Self-adaptive adjusting system of indoor intelligent lighting equipment

Abstract

The invention discloses a self-adaptive adjusting system for indoor intelligent lighting equipment, and relates to the technical field of intelligent lighting, the self-adaptive adjusting system comprises a data acquisition module, a data analysis module and an intelligent control module, the data acquisition module is used for carrying out area marking on the indoor intelligent lighting equipment based on an indoor lighting function area, the data acquisition module acquires user primary area track data and environment data of the lighting equipment area data; the data analysis module is used for training a lighting control strategy mechanical learning model for predicting a lighting function area based on lighting equipment area data, user primary area track data and environment data; by means of the lighting control strategy mechanical learning model trained by the data analysis module and used for predicting the lighting function area, it can be guaranteed that lighting equipment in the lighting function area is started at the accurate time point and cannot be delayed, energy conservation and environmental protection are guaranteed, energy consumption is reduced, and the personal experience of a user is also guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent lighting, and particularly relates to an adaptive adjustment system for indoor intelligent lighting equipment. Background Art

[0002] Indoor intelligent lighting equipment refers to the product of combining the Internet of Things with indoor lighting equipment. By integrating core technologies such as digital information processing, wireless sensor networks, broadband power line carrier communication, and intelligent terminal control, a distributed remote monitoring and control system is constructed. This system not only realizes basic functions such as stepless dimming, progressive lighting switching, and timed power management, but also supports users to customize multi-scene light effect modes (such as meeting mode, cinema mode, energy-saving mode, etc.), and achieves precise energy control and personalized lighting experience through intelligent algorithm optimization;

[0003] In the existing adaptive adjustment system of indoor intelligent lighting equipment, for example, a lighting unit adaptive adjustment method of an intelligent lighting system disclosed in a Chinese patent application with the publication number CN118102545A cannot accurately and timely control the startup of lighting equipment, resulting in a very poor experience for individual users. This is because the startup time of the startup device of the lighting equipment is not accurate enough and has a certain lag. Suddenly turning on the lighting equipment when just entering the dark will stimulate people's eyes and affect people's eyesight. Therefore, the actual experience is very bad. Summary of the Invention

[0004] In order to overcome the above technical problems, the purpose of the present invention is to provide an adaptive adjustment system for indoor intelligent lighting equipment to solve the problem in the prior art that due to the inaccurate startup time of the lighting equipment and a certain lag, suddenly turning on the lighting equipment when just entering the dark will stimulate people's eyes and affect people's eyesight, resulting in a very poor experience for individual users.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] Specifically, an adaptive adjustment system for indoor intelligent lighting equipment is provided, including a data acquisition module, a data analysis module, and an intelligent control module. The data acquisition module is used to: mark the indoor intelligent lighting equipment based on the lighting functional areas in the room to obtain lighting equipment area data. The data acquisition module collects the user's primary area trajectory data and environmental data of the lighting equipment area data. The data analysis module is used to: train a lighting control strategy machine learning model for predicting the lighting functional area based on the lighting equipment area data, the user's primary area trajectory data, and the environmental data. The intelligent control module is used to: control the lighting equipment in the lighting functional area based on the lighting control strategy predicted by the lighting control strategy machine learning model. The intelligent control module collects the user's secondary area trajectory data in the lighting functional area after being controlled based on the lighting control strategy through the data acquisition module, and the intelligent control module optimizes the lighting control strategy based on the user's secondary area trajectory data.

[0007] As a further solution of the present invention: The data acquisition module uses Ri to mark the lighting functional area, where Re ∈ [R1, R2, ……, Ri];

[0008] The data acquisition module uses Lj to mark the lighting equipment in the lighting functional area, where Lj ∈ [L1, L2, ……, Lj];

[0009] The lighting equipment area data is Ri of the lighting functional area.

[0010] As a further solution of the present invention: A distance sensing module is provided in the lighting functional area, and the distance sensing module is used to collect the distance d between the user individual and the distance sensing module;

[0011] The data acquisition module generates the user's primary area trajectory data or the user's secondary area trajectory data based on the distance d between the user individual and the distance sensing module.

