Lighting device for heat supply and energy conservation and lighting control method

Abstract

The invention discloses a lighting device for heat supply and energy conservation and a lighting control method. The lighting device for heat supply and energy conservation comprises a sensor integration module, an image recognition module, a lighting control module, a lighting driving module, an upper computer module, a communication module and a fault monitoring module. The illumination control method comprises the following steps: step 1, acquiring environment data information through the sensor integration module and the image recognition module; step 2, the lighting control module controls a lighting device through a self-adaptive PID control model; and step 3, the illumination driving module automatically adjusts the illumination intensity and angle of the energy-saving LED lamp according to the received control signal. Resource waste is greatly reduced, and the illumination control capability is improved.

Description

Technical Field

[0001] The present invention is based on the field of lighting control, and more specifically relates to a lighting device for energy-saving heating and a lighting control method. Background Art

[0002] With the acceleration of urbanization and the continuous improvement of people's requirements for quality of life, the problem of building energy consumption has become increasingly prominent. As the main components of building energy consumption, heating and lighting account for a high proportion of energy consumption and have great energy-saving potential.

[0003] Traditional lighting control systems have the following shortcomings: Lack of intelligent management means makes it difficult to make dynamic adjustments based on the actual situation of people's activities in the building, resulting in serious energy waste; Traditional lighting equipment, incandescent lamps and fluorescent lamps, although low in cost, have low energy efficiency and generate a lot of heat during the lighting process, resulting in energy waste; The lighting brightness cannot be automatically adjusted according to the ambient light, and the control ability is low, so the energy saving effect cannot be maximized.

[0004] Based on the deficiencies of the above-mentioned technology, the existing patented lighting device and method of using the lighting device, patent number CN201811375467.9, proposes to adjust the brightness of the lighting equipment in real time according to the surrounding environment; but no intelligent management is performed, and the lighting device is dynamically adjusted according to the activities of personnel. The intelligent analysis capability of the surrounding environment is poor, and it is difficult to achieve adaptive perception of the environment.

[0005] Therefore, the present invention proposes a lighting device and a lighting control method for energy-saving heating, which can realize intelligent lighting control and reduce resource waste. Summary of the invention

[0006] In view of the shortcomings of the above-mentioned technology, the traditional lighting control system cannot be managed intelligently. The present invention discloses a lighting device for energy-saving heating and a lighting control method that can realize intelligent management, dynamic lighting brightness and lighting angle, and can dynamically perceive the lighting environment.

[0007] A heating energy-saving lighting device, comprising: A sensor integrated module, the sensor integrated module includes a photoresistor circuit, an infrared sensor circuit and a temperature sensor circuit; the photoresistor circuit is used to monitor the natural light intensity indoors and outdoors; the infrared sensor circuit monitors the thermal radiation of the human body, and when the infrared sensor circuit detects human activity, it provides a trigger signal for the image recognition module; the temperature sensor circuit is used to monitor the temperature of the environment, and when the indoor temperature is greater than 30°C, the lighting control module controls the lighting device to reduce the brightness and reduce the heat load; The image recognition module locates the position of the personnel in real time through a camera with a resolution greater than 1080P; performs image feature extraction, data analysis and tracking of active objects in the lighting area through a regional tracking neural network model, wherein the regional tracking neural network model includes a regional feature extraction module, a personnel detection module, a regional tracking module, an activity recognition module and a data analysis and fusion module; the regional tracking neural network model performs feature extraction and regional tracking based on an adaptive convolutional neural network; the signal output end of the regional feature extraction module is connected to the signal input end of the personnel detection module, the signal output end of the personnel detection module is connected to the signal input end of the regional tracking module, the signal output end of the regional tracking module is connected to the signal input end of the activity recognition module, and the signal output end of the activity recognition module is connected to the signal input end of the data analysis and fusion module; The lighting control module adjusts the lighting brightness according to the data information of the sensor integration module and the image recognition module, provides local lighting according to the position of the personnel, or adjusts the lighting intensity for a specific area; A lighting drive module uses energy-saving LED lamps for lighting, and adjusts the switch, angle and light intensity of the lighting device according to the instructions of the lighting control module; The host computer module displays the operating status of the lighting device through the display screen, receives user instructions and adjusts the control strategy, and performs remote control and optimization settings as needed; A communication module, which uses a wireless communication method ZigBee to realize data transmission between the lighting control module and the host computer module; Fault monitoring module, which monitors the operating status of lighting devices in real time through isolated forests; The signal output end of the sensor integrated module is connected to the signal input end of the image recognition module, the signal output end of the sensor integrated module is connected to the signal input end of the lighting control module, the signal output end of the image recognition module is connected to the signal input end of the lighting control module, the signal output end of the lighting control module is connected to the signal input end of the lighting driving module, the lighting control module is bidirectionally connected to the communication module, the communication module is bidirectionally connected to the host computer module, and the fault monitoring module of the lighting control module is bidirectionally connected. As a further embodiment of the present invention, the energy-saving LED lamp includes a focusing module, a reflector module and an angle rotation module; the focusing module is used to control and concentrate the light beam of the energy-saving LED lamp, and concentrate the light on a small area or a specific target; the reflector module effectively reflects and concentrates the light emitted by the lamp to the area that needs to be illuminated, reducing the loss and waste of light; the angle rotation module adjusts the illumination angle according to the data information of the image recognition module to achieve directional lighting and dynamic lighting.

