Festival light contextual model automatic adjusting system based on smart home system

Abstract

The invention relates to the technical field of smart home, and discloses a festival lamplight contextual model automatic adjusting system based on a smart home system, which comprises a main controller, a lamplight contextual model recognition module, a dynamic adjustment execution module, an environmental parameter acquisition module, a user behavior analysis module and the like. The light contextual model recognition module generates an initial contextual model according to festival types and time, the environment parameter acquisition module acquires environment data in real time, the user behavior analysis module analyzes user data to generate preference feature vectors, and the main controller fuses multi-dimensional information to generate a target contextual model instruction. And the dynamic adjustment execution module adjusts the light equipment accordingly. In addition, the system also has the functions of anomaly detection, adaptive correction, contextual model optimization, equipment compatibility detection, energy consumption balance and the like. According to the invention, intelligent and personalized adjustment of the festival light contextual model is realized, the equipment compatibility and the energy consumption management efficiency are improved, and better festival light experience is brought to users.

Description

Technical Field

[0001] The present invention relates to the field of smart home technology, and in particular to a festival lighting scene mode automatic adjustment system based on a smart home system. Background Art

[0002] As people's living standards improve, smart home products are becoming more and more popular, and people's demand for intelligent and personalized home environments is also growing. During the holidays, creating a unique and suitable lighting atmosphere has become the pursuit of many families, which has prompted the development of the holiday lighting scene mode automatic adjustment system. However, the relevant technologies on the current market still have many shortcomings.

[0003] On the one hand, the existing lighting scene mode adjustment system has a low level of intelligence in holiday scenes. Many systems only provide preset fixed lighting modes and cannot be dynamically adjusted according to the actual environment and user needs. For example, in different holiday atmospheres, the color, brightness and dynamic effects of the lights should be different to better fit the holiday theme. However, it is difficult for traditional systems to automatically generate suitable lighting scene modes based on factors such as the type of holiday, time, and ambient light intensity. For example, at Christmas, the outdoor ambient light is dim, and the indoor lights should be brighter and more colorful to create a cheerful atmosphere, but the existing system may not be able to automatically perceive environmental changes and make corresponding adjustments, resulting in the lighting effects not matching the holiday atmosphere and failing to meet users' diverse needs for holiday lights.

[0004] On the other hand, the personalized needs of users are not fully considered. Each user has different preferences and usage habits for lighting, but most existing systems lack in-depth analysis of user behavior and preferences. They cannot accurately capture the user's historical operation data, such as the frequency of scene mode switching, brightness adjustment range, and color preferences, and it is difficult to generate personalized lighting adjustment plans based on the user's real-time activity trajectory. For example, some users like to create a warm lighting atmosphere in the living room, but prefer soft light in the bedroom, but traditional systems cannot accurately adjust the lighting according to these personalized needs, which greatly reduces the user experience.

[0005] In addition, equipment compatibility and energy management issues are also prominent. Lighting equipment of different brands and models differ in functions and parameters. The existing system is not compatible with various types of lighting equipment, which can easily lead to problems such as parameter mismatch and inability to control normally. This not only limits users' choice of lighting equipment, but also affects the overall performance of the system. In terms of energy management, many systems lack effective energy balance strategies and are unable to reasonably regulate and control according to actual power consumption and energy consumption requirements of target scenario modes, resulting in serious energy waste. For example, in some situations where high brightness and complex dynamic effects are not required, lighting equipment still maintains high energy consumption, which increases the user's cost of use and is not in line with the development trend of energy conservation and environmental protection. Summary of the invention

[0006] The object of the present invention is to provide a holiday lighting scene mode automatic adjustment system based on a smart home system to solve the problems raised in the above background technology.

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a holiday lighting scene mode automatic adjustment system based on a smart home system, comprising a main controller, a lighting scene mode recognition module, a dynamic adjustment execution module, an environmental parameter acquisition module and a user behavior analysis module; The lighting scene mode recognition module is used to generate an initial scene mode according to the preset festival type and time information, and the initial scene mode includes lighting color parameters, brightness parameters and dynamic effect parameters; The environmental parameter acquisition module collects indoor light intensity, user location distribution data and external ambient light data in real time, and sends the data to the main controller; The user behavior analysis module generates a user preference feature vector by analyzing the user's historical operation data and real-time activity trajectory; The main controller receives the initial scene mode, environmental parameters and user preference feature vector, and generates a target scene mode instruction through a multi-dimensional fusion algorithm; The dynamic adjustment execution module adjusts the operating parameters of the lighting equipment according to the target scene mode instruction, and feeds back the adjustment status to the main controller in real time; The system also includes an abnormal mode detection module, which generates an abnormal adjustment signal if the deviation between the adjustment state fed back by the dynamic adjustment execution module and the target scenario mode instruction exceeds a preset threshold, and triggers an adaptive correction algorithm through the main controller.