[0012] As a further solution of the present invention: The method by which the data acquisition module generates the user's primary area trajectory data or the user's secondary area trajectory data based on the distance d between the user individual and the distance sensing module is as follows:

[0013] The data acquisition module collects the distances d1, d2, ……, dn between the user individual and the distance sensing module at the same time intervals within the time period t by the distance sensing module;

[0014] Based on the distances d1, d1, ……, dn and the time period t, a primary coordinate axis is generated. The horizontal axis of the primary coordinate axis is the time period t, and the vertical axis is the distance d;

[0015] Based on the curve formed by the primary coordinate axis, a primary movement trajectory is generated, and the primary movement trajectory is the user's primary area trajectory data;

[0016] The data acquisition module acquires the distance d between the user individual and the distance sensing module at the same time interval within the moment t , 1. d , 1. d , 2. ……, d , m;

[0017] Based on d , 1. d , 2. ……, d , m and the moment t , generate a secondary coordinate axis, where the horizontal axis of the secondary coordinate axis is the moment t , , and the vertical axis is the distance d , ;

[0018] Generate a secondary movement trajectory based on the curve formed by the secondary coordinate axis, and the secondary movement trajectory is the user's secondary area trajectory data.

[0019] As a further solution of the present invention: an image acquisition module is provided in the lighting function area, and the image acquisition module is used to acquire image data of the user individual in the lighting function area;

[0020] The data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the image data in the lighting function area.

[0021] As a further solution of the present invention: the method for the data acquisition module to generate user primary area trajectory data or user secondary area trajectory data based on the image data in the lighting function area is as follows:

[0022] The data acquisition module obtains the image data P1, P2, ……, Pn acquired by the image acquisition module at the same time interval within the moment T based on the image acquisition module;

[0023] Based on the image data P1, P2, ……, Pn, use a convolutional neural network to identify the primary movement trajectory of the user individual in the lighting function area, and the primary movement trajectory of the user individual in the lighting function area is the user primary area trajectory data;

[0024] The data acquisition module obtains the image data P , acquired by the image acquisition module at the same time interval within the moment T , 1. P , 2. ……, P , m;

[0025] Based on the image data P , 1. P , 2. ……, P ,The convolutional neural network is used to identify the secondary movement trajectory of the user individual within the lighting function area, and the secondary movement trajectory of the user individual within the lighting function area is the user secondary area trajectory data.

[0026] As a further solution of the present invention: The data analysis module pre-collects historical training data and trains a machine learning model for predicting the lighting control strategy of the lighting function area based on the historical training data.

[0027] As a further solution of the present invention: The historical training data includes N sets of training data, where N is a positive integer. Each set of training data includes feature data and label data. The feature data includes lighting device area data, user primary area trajectory data, and environmental data, where:

[0028] The lighting device area data refers to the lighting function area Ri determined by the data acquisition module when each set of training data is collected;

[0029] The user primary area trajectory data refers to the movement trajectory of the user individual within the lighting function area when each set of training data is collected;

[0030] The environmental data refers to the sound intensity value and light intensity value collected by the data acquisition module when each set of training data is collected;

[0031] The label data refers to the lighting control strategy of the intelligent control module when each set of training data is collected, and the lighting control strategy is the startup time of the lighting device.

[0032] As a further solution of the present invention: The method by which the data analysis module trains a machine learning model for predicting the lighting control strategy of the lighting function area is as follows:

[0033] Convert the feature data in the historical data into feature vectors, use the feature vectors as the input of the machine learning model, use the startup time predicted by the machine learning model for the corresponding feature data as the output, use the startup time in the label data corresponding to the feature data as the training target, and use minimizing the sum of prediction accuracies as the training target. The technical formula for prediction accuracy is: ai = (ui - wi) 2 where ai is the prediction accuracy, ui is the predicted startup time corresponding to the i-th set of feature data, wi is the startup time in the label data corresponding to the i-th set of feature data, and train the machine learning model until the prediction accuracy reaches convergence and then stop training.

[0034] As a further solution of the present invention: The method by which the intelligent control module optimizes the lighting control strategy based on the user secondary area trajectory data is as follows:

[0035] The data acquisition module generates the speed variance of the user individual moving over time based on the secondary movement trajectory and sets a preset variance threshold;

[0036] If the speed variance ≤ the variance threshold, the lighting control strategy does not need to be optimized;

[0037] If the speed variance > the variance threshold, the lighting control strategy is optimized through the speed variance.

[0038] Advantages of the present invention:

[0039] In the present invention, the data acquisition module will collect the lighting device area data, the user's primary area trajectory data, and the environmental data in real time, and then convert the lighting device area data, the user's primary area trajectory data, and the environmental data into digital signals and transmit them to the data analysis module. After receiving the digital signals of the lighting device area data, the user's primary area trajectory data, and the environmental data, the data analysis module uses the trained lighting control strategy machine learning model to predict the start time of the lighting devices in the lighting control function area. After receiving the digital signal of the start of the lighting device, the intelligent control module can directly control the lighting devices in the lighting function area to turn on. Through the machine learning model of the lighting control strategy for predicting the lighting function area trained by the data analysis module, it can ensure that the lighting devices in the lighting function area start at the accurate time point without delay, that is, it ensures energy conservation and environmental protection, reduces energy consumption, and also ensures the user's personal experience. Description of the drawings

[0040] The present invention will be further described below with reference to the accompanying drawings.