[0008] As a further embodiment of the present invention, the working principle of the area tracking neural network model is: Collect video image information in real time through a camera with a resolution greater than 1080P installed in the heating area; pre-process the collected images, the pre-processing at least including image cropping, normalization and denoising, to improve image quality and reduce the amount of calculation; Utilizing the regional feature extraction module to extract feature information in an image or video, and performing data analysis to generate a feature map; The image data features are analyzed according to the feature graph, and the personnel detection module, the area tracking module and the activity recognition module detect and locate the target in the image or video; the personnel detection module uses a classifier to detect the target, and the classifier is used to determine whether there is a person in the image; the area tracking module uses a classifier to locate the target and determine the specific position of the person; The data analysis and fusion module processes and analyzes the result data information from the personnel detection module, the area tracking module and the activity recognition module, generates a final personnel tracking report, and feeds the tracking report back to the lighting control module. The lighting control module drives the lighting driving module to adjust the illumination intensity and projection angle of the lighting device according to the movement trajectory and number of personnel to illuminate the dynamic area. As a further embodiment of the present invention, the output of the activation function in the area feature extraction module is: In formula (1), is the number of convolutional unit layers, For the The activation output of the layer, is the activation function, is the activation output of the previous layer, For the The convolution kernel of the layer, is the bias term, is the offset control factor, m is the convolution random coefficient, For the The linear combination output of the layer; the formula of the activation function is: In formula (2), The output probability result of the personnel detection module is: In formula (3), x is whether human activity is detected, j is the category, The probability of class j is is the original output of the classifier for category j, N is the total number of categories, is the output of the classifier for the nth category. As a further embodiment of the present invention, the fault monitoring module includes a data processing unit and a collection unit, an analysis and diagnosis unit, a communication unit and an alarm unit connected to the data processing unit; the alarm unit includes a first alarm unit and a second alarm unit, the first alarm unit is located on the lighting device, and the second alarm unit is located in the host computer module. When the analysis and diagnosis unit detects an abnormality, the first alarm unit performs an audible and visual alarm, and the second alarm unit performs an alarm through the display unit in the host computer module; the collection unit collects environmental parameters in real time through the sensor module; the analysis and diagnosis unit determines whether the light intensity exceeds a threshold and whether the angle selection module works normally, and the analysis and diagnosis unit uses an isolation forest to perform abnormality diagnosis. The working principle of the isolation forest is to extract 256 sub-sample sets from the original data set, and After each sub-sample set, an isolated tree is formed. A feature is randomly selected, and a cutting point q is randomly selected within the value range of the feature to generate a hyperplane. Samples with a value less than q are divided into the left subtree, and samples with a value greater than or equal to q are divided into the right subtree. The above steps are performed recursively until only one sample is left. The anomaly score is the ratio of the average length of the data point to the average path length of the normal sample in the tree. If the anomaly score exceeds the threshold, it is judged as an abnormal point, and the first alarm unit and the second alarm unit start to alarm. The signal output end of the acquisition unit is connected to the signal input end of the data processing unit, and the signal output end of the data processing unit is connected to the signal input end of the analysis and diagnosis unit. The signal output end of the analysis and diagnosis unit is connected to the signal input end of the communication unit, and the signal output end of the communication unit is connected to the signal input end of the alarm unit. As a further embodiment of the present invention, a lighting control method for a heating energy-saving lighting device comprises the following steps: Step 1: Collecting environmental data information through the sensor integration module and the image recognition module; The environmental data information at least includes natural light intensity, presence or absence of human activities, human activity conditions and environmental temperature; the collected data is preprocessed, including at least data cleaning, conversion and standardization operations to eliminate noise and abnormal values ​​in the data; the sensor integration module and the image recognition module transmit the preprocessed data to the lighting control module; Step 2: The lighting control module controls the lighting device through an adaptive PID control model; The adaptive PID control model includes a rule generation module, a rule evaluation module, an adaptive parameter adjustment module, a PID control module and a feedback loop module; the signal output end of the rule generation module is connected to the signal input end of the rule evaluation module, the signal output end of the rule evaluation module is connected to the signal input end of the adaptive parameter adjustment module, the signal output end of the adaptive parameter adjustment module is connected to the signal input end of the PID control module, the signal output end of the PID control module is connected to the signal input end of the feedback loop module, and the signal output end of the feedback loop module is connected to the signal input end of the rule generation module; Step 3: The lighting driving module automatically adjusts the illumination intensity and angle of the energy-saving LED lamp according to the received control signal to meet the lighting requirements and achieve energy-saving effects. As a further embodiment of the present invention, the working principle of the adaptive PID control model is as follows: (1) The actual state of the lighting device is obtained through the sensor integration module and the image recognition module, and the actual state includes at least light intensity, temperature and whether there is human activity, and it is compared with the expected state to obtain an error; (2) The control quantity is calculated by the PID control module according to the error. The calculation formula of the control quantity is: In formula (4), is the system error, is the output light brightness, is the proportionality coefficient, is the integration time, d is the time, c is the compensation coefficient, k is the number of sampling times, is the proportionality coefficient, is the integration coefficient, is the differential coefficient, is the sampling number, r is the number of people; (3) The adaptive adjustment module adjusts the lighting fixture online according to the changes in the lighting fixture and the uncertainty of the external environment. (4) Adjust the control amount As an output signal, it is transmitted to the lighting driving module; the lighting driving module adjusts the light intensity and lighting angle of the energy-saving LED lamp according to the received control signal; the actual state is continuously detected by the sensor integration module and the image recognition module, and it is fed back to the lighting control module to form a closed-loop control. As a further embodiment of the present invention, the rule generation module generates adjustment rules for PID control parameters according to the operating characteristics, energy-saving goals and external environmental conditions of the lighting device; the rule evaluation module compares the calculated performance index with the pre-set expected performance through a threshold comparison algorithm to determine whether the current PID parameter adjustment rule meets the requirements; the adaptive parameter adjustment module: dynamically adjusts the parameters of the PID controller according to the output of the rule evaluation module and the real-time feedback of the operating state of the lighting device The PID control module calculates a control signal based on the error between the set target light intensity and the actual light intensity to adjust the brightness of the lighting device; the feedback loop module feeds back the actual light intensity of the lighting device to the PID control module to form a closed-loop control to ensure that the lighting device can accurately track the target light intensity.