[0008] Preferably, the specific implementation of the lighting scene mode recognition module is as follows: Matching preset holiday types according to calendar data and associating with the standard scene mode library corresponding to the holiday; Divide the scenario mode time periods based on time information, including warm-up period, core period and end period; During the warm-up period, the basic parameters in the standard scene mode library are weighted and fused with the real-time ambient light data to generate the initial brightness gradient value; During the core period, the spatial distribution weight of the lighting color is dynamically adjusted according to the user location distribution data; At the end of the period, the light dynamics parameters are reduced through a time decay function until the shutdown threshold is reached.

[0009] Preferably, the specific analysis process of the user behavior analysis module includes: Collect scene mode switching frequency, brightness adjustment range and color preference sequence from the user's historical operation data; Use clustering algorithms to classify user behaviors into high-activity, medium-activity, and low-activity categories, and generate category labels; The real-time activity trajectory is obtained through the indoor positioning system, and the coordinate set of the user's stay area and the length of stay are extracted; The category labels and the coordinate sets are subjected to a spatiotemporal correlation analysis to generate a user preference feature vector, which includes a region weight value, a color correlation degree, and a dynamic response coefficient.

[0010] Preferably, the specific steps of the multi-dimensional fusion algorithm are: Normalizing the initial scene mode parameters, the ambient light intensity and the user preference feature vector into a first matrix, a second matrix and a third matrix respectively; The implicit features of each matrix are extracted through convolutional neural network, and the feature similarity weights are calculated; Perform weighted summation of implicit features according to weights to generate a fusion feature matrix; The fused feature matrix is ​​mapped to the lighting parameter space, and the target scene mode instructions are output, including color coding values, brightness gradient curves, and dynamic effect frequencies.

[0011] Preferably, the adaptive correction algorithm is implemented as follows: When an abnormal regulation signal is generated, the main controller starts the correction priority queue and sorts the abnormal parameters according to the degree of deviation; For color coding deviation, the color gamut compensation model is used to adjust the gain coefficient of RGB channels; For brightness gradient deviation, the brightness curve is reconstructed by piecewise linear interpolation; For the dynamic effect frequency deviation, the fundamental frequency component is extracted by Fourier transform and the phase parameters are recalibrated.

[0012] Preferably, the system further includes a scenario mode optimization module, which is specifically implemented as follows: Collect the difference data between the target scenario mode instruction and the actual adjustment state to generate a difference feature set; Construct a differential reward function through a reinforcement learning model and optimize the weight allocation strategy of the multi-dimensional fusion algorithm; If the mean of the difference feature set after N consecutive optimizations is lower than the preset threshold, the parameters in the standard scenario model library are updated.

[0013] Preferably, the design of the difference reward function includes: Quantify the color difference, brightness difference and dynamic effect difference into the first reward sub-item, the second reward sub-item and the third reward sub-item respectively; Applying dynamic weights to the third reward sub-item according to the dynamic response coefficient in the user preference feature vector; The weight distribution strategy is iteratively updated through the gradient ascent algorithm until the reward function converges.

[0014] Preferably, the system further includes a device compatibility detection module, and its operation process is as follows: Scan the connected lighting devices to obtain the color range, maximum brightness and response delay parameters supported by the devices; Calculate the matching degree between the device parameters and the required parameters of the target scenario mode instructions; If the matching degree is lower than the preset threshold, a device downgrade instruction is generated to truncate the required parameters according to the device support range; If the matching degree is higher than the preset threshold, a device optimization instruction is generated to activate the device's overclocking mode to improve the response speed.

[0015] Preferably, the specific method for calculating the matching degree is: Perform intersection operation on the device color range and the target color encoding value to calculate the color gamut coverage ratio; Calculate the ratio of the maximum brightness of the device to the peak value of the target brightness gradient curve; Calculate the absolute value of the difference between the device response delay parameter and the inverse of the dynamic effect frequency; The above calculation results are combined into a matching score through a weighted summation formula.

[0016] Preferably, the system further includes an energy consumption balancing module, the implementation of which includes: Monitor the total power consumption of lighting equipment in real time and associate it with the expected energy consumption threshold of the target scenario mode command; If the total power consumption exceeds the preset ratio of the expected energy consumption threshold, the energy consumption hierarchical control strategy is activated: First, reduce the light brightness in low-priority areas, then reduce the frequency of dynamic effects, and finally adjust color saturation; If the total power consumption is lower than the preset ratio of the expected energy consumption threshold, the redundant energy consumption allocation model is activated to allocate the remaining power consumption to the area with the highest weight in the user preference feature vector.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The present invention generates an initial scene mode according to the preset festival type and time information through the lighting scene mode recognition module, and combines the data collected in real time by the environmental parameter acquisition module, such as indoor light intensity, user location distribution data and external ambient light data, and the user preference feature vector generated by the user behavior analysis module, and uses the multi-dimensional fusion algorithm of the main controller to generate the target scene mode instruction. This enables the system to automatically adjust the color, brightness and dynamic effect of the light according to different festivals, different time periods and real-time environmental changes. For example, during the Spring Festival, the system automatically increases the light brightness in the evening (preheating period) according to the gradual dimming of the external ambient light, and combines the user location distribution to enhance the color weight of the red light in the living room area where the family gathers, creating a festive atmosphere; in the core period of the evening, according to the characteristics of frequent user activities in the living room, the dynamic effect of the light is further optimized, such as increasing the flashing frequency, so that the festive atmosphere is more enthusiastic. This intelligent adjustment method greatly improves the fit between the lighting scene mode and the festive atmosphere, and provides users with a better holiday lighting experience.

