; This is the utilization factor; For maintenance factor; The target area is expressed in square meters.

[0041] Therefore, when the light source parameters are known, the ambient stray light components at each point can be solved by the illuminance superposition relationship. The ambient stray light components at each point covered by the target illumination area constitute the ambient stray light component input of that area.

[0042] Furthermore, determining the ambient stray light component of the target illumination area based on the ambient light signal specifically includes: Obtain the illuminance values ​​collected by each illuminance collection point arranged within the learning area; based on the illuminance superposition relationship, decompose the illuminance value of each illuminance collection point into the light source contribution component and the ambient stray light component; read the illuminance collection points covered by the target illumination area; and use the ambient stray light component of the covered illuminance collection points as the ambient stray light component of the target illumination area.

[0043] Specifically, the target illumination area only covers a portion of the illuminance collection points, so the entire area is first decomposed and then read according to the coverage relationship to avoid stray light from outside the area affecting the judgment results within the area. The two actions of decomposition and reading are connected end to end, forming a complete calculation path from the illuminance value of each illuminance collection point to the environmental stray light component of the target illumination area.

[0044] Furthermore, the compensation light intensity is determined based on the target illuminance value and the ambient stray light component, specifically including: Read the target illuminance value and the ambient stray light component of the target illuminated area; based on the illuminance superposition relationship, subtract the ambient stray light component from the target illuminance value to calculate the actual illuminance contribution required by the light source to the target illuminated area.

[0045] Among them, the actual illumination contribution decreases as the ambient stray light component increases and increases as the ambient stray light component decreases; based on the required actual illumination contribution and the current output of the light source, the compensation light intensity is calculated so that the total illuminance after the superposition of the actual illumination contribution of the light source and the ambient stray light component in the target illumination area converges to the target illuminance value.

[0046] Specifically, the ambient stray light component changes with time, weather, and environment, and the total illuminance needs to be kept stable. Therefore, the intelligent control module calculates the direction and magnitude of the light adjustment required based on the current light intensity and the target light intensity, so that the light source contribution component follows the ambient stray light component in the opposite direction, so that the updated illuminance values ​​of each illuminance collection point are kept within ±3% of the current illuminance standard value. The compensation light intensity is the light intensity that the light source contribution component needs to be adjusted.

[0047] Furthermore, the output includes a light intensity modulation signal for driving the light source, specifically including: Obtain the actual illuminance of the target illumination area; calculate the deviation between the actual illuminance and the target illuminance value; generate a duty cycle adjustable switching drive signal as a light intensity adjustment signal; perform closed-loop iterative update of the duty cycle based on the calculated deviation to adjust the conduction current of the light source.

[0048] Specifically, the intelligent control module uses a proportional-integral-derivative adjustment algorithm to dynamically output the duty cycle of the switch drive signal, driving the execution unit to generate four 1000 Hz switch drive signals, which respectively drive the dimming circuit of the light-emitting unit and the color temperature adjustment module. The duty cycle can be steplessly adjusted between 0% and 100%, and the brightness linearity error does not exceed ±3%.

[0049] Among them, the execution component refers to the circuit component that converts the calculation output of the intelligent control module into a drive signal; the high-frequency switching characteristic eliminates visible flicker and precisely controls the conduction current of the light-emitting unit, with a control calculation response time of less than 50 milliseconds and an instruction execution delay of less than 200 milliseconds.

[0050] The second implementation path for light intensity scaling is: The brightness-adjustable lighting module consists of a switching voltage regulation circuit composed of unidirectional thyristors. The off-time of the circuit is adjusted by adjusting the resistance of the potentiometer. The shorter the off-time, the greater the output current and voltage, thus achieving the purpose of voltage regulation and dimming.

[0051] The closed-loop process is as follows: After system initialization, the storage unit is cleared and the initial values ​​of each conversion chip are set. The system detects the light on / off status. When the light is on, it performs illuminance sampling and controls the dimming output. When the light is off, it detects whether there is anyone in the learning area. If there is someone, it turns on the light and samples the dimming output. If there is no one, it samples and returns the output.

[0052] In one specific implementation: When the target illuminance is 300 lux and the actual illuminance is 282 lux, the deviation is -18 lux. The proportional-integral-derivative adjustment algorithm increases the duty cycle according to the deviation. After closed-loop iteration, the actual illuminance stabilizes in the range of 300 lux ± 3 lux.