[0041] Figure 1 is a module flowchart of an adaptive adjustment system for an indoor intelligent lighting device of the present invention;

[0042] Figure 2 is a process flowchart of an adaptive adjustment system for an indoor intelligent lighting device of the present invention. Detailed implementation manners

[0043] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0044] As a real-time mode of the present invention, as Figure 1 and Figure 2 shown, an adaptive adjustment system for an indoor intelligent lighting device is disclosed, including a data acquisition module, a data analysis module, and an intelligent control module;

[0045] Among them, the data acquisition module is used to: based on the indoor lighting functional areas, mark the indoor intelligent lighting devices to obtain lighting device area data. The data acquisition module collects the user's primary area trajectory data and environmental data of the lighting device area data. It should be noted that the data acquisition module obtains digital signals through the sensor interface, converts the digital signals into analog signals or converts the analog signals into digital signals. During the process of processing digital signals or analog signals, it is necessary to perform preprocessing such as filtering, calibration, amplification, and denoising on the digital signals or analog signals to ensure data quality. Filtering is the primary step to suppress noise and interference, which is divided into two categories: digital filtering and analog filtering. Digital filtering selectively attenuates signals in the time domain or frequency domain through algorithms (such as finite impulse response FIR and infinite impulse response IIR filters). For example, low-pass filtering can eliminate high-frequency noise, while band-stop filtering can suppress power frequency interference (such as 50 / 60 Hz power supply noise). For non-stationary signals, moving average filtering or median filtering can effectively smooth out burst interference. Calibration eliminates systematic errors through calibration, including static calibration (fitting polynomials or look-up tables based on standard input and output) and dynamic calibration (compensating for sensor response delay or non-linearity). Amplification aims to adjust the signal amplitude to match the dynamic range of the analog-to-digital converter (ADC) to avoid quantization errors. Digital amplification multiplies the signal by a gain factor (such as amplifying a microvolt-level electrical signal to a volt-level), but the risk of amplified noise needs to be vigilant. Usually, it is necessary to combine pre-amplification or a programmable gain amplifier (PGA) to optimize the signal-to-noise ratio; dynamic range compression (such as logarithmic amplification) is used to process wide-range signals. Denoising focuses on eliminating random interference in the signal. Common methods include adaptive filtering (such as the LMS algorithm to track noise characteristics in real time), wavelet threshold denoising (separating high-frequency noise components), and machine learning methods (such as deep learning-based noise suppression models);

[0046] The indoor lighting functional areas can be various indoor environments, such as indoor shopping malls, indoor factories, or indoor residences. It should be noted that although the present invention defines indoor intelligent lighting devices, the intelligent lighting devices extended from indoor to outdoor should also be within the protection scope of the present invention. For example, the adaptive adjustment system of outdoor intelligent lighting devices adopting the technical solution of the present invention;

[0047] For an indoor shopping mall, the lighting functional areas can be divided according to different areas of the indoor shopping mall, such as aisles, bathrooms, underground parking lots, stairwells, etc. The specific division of the lighting functional areas is determined by those skilled in the art according to the actual area size or function of the indoor shopping mall;

[0048] Indoor factories or indoor residences are the same as indoor shopping malls. For example, the division of the lighting functional areas of an indoor residence includes the living room, dining room, kitchen, bathroom, study, balcony, bedroom, etc., which will not be elaborated here;

[0049] The same applies outdoors. The lighting functional areas can be divided according to area or according to their usage functions.

[0050] Then, according to the Ri markers of the lighting functional areas, where Re ∈ [R1, R2, ……, Ri], the data acquisition module uses Lj markers for the lighting devices within the lighting functional areas, where Lj ∈ [L1, L2, ……, Lj]. R1, R2, ……, Ri respectively represent different lighting functional areas, and L1, L2, ……, Lj respectively represent different lighting devices. The installation positions and quantities of L1, L2, ……, Lj can be set arbitrarily within the lighting functional areas. The lighting devices can be LED lights or other lighting fixtures.

[0051] Environmental data refers to the sound intensity value and light intensity value collected by the data acquisition module. Therefore, a sound sensor and a light sensor need to be set within the lighting functional areas. The data acquisition module collects the sound intensity value within the lighting functional areas through the sound sensor, and the data acquisition module collects the light intensity value within the lighting functional areas through the light sensor.