[0018] The user behavior analysis module generates a user preference feature vector containing regional weight values, color associations, and dynamic response coefficients by collecting and analyzing user historical operation data and real-time activity trajectories. This enables the system to gain an in-depth understanding of each user's unique needs and usage habits, and to achieve personalized lighting adjustments. For example, for users who like to read in the bedroom, the system can automatically adjust the brightness and color temperature of the bedroom lights based on their length of stay in the bedroom and their operating habits, making them more suitable for reading; for highly active users who frequently switch lighting scene modes, the system can prioritize the scene modes that suit their preferences based on their historical switching frequency and preferences, thereby improving the convenience and satisfaction of user operations. This personalized design fully respects the individual differences of users, allowing each user to enjoy a holiday lighting environment that meets their needs.

[0019] The setting of the device compatibility detection module effectively solves the compatibility problem of different lighting devices. Before the system is run, the module scans the connected lighting devices, obtains the color range, maximum brightness and response delay parameters supported by the device, and calculates the matching degree with the required parameters of the target scenario mode command. If the matching degree is lower than the preset threshold, a device downgrade instruction is generated to truncate the required parameters according to the device support range; if the matching degree is higher than the preset threshold, a device optimization instruction is generated to activate the overclocking mode of the device to improve the response speed. This function ensures that the system can be well adapted to various brands and models of lighting equipment. Users are more free when choosing lighting equipment and are not restricted by compatibility issues. At the same time, it also improves the control accuracy and stability of the system for different devices and ensures the consistency of the lighting adjustment effect.

[0020] The energy consumption balancing module realizes the effective management of the energy consumption of lighting equipment. By real-time monitoring of the total power consumption of lighting equipment and associating the expected energy consumption threshold of the target scenario mode command, when the total power consumption exceeds the preset ratio of the expected energy consumption threshold, the energy consumption hierarchical control strategy is activated, which gives priority to reducing the light brightness of low-priority areas, then reduces the frequency of dynamic effects, and finally adjusts the color saturation, thereby effectively reducing energy consumption. When the total power consumption is lower than the preset ratio of the expected energy consumption threshold, the redundant energy consumption allocation model is activated, and the remaining power consumption is allocated to the area with the highest weight in the user preference feature vector to achieve the rational use of energy. For example, after the holiday activities are over, some areas are unattended, and the system automatically reduces the light brightness of these areas to reduce unnecessary energy consumption; while in key areas where users often move around, such as the sofa area in the living room, when there is surplus power consumption, the system automatically increases the light brightness of the area to improve the user experience. This energy consumption balancing strategy not only reduces the user's electricity cost, but also conforms to the social development trend of energy conservation and environmental protection, and has significant economic and environmental benefits.

[0021] The scenario mode optimization module generates a difference feature set by collecting the difference data between the target scenario mode instructions and the actual adjustment state, and uses the reinforcement learning model to build a difference reward function to optimize the weight allocation strategy of the multi-dimensional fusion algorithm. If the mean of the difference feature set is lower than the preset threshold after N consecutive optimizations, the parameters in the standard scenario mode library are updated. This enables the system to continuously learn and adapt to actual usage and continuously optimize the lighting adjustment effect. Over time, the system will respond more accurately to user needs and environmental changes, providing users with a higher quality and more stable holiday lighting scenario mode automatic adjustment service. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a working principle diagram of the holiday lighting scene mode automatic adjustment system of the present invention; Figure 2This is the workflow architecture diagram of the lighting scene mode recognition module; Figure 3 This is the working logic architecture diagram of the user behavior analysis module. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1-Figure 3 The present invention provides a technical solution: a holiday lighting scene mode automatic adjustment system based on a smart home system, the system comprising: Lighting scene mode recognition module: This module generates an initial scene mode based on the preset holiday type and time information. The initial scene mode covers lighting color parameters, brightness parameters and dynamic effect parameters. In actual operation, it obtains calendar data, accurately matches the preset holiday type, and associates the standard scene mode library of the corresponding holiday, providing a basis for subsequent operations.

[0025] Environmental parameter acquisition module: This module undertakes the important task of real-time acquisition of indoor light intensity, user location distribution data and external ambient light data, and sends these data to the main controller in a timely manner. Among them, indoor light intensity acquisition can be achieved with the help of light sensors distributed in different locations indoors. These sensors convert light signals into electrical signals and transmit them to the main controller after analog-to-digital conversion; user location distribution data can be obtained through indoor positioning systems, such as Bluetooth positioning, Wi-Fi positioning or UWB positioning technology; external ambient light data is collected by ambient light sensors installed in suitable locations outdoors.