[0053] Furthermore, after determining the target illuminance value corresponding to the current learning situation based on the illumination mode in S3, and before determining the compensation light intensity based on the target illuminance value and the ambient stray light component, the following operations are performed: The system reads the lower and upper illuminance thresholds corresponding to the currently identified action type; determines whether the target illuminance value is lower than the lower illuminance threshold. If it is, the target illuminance value is set as the lower illuminance threshold. It reads the stored eye strain index; determines whether the eye strain index exceeds a preset strain threshold. If it is, the target illuminance value is increased to the upper illuminance threshold, and a prompt signal is output to the user's device. The set or increased target illuminance value is used as the input for determining the compensation light intensity in step S3. The lower and upper illuminance thresholds refer to the illuminance boundary values ​​defined for each action type according to the visual task illuminance standard. The upper and lower limits of the same action type constitute the illuminance safety range for that action type.

[0054] Among them, the load threshold refers to the judgment boundary that triggers the increase action based on the eye load index, and is calibrated according to the correspondence between the eye load index and the subjective fatigue score; the prompting device refers to the terminal that receives prompting signals and issues reminders to the user, which comes from the user's possession.

[0055] Specifically, the instantaneous recognition result of the action type may fall too low, so a lower threshold is used as a fallback. Since the eye load index reflects cumulative visual fatigue, the higher the load, the higher the illumination requirement. Therefore, when the load threshold is exceeded, the upper limit is increased and a prompt signal is output to remind the user to rest. The stored eye load index is read, which corresponds to the location where the generated eye load index is stored. Both paths produce the target illumination value after being set or increased. This value is used as the input to determine the compensation light intensity.

[0056] In one specific implementation: The lower limit threshold for illuminance for reading activities is set at 300 lux, and the upper limit threshold is set at 500 lux, based on the standard illuminance range for reading visual tasks; the load threshold is set at 0.7; when the eye load index accumulates to 0.8, the target illuminance value is increased from 300 lux to 500 lux, and a rest reminder is simultaneously sent to the user's mobile phone.

[0057] S4. Output a color temperature adjustment signal according to the lighting mode to adjust the color temperature of the light source; S1 to S4 are executed in a loop.

[0058] Among them, the color temperature adjustment signal refers to the driving signal that controls the current of each channel of the multicolor light-emitting unit to achieve light mixing, which comes from the calculation output of the intelligent control module based on the target color temperature value; the multicolor light-emitting unit refers to the light-emitting component composed of multiple color light-emitting bodies, which outputs the target color temperature after light mixing.

[0059] Specifically, the color temperature control layer uses a constant current driver chip with a built-in 12-bit pulse width modulation generator and a three-channel constant current controller, which supports independent output of red, green, blue and white channels. The 16 MHz clock frequency and ±1% current matching accuracy ensure multi-channel color synchronization and smooth transition. The constant current driver chip refers to the driving component that provides constant current channels and pulse width modulation output.

[0060] Furthermore, a color temperature adjustment signal is output according to the lighting mode to adjust the color temperature of the light source, specifically including: The system reads the target color temperature value from the lighting mode; generates a color temperature adjustment signal based on the target color temperature value; distributes the color temperature adjustment signal to multiple constant current channels; drives the corresponding multi-color light-emitting units to emit light through each constant current channel, and mixes the light to obtain the color temperature corresponding to the target color temperature value; performs gamma correction on the constant current channels to make the obtained color temperature continuously adjustable. The red, green, and blue channels use a gamma correction algorithm to ensure linear color transition, achieving continuous color temperature adjustment from 3,000 Kelvin to 6,500 Kelvin; the progressive brightness adjustment uses a gradient algorithm to ensure that there are no visual jumps in the lighting transition.

[0061] Specifically, the human eye's perception threshold for color temperature changes is lower than its perception threshold for brightness changes. Therefore, gamma correction and gradient algorithms work together to make the color temperature change gradually and smoothly.

[0062] In one specific implementation: It adopts a spectrally tunable light-emitting array, supports color temperature from 2700 to 6500 Kelvin and five-channel mixing of red, green, blue, amber and white light, and is equipped with voice control and remote control interaction. The shell is made by 3D printing and is assembled with development board, sensor, programmable light strip and voice module.

[0063] After S4 of each loop is completed, the following operations are performed: The system records the changes in the light intensity modulation signal output by S3 to drive the light source over time during this cycle and accumulates the learning period duration; it determines whether the learning period has ended; when the learning period has ended, it generates usage record data based on the recorded changes in the light intensity modulation signal and the accumulated learning period duration, stores the usage record data in the cloud server, updates the mapping relationship between action type and emotional state and lighting mode based on the usage record data stored in the cloud server, and distributes the updated mapping relationship; the distributed mapping relationship serves as the basis for determining the lighting mode based on action type and emotional state in the next learning period.