[0009] Positive beneficial effects A lighting device and lighting control method for energy saving in heating uses a regional tracking neural network model to track active objects in the lighting area, and transmits the data results to an adaptive PID control model, so that the heating and lighting systems can dynamically adjust the output according to the actual human activity, thereby avoiding energy waste in non-essential areas or time periods; through accurate tracking of active objects in the lighting area, the system can identify which areas are densely populated and which areas are sparsely populated. In densely populated areas, the system will increase the output of heating and lighting to ensure that comfort and lighting needs are met; while in sparsely populated areas, the output will be reduced accordingly to achieve the purpose of energy saving; the dynamic adjustment mechanism helps to optimize the allocation of energy resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which: Figure 1 This is a schematic diagram of the overall structure of a heating energy-saving lighting device of the present invention; Figure 2 A specific flow chart of a lighting control method for a heating energy-saving lighting device according to the present invention; Figure 3This is a working principle diagram of a regional tracking neural network model of a heating energy-saving lighting device of the present invention; Figure 4 It is a working principle diagram of an adaptive PID control model of a lighting control method for a lighting device for energy-saving heating according to the present invention; Figure 5 This is a working principle diagram of a PID control module of a lighting control method for a lighting device for energy-saving heating according to the present invention. DETAILED DESCRIPTION

[0011] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0012] like Figure 1 like Figure 5 As shown, a lighting device for heating and energy saving comprises: A sensor integrated module, the sensor integrated module includes a photoresistor circuit, an infrared sensor circuit and a temperature sensor circuit; the photoresistor circuit is used to monitor the natural light intensity indoors and outdoors; the infrared sensor circuit monitors the thermal radiation of the human body, and when the infrared sensor circuit detects human activity, it provides a trigger signal for the image recognition module; the temperature sensor circuit is used to monitor the temperature of the environment, and when the indoor temperature is greater than 30°C, the lighting control module controls the lighting device to reduce the brightness and reduce the heat load; In the above scheme, after the sensor integrated module is powered on, each circuit performs an initialization self-test, including signal calibration and threshold preset of the photoresistor, infrared sensor, and temperature sensor. The current natural light intensity value is collected in real time through the photoresistor circuit as the reference parameter for subsequent lighting adjustment; the temperature sensor records the current ambient temperature for subsequent trigger condition judgment. The ambient light intensity (such as day / night state) is sensed by the change in the resistance value of the photoresistor, and the analog signal is converted into a voltage value and transmitted to the main control module. According to the preset light intensity threshold (such as 500lux), when the natural light intensity is detected to be lower than the threshold, the lighting control module is triggered to turn on the basic lighting. The pyroelectric infrared (PIR) sensor is used to focus the infrared energy radiated by the human body through the Fresnel lens (wavelength is about 10μm). When the infrared radiation change caused by human movement is detected: a high-level trigger signal is output to the image recognition module to start video monitoring or abnormality detection; at the same time, a personnel presence signal is sent to the lighting control module to maintain or enhance the current lighting brightness. A digital temperature sensor (such as DS18B20) is used to collect the ambient temperature in real time. When the detection value exceeds 30℃: a temperature over-limit signal is sent to the lighting control module; Trigger the dynamic dimming strategy of the lighting device (such as reducing the brightness by 5% for every 1℃ increase). When the natural light intensity is ≥800lux, the artificial lighting is automatically turned off; when the light intensity is between 300800lux, the LED brightness is linearly adjusted according to the photoresistor data. When the infrared sensor detects human activity, the lighting brightness is temporarily increased to the preset maximum value (such as 100%), and the original brightness is restored after 5 minutes of activity cessation. When the temperature is greater than 30℃, the lighting brightness is forced to be reduced to less than 50%, and PWM dimming is used to reduce heat. When the infrared sensor detects human activity, the image recognition module is synchronously triggered: the main control module stores data such as light intensity, temperature, and human activity frequency according to timestamps to generate environmental parameter change curves; optimize lighting control strategies (such as statistical light intensity distribution during peak hours). Upload real-time data to the monitoring center through RS485 or WiFi modules, supporting remote viewing and parameter adjustment. More accurate environmental perception is achieved by integrating three-dimensional data of light, heat, and human activity. A smooth transition of lighting brightness is achieved through fuzzy control algorithms to avoid flickering and abrupt changes. Using the sleep wake-up mechanism, the infrared sensor enters low-power mode when there is no activity, and the overall power consumption is less than 2W.