[0026] User behavior analysis module: Generates user preference feature vectors by conducting in-depth analysis of user historical operation data and real-time activity trajectories. Specifically, first collect information such as scene mode switching frequency, brightness adjustment amplitude, and color preference sequence in the user's historical operation data, and then use clustering algorithms to divide user behaviors into high-activity, medium-activity, and low-activity categories, and generate corresponding category labels. The real-time activity trajectory is obtained through the indoor positioning system, from which the coordinate set and duration of the user's stay area are extracted. Finally, the category label and the coordinate set are subjected to spatiotemporal correlation analysis to generate a user preference feature vector containing regional weight values, color correlation, and dynamic response coefficients.

[0027] Main controller: As the core control unit of the system, the main controller receives the initial scene mode from the lighting scene mode recognition module, the environmental parameters from the environmental parameter acquisition module, and the user preference feature vector from the user behavior analysis module. The main controller uses a multi-dimensional fusion algorithm to comprehensively process this information and generate a target scene mode instruction. The instruction contains key parameters such as the color coding value of the light, the brightness gradient curve, and the dynamic effect frequency, providing an operating basis for the dynamic adjustment execution module.

[0028] Dynamic adjustment execution module: According to the target scenario mode command issued by the main controller, this module is responsible for adjusting the operating parameters of the lighting equipment, such as controlling the color, brightness and dynamic effects of the light. At the same time, it will feed back the adjustment status to the main controller in real time so that the main controller can understand the working status of the lighting equipment in time and realize closed-loop control.

[0029] Abnormal mode detection module: The system also has an abnormal mode detection module to monitor the deviation between the adjustment state fed back by the dynamic adjustment execution module and the target scenario mode instruction. If the deviation exceeds the preset threshold, the module will generate an abnormal adjustment signal and trigger the adaptive correction algorithm through the main controller to handle the abnormal situation and ensure that the operation of the lighting equipment meets expectations.

[0030] The present invention will be further described below in conjunction with Examples 1 to 5: Embodiment 1: This embodiment aims to explain in detail the specific working process of the lighting scene mode recognition module so that it can generate the initial scene mode more accurately to meet the lighting needs of different festivals and time periods.

[0031] The system reads the calendar data to identify the type of holiday corresponding to the current date. For example, if the detected date is December 25, it is determined to be Christmas; if it is the first day of the first lunar month, it is determined to be the Spring Festival. Each holiday has a corresponding standard scene mode library pre-set, which is stored in the system's storage device and contains common lighting scene mode parameters for that holiday. Taking Christmas as an example, the standard scene mode library may contain parameters such as red and green as the main colors, with flashing dynamic effects; the standard scene mode library for the Spring Festival may be red as the main color, with high brightness and festive and cheerful dynamic effects.

[0032] Based on time information, the lighting scene modes during the holiday season are divided into a warm-up period, a core period, and an end period. For example, for Christmas, the warm-up period can be set to 18:00 - 24:00 on December 24, the core period to 00:00 - 12:00 on December 25, and the end period to 12:00 - 24:00 on December 25. The lighting scene modes at different times will be different to create different atmospheres.

[0033] Warm-up period operation: In the warm-up period, in order to make the lighting effect better adapt to the changes in ambient light, the basic parameters in the standard scene mode library are weighted and fused with the real-time ambient light data to generate the initial brightness gradient value. Suppose the basic brightness parameter in the standard scene mode library is , the real-time ambient light intensity is , and the weighting coefficients are and ( ), then the initial brightness gradient value The calculation formula is: For example, if the base brightness in the Christmas standard scene mode library is 50 (assuming the brightness value range is 0 - 100), and the current ambient light intensity is 30, the weighting coefficient , , then the initial brightness gradient value .

[0034] During the core period, the spatial distribution weight of the lighting color is dynamically adjusted according to the user location distribution data. Assuming that the system obtains the user concentration in the living room area through the indoor positioning system, the color weights of red and green (for example, Christmas) in the living room area can be increased to make the lighting in this area more festive. In specific implementation, the color weight adjustment coefficient of each area can be calculated according to the density of user location distribution. Suppose the user density of a certain area is , the initial color weight of the region is , the adjusted color weight is , the adjustment factor is ( and Related, The bigger, The larger the .

[0035] At the end of the period, in order to gradually turn off the lights and avoid abruptness, the light dynamic effect parameters are reduced through the time decay function until the turning-off threshold is reached. Set the initial value of the light dynamic effect parameter to , the time decay function is ,in is the difference between the current time and the start time of the end period, and the elapsed time After that, dynamic effect parameters The calculation formula is: For example, if the initial value of the dynamic effect parameter is 80 (assuming the value range is 0 - 100), the time decay function is ( In hours), dynamic effect parameters 1 hour after the end period starts , as time goes by, the dynamic effect parameters gradually decrease. When it is less than the shutdown threshold (such as 10), the dynamic effect of the light is turned off.