[0052] It should be noted that the installation quantities and installation positions of the sound sensor and the light sensor are both arbitrarily set by those skilled in the art within the lighting functional areas, as long as it helps the data acquisition module accurately collect the sound intensity value and the light intensity value. The sound intensity value and the light intensity value can also be the average value or the average value after removing abnormal items. For example, remove the obviously too large sound intensity value (external accidental noise) and the obviously too large light intensity value (external accidental light source) to ensure the accuracy of the collection of the sound intensity value and the light intensity value.

[0053] The data analysis module is used to: train a machine learning model for predicting the lighting control strategy of the lighting functional areas based on the lighting device area data, the user's primary area trajectory data, and the environmental data. It should be noted that after the data acquisition module collects the lighting device area data, the user's primary area trajectory data, and the environmental data, it converts the lighting device area data, the user's primary area trajectory data, and the environmental data into digital signals and transmits them to the data analysis module. It should be noted that when the data acquisition module collects a large amount of lighting device area data, the user's primary area trajectory data, and the environmental data, the data analysis module can analyze and process the large amount of lighting device area data, the user's primary area trajectory data, and the environmental data collected by the data acquisition module. Specifically:

[0054] First, the data analysis module pre-collects historical training data, that is, a large amount of lighting device area data, the user's primary area trajectory data, and the environmental data pre-collected by the data acquisition module.

[0055] Then, the data analysis module trains a machine learning model for predicting the lighting control strategy of the lighting functional area based on historical training data;

[0056] The historical training data includes N sets of training data, where N is a positive integer. Usually, the value of N is from one hundred thousand to over one million. Each set of training data includes feature data and label data. The feature data includes lighting device area data, user's one-time area trajectory data, and environmental data, where:

[0057] The lighting device area data refers to the lighting functional area Ri determined by the data acquisition module when collecting each set of training data. As explained before, the lighting functional area is adaptively divided by those skilled in the art. Here, taking an indoor residence as an example, the living room, dining room, kitchen, bathroom, study, balcony, and bedroom are marked as R1, R2, R3, R4, R5, R6, and R7 in sequence;

[0058] The user's one-time area trajectory data refers to the movement trajectory of the user individual within the lighting functional area when collecting each set of training data, that is, the walking pattern of the user individual within the lighting functional area, generally the walking speed of the user individual;

[0059] Here, the walking posture of the user individual within the lighting functional area can also be recognized through a convolutional neural network to further judge the accuracy of the movement trajectory of the user individual. However, the walking posture may involve personal privacy, so the walking speed is preferentially collected here, which not only ensures the accuracy of the movement trajectory collection but also protects personal privacy;

[0060] The environmental data refers to the sound intensity value and light intensity value collected by the data acquisition module when collecting each set of training data. The environmental data refers to the sound intensity value and light intensity value collected by the data acquisition module. Therefore, a sound sensor and a light sensor need to be set within the lighting functional area. The data acquisition module collects the sound intensity value within the lighting functional area through the sound sensor, and the data acquisition module collects the light intensity value within the lighting functional area through the light sensor;

[0061] It should be noted that the installation quantity and installation position of the sound sensor and the light sensor are arbitrarily set by those skilled in the art within the lighting functional area, as long as it helps the data acquisition module accurately collect the sound intensity value and the light intensity value. The sound intensity value and the light intensity value can also adopt the average value or the average value after removing abnormal items. For example, the obviously too large sound intensity value (external accidental noise) and the obviously too large light intensity value (external accidental light source) are removed to ensure the accuracy of the collection of the sound intensity value and the light intensity value;

[0062] Label data refers to the lighting control strategy of the intelligent control module when collecting each set of training data. The lighting control strategy is the startup time of the lighting device, that is, the startup time point of the corresponding lighting device in the corresponding lighting function area. For example, if it is predicted that the startup time of the lighting device is 16:00:30, then the lighting device needs to be turned on at 16:00:30;

[0063] If the startup time is one second late, it will give people the feeling that they have already reached the corresponding lighting function area, and the lighting devices in this lighting function area only start up. This has a certain lag, and the actual experience is very bad. Suddenly turning on the lighting device when just entering the darkness will stimulate people's eyes and affect people's eyesight. However, if it is not turned on late, before people enter this lighting function area, the lighting devices in this lighting function area have just been turned on, giving people the visual impression that the lighting devices here are always on. This not only meets the energy conservation and environmental protection indicators but also improves the user experience;

[0064] The method for the data analysis module to train a machine learning model for predicting the lighting control strategy of the lighting function area is as follows:

[0065] Convert the feature data in the historical data into feature vectors, use the feature vectors as the input of the machine learning model, use the startup time predicted by the machine learning model for the corresponding feature data as the output, use the startup time in the label data corresponding to the feature data as the training target, and use minimizing the sum of prediction accuracies as the training target. The technical formula for prediction accuracy is: ai = (ui - wi) 2 , where ai is the prediction accuracy, ui is the predicted startup time corresponding to the i-th group of feature data, wi is the startup time in the label data corresponding to the i-th group of feature data. Train the machine learning model until the prediction accuracy reaches convergence and then stop training. It should be noted that the convergence criterion is adaptively set by those skilled in the art according to the specific model training situation. For example, set ai ≤ 0.01, that is, when the error between the predicted startup time and the actual startup time is less than 0.01, it is considered that the training of the lighting control strategy machine learning model is completed.

[0066] The intelligent control module is used to: control the lighting devices in the lighting functional area based on the lighting control strategy predicted by the machine learning model of the lighting control strategy. It should be noted that in a certain lighting functional area, the data acquisition module will collect the lighting device area data, the user's primary area trajectory data, and the environmental data in real time, and then convert the lighting device area data, the user's primary area trajectory data, and the environmental data into digital signals and transmit them to the data analysis module. After receiving the digital signals of the lighting device area data, the user's primary area trajectory data, and the environmental data, the data analysis module uses the trained machine learning model of the lighting control strategy to predict the startup time of the lighting devices in the lighting functional area. For example, if the predicted startup time of the lighting device is 16:00:30, then the data analysis module converts the startup time of the lighting device being 16:00:30 into a digital signal and transmits it to the intelligent control module. After receiving the digital signal that the startup time of the lighting device is 16:00:30, the intelligent control module can directly control the lighting devices in the lighting functional area to turn on at 16:00:30. The opening time of the lighting device is preset by those skilled in the art, or another machine learning model is trained simultaneously to predict the closing time of the lighting device to ensure that the user experience is not affected;

[0067] The machine learning model of the lighting control strategy for predicting the lighting functional area trained by the data analysis module can ensure that the lighting devices in the lighting functional area start at the accurate time point without delay, which not only ensures energy conservation and environmental protection, reduces energy consumption, but also ensures the personal experience of the user.

[0068] As a real-time mode of the present invention, such as Figure 1 and Figure 2As shown in the figure, a distance sensing module is provided in the lighting function area. The distance sensing module is used to collect the distance d between the user individual and the distance sensing module. It should be noted that the distance sensing module is an electronic component that uses non-contact measurement technology to detect the distance between the target object and the device in real time. Its core principle is based on the physical process of signal emission - reflection - reception and signal processing algorithms. Common implementation technologies include infrared ranging (IR), ultrasonic ranging, lidar (LiDAR), millimeter wave radar (mmWave), and vision ranging. Infrared ranging calculates the distance by emitting modulated infrared light and measuring the flight time (ToF) or intensity attenuation of the reflected light. It has low cost but is easily affected by ambient light interference and is suitable for short-distance scenarios (such as automatic induction lights). Ultrasonic ranging calculates the distance using the time difference of sound wave reflection. It has strong anti-light interference ability, but the speed of sound is affected by temperature and humidity and requires dynamic calibration. It is mostly used for medium and short-distance detection. Lidar generates 3D point clouds through the ToF or phase difference of high-precision laser beams, with millimeter-level accuracy and long ranging ability (up to 200 meters), but it has high cost and limited performance in complex environments (rain and fog). It is commonly used in autonomous driving and robot navigation. Millimeter wave radar realizes penetration detection by analyzing the Doppler effect and phase change of high-frequency electromagnetic waves (24 - 100 GHz), can identify targets behind partitions and adapt to bad weather, and is suitable for presence sensing in smart homes and industrial security. Vision ranging relies on binocular camera parallax or monocular deep learning models to estimate depth information and has the ability to identify targets, but it has high computing power requirements and depends on lighting conditions. In practical applications, the module needs to combine filtering algorithms (such as Kalman filtering to eliminate noise), environmental compensation (temperature and humidity calibration), and multi-sensor fusion (such as infrared + ultrasonic redundancy verification) to improve reliability. At the same time, it is necessary to balance accuracy, range, power consumption, and cost. Therefore, the specific model and type of the distance sensing module are adaptively determined by the technicians in this field according to the use of the lighting function area;

[0069] The distance sensing module only collects the distance between the user individual and the distance sensing module and does not involve other privacy of the user individual. Therefore, it is very safe and reliable, and the user individual also has no risk of leaking personal privacy.