[0013] The image recognition module locates the position of the personnel in real time through a camera with a resolution greater than 1080P; performs image feature extraction, data analysis and tracking of active objects in the lighting area through a regional tracking neural network model, wherein the regional tracking neural network model includes a regional feature extraction module, a personnel detection module, a regional tracking module, an activity recognition module and a data analysis and fusion module; the regional tracking neural network model performs feature extraction and regional tracking based on an adaptive convolutional neural network; the signal output end of the regional feature extraction module is connected to the signal input end of the personnel detection module, the signal output end of the personnel detection module is connected to the signal input end of the regional tracking module, the signal output end of the regional tracking module is connected to the signal input end of the activity recognition module, and the signal output end of the activity recognition module is connected to the signal input end of the data analysis and fusion module; The lighting control module adjusts the lighting brightness according to the data information of the sensor integration module and the image recognition module, provides local lighting according to the position of the personnel, or adjusts the lighting intensity for a specific area; A lighting drive module uses energy-saving LED lamps for lighting, and adjusts the switch, angle and light intensity of the lighting device according to the instructions of the lighting control module; The host computer module displays the operating status of the lighting device through the display screen, receives user instructions and adjusts the control strategy, and performs remote control and optimization settings as needed; A communication module, which uses a wireless communication method ZigBee to realize data transmission between the lighting control module and the host computer module; Fault monitoring module, which monitors the operating status of lighting devices in real time through isolated forests; The signal output end of the sensor integrated module is connected to the signal input end of the image recognition module, the signal output end of the sensor integrated module is connected to the signal input end of the lighting control module, the signal output end of the image recognition module is connected to the signal input end of the lighting control module, the signal output end of the lighting control module is connected to the signal input end of the lighting driving module, the lighting control module is bidirectionally connected to the communication module, the communication module is bidirectionally connected to the host computer module, and the fault monitoring module of the lighting control module is bidirectionally connected.

[0014] In the above embodiment, the regional feature extraction module, the personnel detection module, the regional tracking module, the activity recognition module and the data analysis and fusion module are described in detail below: The regional feature extraction module is the basis of the entire regional tracking neural network model. Its core task is to extract representative feature information from real-time images captured by high-resolution (greater than 1080P) cameras. These features can be local information such as texture, color, and edges of the image, or some more advanced semantic features. Based on the adaptive convolutional neural network, this module can automatically learn and adapt to image features in different scenarios, has strong generalization ability and robustness, and provides a key feature basis for subsequent personnel detection and tracking. In specific work, the input high-resolution image is preprocessed, including image normalization, scaling, cropping, etc., to improve the quality and consistency of the image, which is convenient for subsequent feature extraction. The convolution layer in the adaptive convolutional neural network is used to perform convolution operations on the preprocessed image. The convolution kernel slides on the image to extract local features at different positions. By stacking multiple convolutional layers, more advanced and abstract features can be gradually extracted. The feature map obtained by the convolution operation is subjected to nonlinear activation processing, such as using the ReLU activation function, to enhance the expressiveness of the feature. At the same time, the feature map is downsampled through the pooling layer to reduce the size of the feature map and reduce the computational complexity. Finally, the final feature map is taken as the output of the module and passed to the personnel detection module.

[0015] The main purpose of the person detection module is to accurately detect the position and bounding box of the person in the feature map output by the regional feature extraction module. This module uses the powerful classification and positioning capabilities of the adaptive convolutional neural network to analyze the feature map through the trained model to determine whether there is a person in the image and determine the specific position and size of the person. Its technical essence lies in the ability to accurately identify the person target in a complex background environment, providing a basis for subsequent tracking and activity recognition. By receiving the feature map output by the regional feature extraction module as input. Use the classifier in the adaptive convolutional neural network to classify the feature map to determine whether each area contains a person target. The classifier is usually implemented using a fully connected layer or a convolutional layer, and the feature differences between people and non-people are learned through training. For the area classified as a person, a regressor is used to predict the position and size of the person's bounding box. The regressor outputs the predicted bounding box coordinates by learning the mapping relationship between the feature map and the true bounding box. The predicted bounding box is post-processed, such as non-maximum suppression (NMS), to remove redundant bounding boxes with high overlap and improve the accuracy of the detection result. The final detected person bounding box information is passed to the regional tracking module as the output of the module.

[0016] The core task of the region tracking module is to track the detected human targets in continuous image frames and maintain the identity consistency of the targets. Based on the adaptive convolutional neural network, the module can learn the appearance features and motion patterns of human targets, and accurately predict the position of the person in the subsequent frames by matching the features and analyzing the motion between adjacent frames. Its technical essence lies in the ability to track human targets in real time and accurately in complex scenes, and provide continuous target information for activity recognition. In the first frame of the image, the feature template and state information of the human target are initialized according to the bounding box information output by the human detection module. In the subsequent image frames, the feature map of the current frame is extracted and matched with the feature template of the target. By calculating the similarity between the features, the area that best matches the target is found. According to the motion trajectory and speed of the target in the previous frame, the possible position of the target in the current frame is predicted. Combined with the feature matching results, the accurate position of the target is further determined. According to the actual position and features of the target in the current frame, the feature template and state information of the target are updated to adapt to the appearance changes and motion changes of the target. The position information of the tracked human target is passed to the activity recognition module as the output of the module. The main function of the activity recognition module is to identify the activity type of the person according to the continuous position information of the human target provided by the region tracking module. This module analyzes the characteristics of the movement trajectory and posture changes of the personnel, and combines the preset activity model to determine the activities that the personnel are performing, such as walking, standing, sitting, etc. Its technical essence lies in the ability to extract meaningful information from the dynamic behavior of the personnel and realize intelligent recognition of personnel activities. The continuous position information of the personnel target output by the area tracking module is analyzed to extract the movement trajectory of the personnel. The movement characteristics of the personnel are understood by calculating the parameters such as the length, speed, and direction of the trajectory. The posture information of the personnel, such as joint angles and limb movements, is extracted from the image using human posture estimation technology. Posture estimation can help identify the activity type of the personnel more accurately. The feature information obtained by trajectory analysis and posture estimation is input into the preset activity classification model to classify the activity type. The activity classification model can be a classifier based on machine learning, such as support vector machine (SVM), decision tree, etc., or a neural network model based on deep learning. The identified personnel activity type information is passed to the data analysis and fusion module as the output of the module.