[0036] Embodiment 2: The system continuously records the user's operation behavior on the lighting scene mode, including the scene mode switching frequency, brightness adjustment amplitude and color preference sequence. For example, in the past week, the user switched the lighting scene mode from "daily mode" to "romantic mode" 5 times, which is the scene mode switching frequency information; each time the user adjusts the brightness, the adjustment amplitude is recorded, such as adjusting from brightness 50 to 60, the adjustment amplitude is 10; at the same time, the different lighting colors selected by the user are recorded to form a color preference sequence. For example, if the user selects warm yellow light many times, the weight of warm yellow in the color preference sequence is higher.

[0037] Use clustering algorithms to analyze the collected user historical operation data, divide user behaviors into high-activity, medium-activity, and low-activity categories, and generate category labels. Common clustering algorithms such as the K-Means algorithm assume that (i.e. divided into 3 categories), the algorithm will automatically divide users into different categories based on the characteristics of user behavior data. For example, if a user frequently switches scene modes, adjusts brightness significantly, and chooses a variety of colors, his behavior may be classified as a high-activity category; while users who perform fewer operations and do not change much may be classified as a low-activity category.

[0038] The indoor positioning system is used to obtain the user's real-time activity trajectory and extract the coordinate set and duration of the user's stay area. For example, the coordinate range of the user in the living room is - , the stay time is 30 minutes. This information is combined with the category labels obtained by clustering user behaviors for spatiotemporal correlation analysis. Suppose the user category label is , the coordinate set of the stop area is , the duration of stay is , regional weight value The calculation of can consider the relationship between the length of stay and the activity of the category, such as ,in For and category Corresponding activity coefficient (high activity category Larger values, low activity categories value is smaller), and are the length of stay in other areas and the corresponding activity coefficients respectively. The color association can be determined based on the color preference selected by the user when staying in the area. For example, if the user selects blue light multiple times in a certain area, the color association of the area with blue is higher. The dynamic response coefficient can be calculated based on the user's operation frequency and amplitude in different areas. The higher the operation frequency and the larger the amplitude, the larger the dynamic response coefficient. Through these calculations, a user preference feature vector containing area weight value, color association and dynamic response coefficient is generated.

[0039] Embodiment 3: This embodiment describes in detail how the multi-dimensional fusion algorithm fuses the initial scene mode parameters, the ambient light intensity, and the user preference feature vector to generate a more reasonable target scene mode instruction.

[0040] The initial scene mode parameters, ambient light intensity and user preference feature vector are normalized into the first matrix, the second matrix and the third matrix respectively. Assume that the initial scene mode parameters include the light color parameters (The possible value range is RGB value), brightness parameter (like ), dynamic effect parameters (like The flashing frequency of , through the normalization formula (in is the original value, and is the minimum and maximum value of the parameter), normalized to interval, and obtain the first matrix. For the ambient light intensity (Assuming the value range is ), and the second matrix is ​​obtained by normalizing the same process. The user preference feature vector contains the regional weight value , color correlation , Dynamic response coefficient , and normalization is also performed to obtain the third matrix.

[0041] Convolutional neural network (CNN) is used to extract implicit features from each matrix. CNN contains multiple convolutional layers, pooling layers, and fully connected layers. It extracts features at different levels by sliding convolution kernels on the matrix. For example, after the convolutional layer, the first matrix extracts features related to the lighting scene mode, such as color matching features, brightness change trend features, etc.; the second matrix extracts ambient light intensity change features; and the third matrix extracts user preference related features. Then the similarity weights between the features of each matrix are calculated. Let the first matrix feature be , the second matrix characteristic is , the third matrix characteristic is , feature similarity weight , , The calculation of can be done by calculating the cosine similarity between feature vectors, such as (Here we calculate and ), and normalize them so that .

[0042] According to the calculated weights, the implicit features are weighted and summed to generate a fusion feature matrix. Suppose the fusion feature matrix is ,but This fused feature matrix combines information from the initial scene mode, ambient light, and user preferences.

[0043] The fused feature matrix is ​​mapped to the light parameter space and the target scene mode instruction is output. The fused feature matrix is ​​converted into the color coding value, brightness gradient curve and dynamic effect frequency of the light through the fully connected layer and other methods. For example, after the fused feature matrix is ​​processed by the linear transformation and activation function of the fully connected layer, the color coding value is (Red), the brightness gradient curve gradually increases from 50 to 80 in the next hour, and the dynamic effect frequency flashes once every 5 seconds, etc., indicating the target scenario mode.

[0044] Embodiment 4: This embodiment describes in detail how the adaptive correction algorithm corrects the lighting parameters when the system detects an adjustment abnormality to ensure the normal operation of the lighting equipment.