[0070] Therefore, the data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the distance d between the user individual and the distance sensing module, as shown below:

[0071] The method for the data acquisition module to generate user primary area trajectory data or user secondary area trajectory data based on the distance d between the user individual and the distance sensing module is as follows:

[0072] The data acquisition module collects the distances d1, d2, ……, dn between the user individual and the distance sensing module at the same time intervals within the time period t. Specifically, the length of the time period t and the length of the time intervals are arbitrarily set by those skilled in the art, as long as it is ensured that the intelligent control module is not affected in controlling the lighting device to be turned on on time;

[0073] Generate a primary coordinate axis based on the distances d1, d1, ……, dn and the time period t. The horizontal axis of the primary coordinate axis is the time period t, and the vertical axis is the distance d. It should be noted that the primary coordinate axis generated based on the distance sequence d1, d1, ……, dn and the corresponding time periods t1, t1, ……, tn is a two-dimensional data visualization and analysis method with the time-distance relationship as the core. Its principle lies in revealing the dynamic law of distance change through spatio-temporal mapping. The horizontal axis is the time t and the vertical axis is the distance d. Each data point (ti, di) represents the real-time distance between the target object and the observation point (such as a distance sensor) at a specific moment. The essence of this coordinate system is the visualization of time series analysis, which can intuitively reflect the trend, periodicity or abnormal fluctuations of the distance changing over time;

[0074] Generate a primary movement trajectory based on the curve formed by the primary coordinate axis. The primary movement trajectory is the user's primary area trajectory data, that is, the curve on the generated primary coordinate axis is used as the primary movement trajectory, and the slope of the curve reflects the walking speed of the user individual;

[0075] The data acquisition module is based on the distance sensing module at the time period t , and collects the distances d , 1, d , 2, ……, d , m between the user individual and the distance sensing module at the same time intervals within the time period t; based on d , 1, d , 2, ……, d , m and the time period t , generate a secondary coordinate axis. The horizontal axis of the secondary coordinate axis is the time period t , and the vertical axis is the distance d , ; generate a secondary movement trajectory based on the curve formed by the secondary coordinate axis. The secondary movement trajectory is the user's secondary area trajectory data, and the generation method of the secondary area trajectory data is exactly the same as that of the primary area trajectory data, which will not be elaborated here.

[0076] As a real-time mode of the present invention, such as Figure 1 and Figure 2As shown, an image acquisition module is provided in the lighting function area. The image acquisition module is used to acquire image data of a user individual in the lighting function area. It should be noted that the image acquisition module is a hardware system that converts visible or non-visible light (such as infrared, X-ray) information in the real world into digital signals through optical imaging technology. Its core principle is based on the collaborative work of photoelectric conversion and digital signal processing, covering key components such as optical lenses, image sensors, analog-to-digital converters (ADCs), and processing units. The optical lens is responsible for focusing the light of the target scene onto the surface of the image sensor. The sensor (such as CCD or CMOS) converts the optical signal into an analog electrical signal through an array of photodiodes, where each pixel point corresponds to the charge amount of the light intensity. CMOS sensors are widely used in consumer electronics and industrial fields due to their advantages of high integration, low power consumption, and moderate cost, while CCDs dominate high-end scientific imaging with high dynamic range and low noise characteristics. The ADC quantizes the analog signal pixel by pixel to generate the original digital image (RAW data), and preprocesses it through an image signal processor (ISP), including demosaicing (Bayer interpolation), noise suppression (such as Gaussian filtering or non-local means denoising), color correction (white balance, gamma correction), and sharpening enhancement to optimize the image quality. The module can integrate intelligent algorithms (such as edge computing units or AI acceleration chips). The image acquisition module has a certain possibility of leaking the personal privacy of the user individual compared to the distance acquisition module. Therefore, here the image acquisition module directly acquires the grayscale pictures of the user individual to minimize the possibility of leaking the personal privacy of the user individual;

[0077] The data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the image data in the lighting function area.