[0017] The main task of the data analysis and fusion module is to comprehensively analyze and fuse the personnel activity information output by the activity recognition module, and combine it with other relevant data, such as environmental information, time information, etc., to provide users with more comprehensive and valuable decision support. This module can mine and analyze a large amount of multi-source data, discover the associations and laws between the data, and improve the intelligence level and decision-making ability of the system. Receive the personnel activity information output by the activity recognition module, as well as other relevant data, such as temperature, humidity, light, etc., collected by environmental sensors, time information, etc. Clean and preprocess the input data, remove abnormal data such as noise and missing values, and normalize and standardize different types of data for subsequent analysis and fusion. Use data analysis techniques, such as statistical analysis, machine learning, data mining, etc., to analyze the preprocessed data. By analyzing the frequency, time distribution, and correlation of personnel activities with environmental factors, the associations and laws between the data are discovered. In a specific embodiment, data from different sources are fused to comprehensively consider the impact of various factors on personnel activities. A rule-based fusion method, a model-based fusion method, or a machine learning-based fusion method can be used to fuse multi-source data into a unified information representation. The results of analysis and fusion are presented to users in a visual way, such as generating reports and charts, to provide decision support for users. At the same time, the results are fed back to other modules of the system, such as the lighting control module, to achieve intelligent lighting adjustment.

[0018] In the present invention, the energy-saving LED lamp includes a focusing module, a reflector module and an angle rotation module; the focusing module is used to control and focus the light beam of the energy-saving LED lamp, focusing the light on a small area or a specific target; the reflector module effectively reflects and focuses the light emitted by the lamp to the area that needs to be illuminated, reducing the loss and waste of light; the angle rotation module adjusts the irradiation angle according to the data information of the image recognition module to achieve directional lighting and dynamic lighting. The following is the specific working process of each module of the energy-saving LED lamp: When the focusing module is working, the LED lamp beads of the energy-saving LED lamp are powered on and emit light, generating initial light. These rays diverge in all directions and have a wide distribution range. The focusing module is generally composed of optical components such as optical lenses. When the light is directed to the focusing module, the optical lens uses its refraction principle to control and concentrate the light. For example, a convex lens refracts the light toward the central axis, converging the originally scattered light. According to actual lighting needs, the focusing module can concentrate the light on a small area or a specific target. For example, in museum lighting, the light is focused on the exhibits to highlight the details of the exhibits, while reducing unnecessary lighting of the surrounding environment to achieve energy saving.

[0019] Working process of reflector module The light emitted by the lamp propagates in space, and part of the light will be directed to the reflector module. The reflector is usually made of materials with high reflectivity, such as mirror aluminum. When the light hits the inner surface of the reflector, it will be reflected according to the law of light reflection. The shape of the reflector is carefully designed, and common shapes include parabolic and elliptical shapes. Taking the parabolic reflector as an example, the light emitted by the LED lamp beads at the focus of the parabola will form a nearly parallel beam after reflection, thereby effectively guiding the light to the area that needs to be illuminated. Through the reflection of the reflector, the light that was originally scattered in all directions is concentrated in a specific direction and area, reducing the loss and waste of light in other unnecessary directions and improving the utilization efficiency of light.

[0020] When the angle rotation module is working, the angle rotation module establishes a data connection with the image recognition module. The image recognition module will identify and analyze information such as the location of people in the environment and the activity area, and pass this data information to the angle rotation module. After receiving the data information from the image recognition module, the angle rotation module analyzes it. For example, if the image recognition module detects that a person is active in a specific area, the angle rotation module will calculate the illumination angle that needs to be adjusted for the lamp based on the location information of the area. In a specific embodiment, the angle rotation module is composed of a motor, a transmission mechanism, etc. According to the calculated illumination angle, the angle rotation module controls the rotation of the motor, and drives the entire or partial structure of the lamp to rotate through the transmission mechanism, thereby adjusting the illumination angle.

[0021] As people move or their activity areas change, the image recognition module will continuously update data information and pass it to the angle rotation module. The angle rotation module adjusts the lighting angle of the lamp in real time according to the new data information to achieve directional lighting and dynamic lighting, ensuring that there is always sufficient light in the area where people are active, while avoiding unnecessary lighting in unmanned areas, further saving energy.