[0064] Among them, the recorded data refers to the data set recorded by the intelligent control module according to the set sampling period, which includes the duty cycle sequence of the light intensity modulation signal and the start and end times of the learning period, and comes from round-by-round recording and cumulative calculation; the cloud server refers to the remote storage and calculation node connected to the intelligent control module through the communication module; the mapping relationship refers to the correspondence table between action type and emotional state and target illuminance value and color temperature target value, which comes from the mode preset and is updated by the cloud server.

[0065] Specifically, the mapping relationship is updated based on time-period statistics rather than single-round instantaneous values. Therefore, at the end of each cycle, only two lightweight actions, recording and accumulating, are performed. Generation, storage, updating, and distribution are only performed once after the learning period ends. Updating round by round is unnecessary and increases the communication and computational burden. The system connects to a large model for data analysis, recording data such as the time learners use the desk lamp and changes in light brightness, and storing this data in the cloud. After data processing and analysis, the lighting is adjusted in a targeted manner: when a certain action type and emotional state combination is accompanied by the user's adjustment of illumination multiple times in the historical record, the cloud server updates the target illumination value corresponding to the combination to the user's preferred value, and sends the updated mapping table to the intelligent control module via the communication module. When determining the lighting mode in the next learning period, the updated mapping table is directly used. Users can also remotely control the device via a mobile app, voice assistant, or remote control. The remote control is divided into three parts: signal transmission, interrupt handling, and signal reception. When a function key is pressed, the infrared diode emits a pulse signal corresponding to the function. The transmission lasts for three milliseconds, and the transmission stops for one millisecond. During the reception process, the dimming pulse delay value is set according to the set brightness data. If the low-level pulse width is less than two milliseconds, the device returns to an interrupt; if it is greater than two milliseconds, the device continues to execute. If the high-level pulse width is greater than three milliseconds, the function program is executed according to the number of pulses.

[0066] In one specific implementation: the learning period is set to 2 hours based on the duration of a complete task; the duty cycle sequence is recorded once per second during the period, and the data is packaged and uploaded after the period ends. The cloud updates the target illuminance value corresponding to the combination of fatigue state and reading action type from 300 lux to 320 lux and sends it out. When determining the lighting mode for the next period, the updated correspondence table is directly used.

[0067] The above S1 to S4 are executed in a loop, continuously collecting, determining, identifying and adjusting, so that the illuminance and color temperature of the target lighting area are continuously matched with the learner's learning state.

[0068] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent adjustment of learning lighting environment, characterized in that, include: S1. Collect ambient light signals and image signals of the learning area, determine the learner's position based on the image signals, and determine the target illumination area based on the learner's position; S2. Based on image signal recognition, identify the learner's action type and emotional state, and determine the lighting pattern according to the action type and emotional state. The lighting pattern includes the target illuminance value and the target color temperature value. S3. Determine the ambient stray light component of the target illumination area based on the ambient light signal, determine the target illuminance value corresponding to the current learning situation according to the illumination mode, determine the compensation light intensity according to the target illuminance value and the ambient stray light component, and output the light intensity adjustment signal to drive the light source so that the actual illuminance of the target illumination area reaches the target illuminance value. Here, the learning situation refers to the combination of the currently recognized action type and the recognized emotional state. S4. Output a color temperature adjustment signal according to the lighting mode to adjust the color temperature of the light source; S1 to S4 are executed in a loop.

2. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, Based on image signal recognition, learners' action types and emotional states are identified, and a lighting pattern is determined according to the action type and emotional state. Specifically, this includes: inputting image signals into a pre-trained behavior recognition model, which outputs the learner's action type; extracting facial regions from the image signals, extracting emotional features from the facial regions, and identifying the learner's emotional state based on the emotional features; determining the target illuminance value in the lighting pattern based on the action type, and determining the target color temperature value in the lighting pattern based on the emotional state; determining whether the emotional state is negative; when it is determined to be a negative emotional state, shifting the target color temperature value towards the warm color range and reducing the target illuminance value; and combining the determined, shifted, or reduced target illuminance value with the target color temperature value to form the lighting pattern.

3. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, Between S2 and S3, the following steps are also performed: based on the image signal used in S2, further operations are performed: extracting the face region and head posture from the image signal; determining whether the face region extraction was successful; when the extraction is successful, extracting three features from the face region and head posture: blink frequency, percentage of eye-closing time, and viewing distance; normalizing and weighting the three features to generate an eye load index; when the extraction fails, extracting the head tilt angle from the head posture; replacing the blink frequency and percentage of eye-closing time with the head tilt angle and viewing distance, normalizing and weighting the sum to generate an eye load index; and storing the generated eye load index for verification processing before determining the compensation light intensity in S3.

4. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, Determining the ambient stray light component of the target illumination area based on ambient light signals specifically includes: acquiring the illuminance values ​​collected by each illuminance collection point arranged within the learning area; decomposing the illuminance value of each illuminance collection point into a light source contribution component and an ambient stray light component according to the illuminance superposition relationship; reading the illuminance collection points covered by the target illumination area; and using the ambient stray light component of the covered illuminance collection points as the ambient stray light component of the target illumination area.

5. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, The compensation light intensity is determined based on the target illuminance value and the ambient stray light component. Specifically, this includes: reading the target illuminance value and the ambient stray light component of the target illuminated area; calculating the actual illumination contribution required by the light source for the target illuminated area by subtracting the ambient stray light component from the target illuminance value based on the illuminance superposition relationship, wherein the actual illumination contribution decreases as the ambient stray light component increases and increases as the ambient stray light component decreases; and calculating the compensation light intensity based on the required actual illumination contribution and the current output of the light source, so that the total illuminance after superimposing the actual illumination contribution of the light source and the ambient stray light component in the target illuminated area converges to the target illuminance value.

6. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, The output is a light intensity adjustment signal for driving the light source, specifically including: obtaining the actual illuminance of the target illumination area; calculating the deviation between the actual illuminance and the target illuminance value; generating a duty cycle adjustable switching drive signal as the light intensity adjustment signal; and performing closed-loop iterative update of the duty cycle based on the calculated deviation to adjust the conduction current of the light source.

7. The intelligent adjustment method for learning lighting environment according to claim 3, characterized in that, After determining the target illuminance value corresponding to the current learning situation based on the illumination mode in S3, and before determining the compensation light intensity based on the target illuminance value and the ambient stray light component, the following operations are performed: read the lower illuminance threshold and the upper illuminance threshold corresponding to the currently identified action type; determine whether the target illuminance value is lower than the lower illuminance threshold, and if it is determined to be lower, set the target illuminance value to the lower illuminance threshold; read the stored eye strain index; determine whether the eye strain index exceeds the preset strain threshold, and if it is determined to exceed, raise the target illuminance value to the upper illuminance threshold and output a prompt signal to the prompting device held by the user; use the set or raised target illuminance value as the input for determining the compensation light intensity in S3.

8. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, The color temperature adjustment signal is output according to the illumination mode to adjust the color temperature of the light source. Specifically, it includes: reading the target color temperature value in the illumination mode; generating a color temperature adjustment signal according to the target color temperature value; distributing the color temperature adjustment signal to multiple constant current channels; driving the corresponding multi-color light-emitting unit to emit light through each constant current channel, mixing the light to obtain the color temperature corresponding to the target color temperature value; and performing gamma correction on the constant current channels to make the obtained color temperature continuously adjustable.

9. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, Before executing S1 in each cycle, the following operations are performed: the presence status of the learner in the learning area is monitored by infrared detection to obtain the presence status signal; based on the presence status signal, it is determined whether the learner has not been detected for more than a preset time. If it is determined that the learner has not been detected for a preset time, the light source is turned off and the execution of S1 to S4 is stopped. During the stop period, the learner's presence status is continuously monitored by infrared detection. When the learner is detected again, the execution of S1 to S4 is resumed.

10. The intelligent adjustment method for learning lighting environment according to claim 1, characterized in that, After S4 in each cycle is completed, the following operations are performed: record the change of the light intensity modulation signal output by S3 in that cycle as a function of time, and accumulate the duration of the learning period. Determine whether the learning period has ended; when it is determined that the learning period has ended, generate usage record data based on the recorded changes in light intensity modulation signals and the cumulative duration of the learning period, store the usage record data in the cloud server, update the mapping relationship between action type and emotional state and light pattern based on the usage record data stored in the cloud server, and send the updated mapping relationship down. The mapping relationship will be used as the basis for determining the lighting pattern based on the action type and emotional state in the next learning period.

Images