[0078] The method by which the data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the image data in the lighting function area is as follows:

[0079] The data acquisition module acquires the image data P1, P2, ……, Pn collected by the image acquisition module at the same time intervals within the time period T by the image acquisition module;

[0080] Based on the image data P1, P2, ……, Pn, a convolutional neural network is used to identify a primary movement trajectory of a user individual within the lighting functional area. A primary movement trajectory of a user individual within the lighting functional area is user primary area trajectory data. It should be noted that a convolutional neural network (CNN) is a deep learning model specifically designed for processing grid-structured data (such as images and videos). Its core principle is to automatically learn the spatial or temporal patterns of input data through local perception, weight sharing, and hierarchical feature extraction. A CNN consists of a convolutional layer, a pooling layer, an activation function, and a fully connected layer: The convolutional layer slides a learnable filter (convolution kernel) over the input data to calculate the feature map of the local area. Through the stacking of multi-channel convolution kernels, it captures low-level features such as edges and textures to high-level features such as semantic objects in a layer-by-layer abstraction. Weight sharing significantly reduces the number of parameters and enhances translational invariance (such as recognizing an object regardless of its location in the image). The pooling layer (such as max pooling, average pooling) downsamples the feature map, compresses the data dimension, and enhances robustness (such as resistance to minor deformations). The activation function (such as ReLU, Sigmoid) introduces non-linearity, enabling the network to fit complex functions. The fully connected layer then integrates global features for classification or regression output. During the training process, the loss function (such as cross-entropy, mean squared error) is optimized through the backpropagation algorithm, and the convolution kernel parameters are updated in combination with gradient descent (such as the Adam optimizer) to gradually improve the feature representation and task performance. The key technological breakthroughs of CNN include residual connections (ResNet solves the problem of gradient disappearance in deep networks), batch normalization (accelerates training and improves generalization), dilated convolution (expands the receptive field), and attention mechanisms (focus on key areas). Through the convolutional neural network, the secondary movement trajectory of the user individual within the lighting functional area can be directly identified;

[0081] The data acquisition module is based on the image acquisition module at time T , to obtain the image data P , 1, P , 2, ……, P , m collected by the image acquisition module at the same time interval within;

[0082] Based on the image data P , 1, P , 2, ……, P , m, a convolutional neural network is used to identify the secondary movement trajectory of the user individual within the lighting functional area. The secondary movement trajectory of the user individual within the lighting functional area is user secondary area trajectory data. The generation method of the secondary area trajectory data is exactly the same as that of the primary area trajectory data and will not be elaborated here.

[0083] As a real-time mode of the present invention, such as Figure 1 andFigure 2 As shown in Figure 2 , the method for the intelligent control module to optimize the lighting control strategy based on the user's secondary area trajectory data is as follows:

[0084] The data acquisition module generates the speed variance of the user's individual movement over time based on the secondary movement trajectory, and preset a variance threshold. For example, if the lighting equipment in the lighting function area is started with a delay, it will inevitably cause the user's individual to stay when entering this lighting function area (wait for the lighting equipment to turn on when entering the dark area), which affects the actual experience of the user's individual. Therefore, the speed variance of the user's individual movement over time is generated through the secondary movement trajectory. If the movement speed of the user's individual does not change significantly, it indicates that the start time of the lighting equipment is good. If the speed variance of the user's individual movement is too large, it means that there is a delay, that is, the start time of the lighting equipment is late;

[0085] If the speed variance ≤ the variance threshold, the lighting control strategy does not need to be optimized;

[0086] If the speed variance > the variance threshold, the lighting control strategy is optimized through the speed variance. As can be seen from the above introduction, the larger the speed variance of the user's individual movement over time, the later the start time of the lighting equipment. Therefore, the speed variance of the user's individual movement over time is in a proportional relationship with the delay time of the lighting equipment start time. Those skilled in the art can use a computer to quantitatively obtain the proportional relationship between the speed variance of the user's individual movement over time and the delay time of the lighting equipment start time, and then optimize the lighting control strategy according to the obtained quantitative formula, that is, the start time of the lighting equipment.

[0087] The above has described a specific embodiment of the present invention in detail, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the scope covered by the patent of the present invention.

Claims

1. An adaptive adjustment system for indoor intelligent lighting equipment, characterized in that: include: A data collection module, which performs regional marking of indoor smart lighting equipment based on the indoor lighting function area to obtain lighting equipment regional data. The data collection module collects user's regional trajectory data and environmental data of the lighting equipment regional data; A data analysis module, which trains a lighting control strategy machine learning model for predicting lighting function areas based on lighting equipment area data, user primary area trajectory data and environmental data; An intelligent control module controls lighting equipment within a lighting function area based on a lighting control strategy predicted by a lighting control strategy machine learning model; The smart control module collects user secondary area trajectory data in the lighting function area controlled by the lighting control strategy through the data acquisition module, and the smart control module optimizes the lighting control strategy based on the user secondary area trajectory data.

2. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 1, characterized in that: The data acquisition module uses Ri to mark the lighting function area, Re∈[R1, R2, ..., Ri]; The data acquisition module uses Lj to mark the lighting equipment in the lighting function area, Lj∈[L1, L2, ..., Lj]; The lighting equipment area data is Ri of the lighting function area.

3. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 2, characterized in that: A distance sensing module is provided in the lighting functional area, and the distance sensing module is used to collect the distance d between the user individual and the distance sensing module; The data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the distance d between the individual user and the distance sensing module.

4. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 3, characterized in that: The method in which the data acquisition module generates the user's primary area trajectory data or the user's secondary area trajectory data based on the distance d between the user individual and the distance sensing module is: The data collection module collects the distances d1, d2, ..., dn between the individual user and the distance sensing module at the same time interval within time t based on the distance sensing module; Generate a primary coordinate axis based on the distances d1, d1, ..., dn and the time t, where the horizontal axis of the primary coordinate axis is the time t and the vertical axis is the distance d; Generate a movement trajectory based on the curve formed by the primary coordinate axis, and the primary movement trajectory is the user's regional trajectory data; The data acquisition module is based on the distance sensing module at time t , The distance d between the user and the distance sensing module is collected at the same time interval , 1.d , 2. ..., d , m; Based on d , 1.d , 2. ..., d , m and time t , Generate a secondary coordinate axis, the horizontal axis of which is time t , , the vertical axis is the distance d , ; A secondary movement trajectory is generated based on the curve formed by the secondary coordinate axes, and the secondary movement trajectory is the user's secondary area trajectory data.

5. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 2, characterized in that: An image acquisition module is provided in the lighting functional area, and the image acquisition module is used to collect image data of individual users in the lighting functional area; The data acquisition module generates user primary area trajectory data or user secondary area trajectory data based on the image data in the lighting function area.

6. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 5, characterized in that: The method for the data acquisition module to generate the user's primary area trajectory data or the user's secondary area trajectory data based on the image data in the lighting function area is: The data acquisition module acquires the image data P1, P2, ..., Pn acquired by the image acquisition module at the same time interval within the time T based on the image acquisition module; Based on the image data P1, P2, ..., Pn, a convolutional neural network is used to identify a movement trajectory of an individual user in the lighting function area, and the movement trajectory of an individual user in the lighting function area is the user's primary area trajectory data; The data acquisition module is based on the image acquisition module at time T , The image data P collected by the image acquisition module is obtained at the same time interval , 1. P , 2. ..., P , m; Based on the image data P , 1. P , 2. ..., P , m uses a convolutional neural network to identify the secondary movement trajectory of individual users in the lighting function area, and the secondary movement trajectory of individual users in the lighting function area is the user's secondary area trajectory data.

7. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 4 or 6, characterized in that: The data analysis module collects historical training data in advance, and trains a lighting control strategy machine learning model for predicting lighting function areas based on the historical training data.

8. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 7, characterized in that: The historical training data includes N groups of training data, where N is a positive integer, and each group of training data includes feature data and label data, and the feature data includes lighting device area data, user one-time area trajectory data, and environment data, wherein: The lighting equipment area data refers to the lighting function area Ri determined by the data acquisition module when collecting each set of training data; The user's one-time regional trajectory data refers to the movement trajectory of individual users in the lighting function area when collecting each set of training data; Environmental data refers to the sound decibel value and light intensity value collected by the data acquisition module when collecting each set of training data; Label data refers to the lighting control strategy of the intelligent control module when collecting each set of training data. The lighting control strategy is the start-up time of the lighting equipment.

9. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 8, characterized in that: The method for the data analysis module to train a lighting control strategy machine learning model for predicting lighting function areas is: The feature data in the historical data is converted into a feature vector, and the feature vector is used as the input of the machine learning model. The machine learning model uses the startup time predicted by the corresponding feature data as the output, and the startup time in the label data corresponding to the feature data as the training target. The training target is to minimize the sum of the prediction accuracy. The technical formula for prediction accuracy is: ai=(ui-wi) 2 , where ai is the prediction accuracy, ui is the predicted start time corresponding to the i-th set of feature data, and wi is the start time in the label data corresponding to the i-th set of feature data. The machine learning model is trained until the prediction accuracy reaches convergence and the training is stopped.

10. The adaptive adjustment system for indoor intelligent lighting equipment according to claim 4 or 6, characterized in that: The method for the intelligent control module to optimize the lighting control strategy based on the user's secondary area trajectory data is: The data acquisition module generates the speed variance of individual users moving over time based on the secondary movement trajectory and presets the variance threshold; If the speed variance is ≤ the variance threshold, the lighting control strategy does not need to be optimized; If the speed variance > the variance threshold, the lighting control strategy is optimized by the speed variance.

Images