[0022] In the present invention, the working principle of the area tracking neural network model is: Collect video image information in real time through a camera with a resolution greater than 1080P installed in the heating area; pre-process the collected images, the pre-processing at least including image cropping, normalization and denoising, to improve image quality and reduce the amount of calculation; Utilizing the regional feature extraction module to extract feature information in an image or video, and performing data analysis to generate a feature map; The image data features are analyzed according to the feature graph, and the personnel detection module, the area tracking module and the activity recognition module detect and locate the target in the image or video; the personnel detection module uses a classifier to detect the target, and the classifier is used to determine whether there is a person in the image; the area tracking module uses a classifier to locate the target and determine the specific position of the person; The data analysis and fusion module processes and analyzes the result data information from the personnel detection module, the area tracking module and the activity recognition module, generates a final personnel tracking report, and feeds the tracking report back to the lighting control module. The lighting control module drives the lighting driving module to adjust the illumination intensity and projection angle of the lighting device according to the movement trajectory and number of personnel, so as to illuminate the dynamic area. In a specific embodiment, the working principle of the regional feature extraction module is as follows: For feature extraction, the convolution unit uses multiple convolution kernels to perform local convolution operations on the input image. The convolution kernel slides on the image, multiplies and sums the pixel values ​​of the corresponding area to obtain a feature map; the activation unit introduces a nonlinear activation function to transform the feature map to enhance the expression ability of the model; the pooling unit reduces the dimension and calculation amount of the feature map, while improving the robustness of the model. The pooling methods include maximum pooling and average pooling; the maximization retains the significant features in the image, and the average pooling smoothes the image and reduces noise; the batch normalization layer performs a normalization operation before the output of each layer to ensure the stable distribution of the input data, improve the training speed and stability of the neural network, and reduce the risk of overfitting; Perform feature fusion and classification, stack more than 5 convolution units and pooling units, gradually extract features, each unit extracts and transforms features based on the output of the previous unit, and finally forms a multi-dimensional feature map; the fully connected unit expands the multi-dimensional feature map into a one-dimensional vector, and maps it to the output space through linear transformation and activation function; The fully connected unit obtains a final output result of the model according to the size of the probability distribution, and the output result at least includes whether the human activity is detected and the location of the human; Optimize and train, define a loss function to measure the difference between the model output and the true label, calculate the gradient of the loss function to the model parameters through the back propagation algorithm, and use gradient descent optimization to update the parameters, and iterate continuously until the performance of the model on the training set reaches the preset standard or reaches a certain number of iterations. In the present invention, the output of the activation function in the regional feature extraction module is: In formula (1), is the number of convolutional unit layers, For the The activation output of the layer, is the activation function, is the activation output of the previous layer, For the The convolution kernel of the layer, is the bias term, is the offset control factor, m is the convolution random coefficient, For the The linear combination output of the layer; the formula of the activation function is: In formula (2), The output probability result of the personnel detection module is: In formula (3), x is whether human activity is detected, j is the category, The probability of class j is is the original output of the classifier for category j, N is the total number of categories, is the output of the classifier for the nth category. In a specific embodiment, the original output calculation formula of the activation function in the regional feature extraction module is: Introducing the offset control factor Adjust the output range of the activation function; the convolution random coefficient m introduces randomness to enhance the generalization ability of the model; the convolution kernel It is the core component in the convolutional neural network, which is used to slide on the input data and calculate the dot product to extract features; bias term Adjust the output of the activation function; The output of the activation function is obtained by weighted summing, and the output may be used as the input of subsequent layers or as the final output of the network. The original formula of the activation function is: Weight coefficient Represents the result after convolution and pooling of input data; The original calculation formula of the output probability result of the personnel detection module is: x is a variable indicating whether human activity is detected, which is a binary variable with a value of 0 or 1, where 0 indicates that human activity is not detected and 1 indicates that human activity is detected; j is an index of a category, which includes at least human activities in different areas and different activity types; In machine learning, a classifier takes input data and produces an output, which is usually a number or a vector representing the probability or confidence that the input data belongs to each category.

[0023] In a specific embodiment, the test results of the area tracking neural network model are shown in Table 1. In the present invention, the fault monitoring module includes a data processing unit and a collection unit, an analysis and diagnosis unit, a communication unit and an alarm unit connected to the data processing unit; the alarm unit includes a first alarm unit and a second alarm unit, the first alarm unit is located on the lighting device, and the second alarm unit is located in the host computer module. When the analysis and diagnosis unit detects an abnormality, the first alarm unit performs an audible and visual alarm, and the second alarm unit performs an alarm through the display unit in the host computer module; the collection unit collects environmental parameters in real time through the sensor module; the analysis and diagnosis unit determines whether the light intensity exceeds a threshold value and whether the angle selection module works normally, and the analysis and diagnosis unit uses an isolation forest for abnormal diagnosis. The working principle of the isolation forest is to extract 256 sub-sample sets from the original data set, and for each sub-sample set, the first alarm unit performs an audible and visual alarm, and the second alarm unit performs an alarm through the display unit in the host computer module; the collection unit collects environmental parameters in real time through the sensor module; the analysis and diagnosis unit determines whether the light intensity exceeds a threshold value and whether the angle selection module works normally, and the analysis and diagnosis unit uses an isolation forest for abnormal diagnosis. After the sample set, an isolated tree is seen, a feature is randomly selected, a cutting point q is randomly selected within the value range of the feature, a hyperplane is generated, the samples less than q are divided into the left subtree, and the samples greater than or equal to q are divided into the right subtree, and the above steps are recursively performed until only one sample is left; the anomaly score is the ratio of the average length of the data point to the average path length of the normal sample in the tree. If the anomaly score exceeds the threshold, it is judged as an abnormal point, and the first alarm unit and the second alarm unit start to alarm; the signal output end of the acquisition unit is connected to the signal input end of the data processing unit, the signal output end of the data processing unit is connected to the signal input end of the analysis and diagnosis unit, the signal output end of the analysis and diagnosis unit is connected to the signal input end of the communication unit, and the signal output end of the communication unit is connected to the signal input end of the alarm unit.