[0045] When the deviation between the adjustment state fed back by the dynamic adjustment execution module and the target scenario mode instruction exceeds the preset threshold, the abnormal mode detection module generates an adjustment abnormality signal, and the main controller starts the correction priority queue. For example, if the preset color deviation threshold is 10 (assuming that it is calculated based on the RGB value deviation), the brightness deviation threshold is 5, and the dynamic effect frequency deviation threshold is 2. When the color coding deviation is detected to be 15, the brightness gradient deviation is 8, and the dynamic effect frequency deviation is 3, the abnormal parameters are sorted according to the degree of deviation, and the brightness gradient deviation is processed first, then the color coding deviation, and finally the dynamic effect frequency deviation.

[0046] For color coding deviation, the color gamut compensation model is used to adjust the gain coefficient of the RGB channel. Assume that the original RGB channel value is , the deviation is , the adjusted RGB channel values ​​are , the gain coefficients are , , ,but , , . Gain factor The calculation of can be adjusted according to the deviation size and color gamut range, such as ( is the absolute value of the deviation) to ensure that the adjusted color is within a reasonable color gamut.

[0047] For the brightness gradient deviation, the brightness curve is reconstructed by piecewise linear interpolation. Assume that the target brightness gradient curve is , the actual measured brightness curve is , select multiple interpolation nodes during the time period with large deviations and According to the piecewise linear interpolation formula (when ), reconstruct the brightness curve to make the actual brightness curve closer to the target brightness curve.

[0048] For the dynamic effect frequency deviation, the base frequency component is extracted by Fourier transform and the phase parameters are recalibrated. Assume that the actual dynamic effect frequency signal is , through Fourier transform Get its spectrum and extract the fundamental frequency component Calculate the actual phase Phase with target Deviation , by adjusting the phase parameters, such as , recalibrate the phase of the dynamic effect frequency to restore the dynamic effect to normal.

[0049] Embodiment 5: This embodiment focuses on the specific implementation of the scenario mode optimization module, the device compatibility detection module and the energy consumption balancing module, and is committed to further improving system functions and enhancing system performance and user experience.

[0050] ①Scenario mode optimization module Difference data collection and feature set generation: The system continuously monitors and collects the difference data between the target scene mode command and the actual adjustment state. Taking color as an example, assuming that the target color encoding value is RGB (255, 0, 0), and the color value obtained after the actual adjustment is RGB (245, 5, 5), by calculating the difference between the two in each RGB channel, such as the red channel difference is 255 - 245 = 10, the green channel difference is 0 - 5 = -5, and the blue channel difference is 0 - 5 = -5, these differences are integrated into color difference data. For brightness, if the target brightness gradient curve is set to 60 at a certain moment (assuming the brightness value range is 0 - 100), and the actual measured brightness value is 55, then the brightness difference is 60 - 55 = 5. For dynamic effects, if the target dynamic effect frequency is flashing once every 4 seconds, and the actual measurement is flashing once every 5 seconds, then the dynamic effect difference is reflected as the frequency deviation. The difference data of these colors, brightness and dynamic effects are summarized to generate a difference feature set as the basis for subsequent optimization.

[0051] Difference reward function construction and weight optimization: Use the reinforcement learning model to construct a difference reward function. Quantify the color difference, brightness difference and dynamic effect difference as the first reward sub-item , Second reward item And the third reward sub-item For color differences, the Euclidean distance formula in color space is used for quantification, namely:

[0052] The larger the distance value calculated by this formula, the greater the color difference, and the higher the reward value. The smaller the value, the better the system is at reducing the color difference. The brightness difference is quantified as , directly take the absolute value of the difference between the target brightness and the actual brightness. The larger the difference, the smaller the reward value, which makes the actual brightness close to the target brightness. For the dynamic effect difference, set the target frequency to , the actual frequency is , the quantitative formula is . Based on the dynamic response coefficient in the user preference feature vector , dynamic weight is applied to the third reward sub-item, that is, the final dynamic effect reward sub-item is The weight allocation strategy of the multi-dimensional fusion algorithm is iteratively updated through the gradient ascent algorithm. In each iteration, the weight is adjusted according to the feedback of the reward function, and different weight combinations are continuously tried until the reward function converges. At this time, the weight allocation strategy obtained can enable the system to better meet actual needs and reduce differences when generating target scenario mode instructions.

[0053] Standard scenario library update: system sets continuous optimization threshold , after each optimization, calculate the mean of the difference feature set. After the optimization, the mean of the difference feature set is lower than the preset threshold, which indicates that the current optimization strategy is effective and the difference between the actual adjustment effect of the system and the target scene mode instruction is within an acceptable range. At this time, the optimized parameters are updated to the standard scene mode library, so that the standard scene mode library can keep pace with the times, adapt to the usage habits and environmental changes of different users, and provide more accurate basic data for subsequent holiday lighting scene mode adjustments.

[0054] ②Device compatibility detection module Device parameter acquisition: When the system starts the device compatibility detection process, it first scans the connected lighting devices. By interacting with the communication protocol of the lighting device, the color range supported by the device is obtained, such as the RGB color value range supported by some devices is (0 - 200, 0 - 200, 0 - 200); the maximum brightness of the device is obtained, assuming that the maximum brightness of a device is 800 lumens; and the response delay parameter of the device is obtained, such as the response delay of a device is 0.1 seconds. These parameters are the key basis for evaluating the compatibility of the device with the target scene mode command.