[0024] In a specific embodiment, the working process of the isolation forest algorithm is as follows: Select a feature from the sample set and randomly select a range of feature values; Split the data points in the sample set into two subsets based on the selected features and ranges; Repeat (1) and (2) to continue splitting each subset until a predetermined stopping condition is reached, such as when the tree height reaches a maximum limit or only one data point remains in the subset; Construct a binary tree where each data point is a tree node and the depth of the tree is the path length of the data point; Repeat (1) to (4) to construct multiple independent random trees; For a new data point, the path length in each tree is calculated to determine whether it is an outlier. If the path length is shorter, the data point is considered to be an outlier.

[0025] In the present invention, the lighting control method of the heating energy-saving lighting device comprises the following steps: Step 1: Collecting environmental data information through the sensor integration module and the image recognition module; The environmental data information at least includes natural light intensity, presence or absence of human activities, human activity conditions and ambient temperature; the collected data is preprocessed, including at least data cleaning, conversion and standardization operations to eliminate noise and abnormal values ​​in the data; the sensor integration module and the image recognition module transmit the preprocessed data to the lighting control module; step 2, the lighting control module controls the lighting device through an adaptive PID control model; The adaptive PID control model includes a rule generation module, a rule evaluation module, an adaptive parameter adjustment module, a PID control module and a feedback loop module; the signal output end of the rule generation module is connected to the signal input end of the rule evaluation module, the signal output end of the rule evaluation module is connected to the signal input end of the adaptive parameter adjustment module, the signal output end of the adaptive parameter adjustment module is connected to the signal input end of the PID control module, the signal output end of the PID control module is connected to the signal input end of the feedback loop module, and the signal output end of the feedback loop module is connected to the signal input end of the rule generation module; Step three, the lighting drive module automatically adjusts the light intensity and angle of the energy-saving LED lamp according to the received control signal to meet the lighting requirements and achieve energy-saving effects.

[0026] In a specific embodiment, the working principle of the adaptive PID control model is as follows: (1) obtaining the actual state of the lighting device through the sensor integration module and the image recognition module, the actual state including at least light intensity, temperature and whether there is human activity, and comparing it with the expected state to obtain an error; (2) The control quantity is calculated by the PID control module according to the error. The calculation formula of the control quantity is: In formula (4), is the system error, is the output light brightness, is the proportionality coefficient, is the integration time, d is the time, c is the compensation coefficient, k is the number of sampling times, is the proportionality coefficient, is the integration coefficient, is the differential coefficient, is the sampling number, r is the number of people; (3) The adaptive adjustment module adjusts the lighting fixture online according to the changes in the lighting fixture and the uncertainty of the external environment. (4) Adjust the control amount As an output signal, it is transmitted to the lighting driving module; the lighting driving module adjusts the light intensity and lighting angle of the energy-saving LED lamp according to the received control signal; the actual state is continuously detected through the sensor integration module and the image recognition module, and fed back to the lighting control module to form a closed-loop control.

[0027] In a specific embodiment, the original calculation formula of the control amount is: Systematic Error is the difference between the expected light intensity and the actual light intensity. The actual brightness of the light emitted by the lighting device at the time of sampling; the compensation coefficient c compensates for some nonlinear or uncertain factors in the system to make the system more stable or respond faster; the number of people r is used to represent the number of people or activity level in the environment. By detecting the number of people, the lighting brightness can be automatically adjusted to save energy or improve comfort. In a specific embodiment, the test results of the adaptive PID control model are shown in Table 2. From the above table, it can be concluded that as the number of people increases, the lighting area increases; when no human activity is detected, the adaptive PID control model does not perform lighting control.

[0028] In a specific embodiment, the control method of the adaptive PID control model is: Sampling: sampling the system output at a certain sampling period to obtain the current deviation; Error calculation, calculate the current deviation, that is, the difference between the expected value and the actual value; Control output calculation, calculate the control output according to the PID control algorithm ; Execution control, based on the calculated control output, adjusts the actuator to affect the system input and achieves control of the system.

[0029] In a specific embodiment, the working steps of the adaptive PID control model are as follows: Initialize the parameters of the PID controller, including: proportional coefficient, integral time and differential time; Monitor the real-time response of the system, including changes in feedback signals and system outputs; Formulate parameter adjustment rules and algorithms to adjust the parameters of the PID controller according to the real-time situation of the system response; Calculate new PID parameters according to parameter adjustment rules and update them to the PID controller; According to the updated PID parameters, the new control output is calculated to control the system; According to the actual control effect and system response, the parameter adjustment rules are revised and optimized to improve the adaptive performance; Continuously monitor the system response, continuously iterate to adjust parameters and calculate control output, and realize adaptive control. In a specific embodiment, the PID algorithm has three adjustment modes: proportional, integral and differential; the proportional adjustment is to proportionally react to the deviation of the system. Once the system deviates, the proportional adjustment immediately produces an adjustment effect to reduce the deviation. A large proportional effect can speed up the adjustment and reduce the error. However, when the proportion is too large, the stability of the system decreases, and even causes instability in the system; the integral adjustment is to eliminate the steady-state error of the system and improve the degree of freedom from error. If there is an error, the integral adjustment is adjusted until there is no error, and the integral adjustment stops. The integral adjustment outputs a constant value. The strength of the integral effect depends on the integral time constant. The stronger the integral effect, the If the differential adjustment is too large, the integral effect is weak. Adding integral adjustment can reduce the stability of the system and slow down the dynamic response. The differential adjustment reflects the rate of change of the system deviation signal, has predictability, and can predict the trend of deviation change, so it can produce advanced control. Before the deviation is formed, it has been eliminated by the differential adjustment, which can improve the dynamic performance of the system. When the differential time is selected appropriately, it can reduce overshoot and adjustment time. The differential action has an amplifying effect on noise interference. Adding too strong differential adjustment at one time is not conducive to the system's anti-interference. In addition, the differential reflects the rate of change, and when the input does not change, the differential action output is 0. In the present invention, the rule generation module generates the adjustment rules of the PID control parameters according to the operating characteristics, energy-saving goals and external environmental conditions of the lighting device; the rule evaluation module compares the calculated performance index with the preset expected performance through the threshold comparison algorithm to determine whether the current PID parameter adjustment rules meet the requirements; the adaptive parameter adjustment module: dynamically adjusts the parameters of the PID controller according to the output of the rule evaluation module and the real-time feedback of the operating status of the lighting device. The PID control module calculates a control signal based on the error between the set target light intensity and the actual light intensity to adjust the brightness of the lighting device; the feedback loop module feeds back the actual light intensity of the lighting device to the PID control module to form a closed-loop control to ensure that the lighting device can accurately track the target light intensity.