[0055] Matching degree calculation: Calculate the matching degree between the device parameters and the required parameters of the target scenario mode instruction. For the device color range and the target color encoding value, the intersection operation is used to calculate the color gamut coverage ratio. Suppose the device color range is represented in RGB space as , the target color encoding value is , then the color gamut coverage ratio The calculation method is: calculate the ratio of the intersection length to the device channel range length on each channel, and then take the average value. For the peak value of the device maximum brightness and the target brightness gradient curve, calculate the ratio of the two. Suppose the device maximum brightness is , the peak value of the target brightness gradient curve is , then the brightness ratio . For the device response delay parameter and the inverse of the dynamic effect frequency, calculate the absolute value of the difference, and set the device response delay to , the dynamic effect frequency is , then the delay difference Finally, the above calculation results are combined into a matching score through the weighted summation formula. ,Right now ,in , , is the weight coefficient, and , adjust the weights according to actual needs to highlight the importance of different parameters.

[0056] Instruction generation: Generate corresponding instructions based on the comparison result between the matching score and the preset threshold. If the matching degree is lower than the preset threshold, it means that the device cannot fully meet the requirements of the target scenario mode instructions, and a device downgrade instruction is generated. This instruction truncates the required parameters according to the device support range. For example, if the target color coding value exceeds the device color range, the color value is adjusted to the maximum or minimum value supported by the device; if the target brightness exceeds the maximum brightness of the device, the brightness is set to the maximum brightness of the device. If the matching degree is higher than the preset threshold, it means that the device is capable of better executing the target scenario mode instructions. At this time, a device optimization instruction is generated to activate the device's overclocking mode (if the device supports it) to increase the response speed and further optimize the lighting effect.

[0057] ③Energy consumption balancing module Power consumption monitoring and threshold association: The energy consumption balancing module monitors the total power consumption of lighting equipment in real time, and obtains the real-time power consumption data of all current lighting equipment by communicating with the power monitoring circuit of the lighting equipment or related smart meters, and accumulates the total power consumption. At the same time, the expected energy consumption threshold of the target scenario mode command is associated The threshold is pre-set based on different holiday scene modes, the number of lighting devices, and the expected usage time. For example, for a holiday lighting scene mode for a small party, the expected usage time is 3 hours. According to the power specifications of the lighting devices, the expected energy consumption threshold is set to 1 kWh (i.e. 1000 watt-hours).

[0058] Energy consumption classification control strategy: When the total power consumption Exceeding the expected energy consumption threshold The preset ratio (such as 110%, i.e. ), the energy consumption hierarchical control strategy is started. First, the light brightness of low-priority areas is reduced first. The system pre-classifies the regional priorities according to the regional weight values ​​in the user preference feature vector, and the areas with lower weight values ​​are low-priority areas. Assume that the current brightness of a low-priority area is , reduce the brightness by a certain ratio (such as 10%), and the new brightness value If the total power consumption still does not meet the standard after reducing the brightness of the low priority area, then reduce the frequency of dynamic effects. For example, if the original dynamic effect frequency is once every 3 seconds, adjust it to once every 5 seconds. If the total power consumption still exceeds the threshold, adjust the color saturation last. By adjusting the color saturation algorithm, reduce the color saturation to reduce the power consumption of the lighting equipment.

[0059] Redundant energy consumption allocation model: When the total power consumption Below expected energy consumption threshold The preset ratio (such as 90%, i.e. ), the redundant energy consumption allocation model is activated. According to the regional weight value in the user preference feature vector, the region with the highest weight is found and the remaining power consumption is allocated to this region. Assume that the remaining power consumption is , the current brightness of the area is , calculate the brightness value that can be increased based on the relationship between power consumption and brightness (such as a linear relationship) , ( is the proportional coefficient of power consumption and brightness change, which can be determined through equipment parameters and experiments), thereby improving the lighting brightness in the area and meeting users' lighting needs for key areas.

[0060] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0061] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A holiday lighting scene mode automatic adjustment system based on a smart home system, characterized in that: It includes a main controller, a lighting scenario pattern recognition module, a dynamic adjustment execution module, an environmental parameter acquisition module and a user behavior analysis module; The lighting scene mode recognition module is used to generate an initial scene mode according to the preset festival type and time information, and the initial scene mode includes lighting color parameters, brightness parameters and dynamic effect parameters; The environmental parameter acquisition module collects indoor light intensity, user location distribution data and external ambient light data in real time, and sends the data to the main controller; The user behavior analysis module generates a user preference feature vector by analyzing the user's historical operation data and real-time activity trajectory; The main controller receives the initial scene mode, environmental parameters and user preference feature vector, and generates a target scene mode instruction through a multi-dimensional fusion algorithm; The dynamic adjustment execution module adjusts the operating parameters of the lighting equipment according to the target scene mode instruction, and feeds back the adjustment status to the main controller in real time; The system also includes an abnormal mode detection module, which generates an abnormal adjustment signal if the deviation between the adjustment state fed back by the dynamic adjustment execution module and the target scenario mode instruction exceeds a preset threshold, and triggers an adaptive correction algorithm through the main controller.

2. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1 is characterized in that: The specific implementation of the lighting scene mode recognition module is as follows: Matching preset holiday types according to calendar data and associating with the standard scene mode library corresponding to the holiday; Divide the scenario mode time periods based on time information, including warm-up period, core period and end period; During the warm-up period, the basic parameters in the standard scene mode library are weighted and fused with the real-time ambient light data to generate the initial brightness gradient value; During the core period, the spatial distribution weight of the lighting color is dynamically adjusted according to the user location distribution data; At the end of the period, the light dynamics parameters are reduced through a time decay function until the shutdown threshold is reached.

3. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1 is characterized in that: The specific analysis process of the user behavior analysis module includes: Collect scene mode switching frequency, brightness adjustment range and color preference sequence from the user's historical operation data; Use clustering algorithms to classify user behaviors into high-activity, medium-activity, and low-activity categories, and generate category labels; The real-time activity trajectory is obtained through the indoor positioning system, and the coordinate set of the user's stay area and the length of stay are extracted; The category labels and the coordinate sets are subjected to a spatiotemporal correlation analysis to generate a user preference feature vector, which includes a region weight value, a color correlation degree, and a dynamic response coefficient.

4. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1 is characterized in that: The specific steps of the multi-dimensional fusion algorithm are: Normalizing the initial scene mode parameters, the ambient light intensity and the user preference feature vector into a first matrix, a second matrix and a third matrix respectively; The implicit features of each matrix are extracted through convolutional neural network, and the feature similarity weights are calculated; Perform weighted summation of implicit features according to weights to generate a fusion feature matrix; The fused feature matrix is ​​mapped to the lighting parameter space, and the target scene mode instructions are output, including color coding values, brightness gradient curves, and dynamic effect frequencies.

5. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1 is characterized in that: The implementation of the adaptive correction algorithm is as follows: When an abnormal regulation signal is generated, the main controller starts the correction priority queue and sorts the abnormal parameters according to the degree of deviation; For color coding deviation, the color gamut compensation model is used to adjust the gain coefficient of RGB channels; For brightness gradient deviation, the brightness curve is reconstructed by piecewise linear interpolation; For the dynamic effect frequency deviation, the fundamental frequency component is extracted by Fourier transform and the phase parameters are recalibrated.

6. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1, characterized in that: The system also includes a scenario mode optimization module, which is specifically implemented as follows: Collect the difference data between the target scenario mode instruction and the actual adjustment state to generate a difference feature set; Construct a differential reward function through a reinforcement learning model and optimize the weight allocation strategy of the multi-dimensional fusion algorithm; If the mean of the difference feature set after N consecutive optimizations is lower than the preset threshold, the parameters in the standard scenario model library are updated.

7. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 6 is characterized in that: The design of the difference reward function includes: Quantify the color difference, brightness difference and dynamic effect difference into the first reward sub-item, the second reward sub-item and the third reward sub-item respectively; Applying dynamic weights to the third reward sub-item according to the dynamic response coefficient in the user preference feature vector; The weight distribution strategy is iteratively updated through the gradient ascent algorithm until the reward function converges.

8. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1, characterized in that: The system also includes a device compatibility detection module, and its operation process is as follows: Scan the connected lighting devices to obtain the color range, maximum brightness and response delay parameters supported by the devices; Calculate the matching degree between the device parameters and the required parameters of the target scenario mode instructions; If the matching degree is lower than the preset threshold, a device downgrade instruction is generated to truncate the required parameters according to the device support range; If the matching degree is higher than the preset threshold, a device optimization instruction is generated to activate the device's overclocking mode to improve the response speed.

9. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 8, characterized in that: The specific method for calculating the matching degree is: Perform intersection operation on the device color range and the target color encoding value to calculate the color gamut coverage ratio; Calculate the ratio of the maximum brightness of the device to the peak value of the target brightness gradient curve; Calculate the absolute value of the difference between the device response delay parameter and the inverse of the dynamic effect frequency; The above calculation results are combined into a matching score through a weighted summation formula.

10. The holiday lighting scene mode automatic adjustment system based on the smart home system according to claim 1, characterized in that: The system also includes an energy consumption balancing module, which is implemented in the following ways: Monitor the total power consumption of lighting equipment in real time and associate it with the expected energy consumption threshold of the target scenario mode command; If the total power consumption exceeds the preset ratio of the expected energy consumption threshold, the energy consumption hierarchical control strategy is activated: First, reduce the light brightness in low-priority areas, then reduce the frequency of dynamic effects, and finally adjust color saturation; If the total power consumption is lower than the preset ratio of the expected energy consumption threshold, the redundant energy consumption allocation model is activated to allocate the remaining power consumption to the area with the highest weight in the user preference feature vector.

Images