[0030] In a specific embodiment, the working steps of the rule generation module are: obtaining system feedback information, obtaining real-time feedback information from the system, including the system output, state variables and deviations; Calculate the control error, which is the difference between the expected output value and the actual output value of the system; Rule base query, querying a predefined rule base, the rule base contains PID parameter adjustment rules corresponding to different control error ranges; Rule matching, the selected rule will match the current control error to determine the PID parameters or adjustment strategies that need to be adjusted. In a specific embodiment, the rule evaluation module includes a performance indicator calculation module, an expected performance comparison module, a rule effectiveness evaluation module, a feedback module, and a continuous monitoring module; the performance indicator calculation module calculates a series of performance indicators through ISE real-time feedback information and control errors; the expected performance comparison module compares the calculated performance indicator with the pre-set expected performance through a threshold comparison algorithm to determine whether the current PID parameter adjustment rule meets the requirements; the rule effectiveness evaluation module evaluates the actual performance data and expected performance through a fuzzy logic algorithm to evaluate the effectiveness of the current PID parameter adjustment rule; the feedback module feeds back the evaluation results to the rule generation module through a logistic regression algorithm; the continuous monitoring module continuously monitors the performance of the system through a sliding window averaging algorithm, and continuously evaluates the effectiveness of the PID parameter adjustment rule based on real-time data. In a specific embodiment, the workflow of the rule evaluation module is as follows: Obtain real-time system performance data, including response time, stability, overshoot indicators, as well as the output of the frequency conversion control module and the feedback information of the drive compression module; Based on the acquired real-time data, the rule evaluation module calculates performance indicators to evaluate the effect of the current PID parameter adjustment rules. The performance indicators can reflect the control quality and stability of the system. The rule evaluation module compares the calculated performance indicators with the pre-set expected performance, which is set according to system requirements and control objectives, such as minimum overshoot and fastest response time; According to the comparison results of actual performance data and expected performance, the rule evaluation module evaluates the effectiveness of the current PID parameter adjustment rules. If the rules can meet the expected performance requirements, the rules are considered effective, otherwise they need to be adjusted; After the evaluation is completed, the rule evaluation module will feed back the evaluation results to the rule generation module. If the current rule is evaluated as valid, the rule generation module can continue to use the rule. If the rule does not meet the requirements, the rule generation module will adjust the PID parameters or generate new rules based on the evaluation results.

[0031] In a specific embodiment, the adaptive parameter adjustment module has the following parameter adjustment methods: Based on the adjustment of the error, the PID parameters are adjusted according to the size and change trend of the error signal. When the error is large, the proportional coefficient is increased; when the error accumulates more, the integral coefficient is increased; when the error changes faster, the differential time is increased; Based on the adjustment of system frequency response, the PID parameters are adjusted according to the frequency response characteristics of the system; Adjustment based on model identification uses system identification technology to dynamically adjust parameters according to the system model.

[0032] The adjustment based on model recognition uses system identification technology to dynamically adjust parameters according to the system model. In a specific embodiment, the workflow of the adaptive parameter adjustment module is as follows: System modeling and parameter identification: Model the system to determine the initial parameters of the PID controller, perform parameter identification, and estimate the dynamic characteristics of the system through the real-time response data of the system; Set expected output: Determine the expected output of the system, the output value that the system should achieve under ideal circumstances; Error calculation: During the control process, the error signal between the actual output and the expected output of the system is calculated; Parameter update rule design: adjust the parameters of the PID controller according to the error signal and system performance indicators; Parameter adjustment: According to the parameter update rules, the proportional, integral and differential parameters of the PID controller are adjusted in real time to make the actual output as close to the expected output as possible; Convergence analysis: Conduct convergence analysis on the parameter adjustment process to ensure that the parameter adjustment can converge to a stable state.

[0033] Performance evaluation and tuning: Continuously monitor the system's performance indicators, such as overshoot and stability, and tune the parameter adjustment strategy based on actual results to improve system control performance.

[0034] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that these specific embodiments are only illustrative, and those skilled in the art may omit, replace, and change the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, merging the above method steps so as to perform substantially the same functions in substantially the same manner to achieve substantially the same results is within the scope of the present invention. Therefore, the scope of the present invention is limited only by the appended claims.

Claims

1. A lighting device for heating and energy saving, characterized in that: Include: A sensor integrated module, the sensor integrated module includes a photoresistor circuit, an infrared sensor circuit and a temperature sensor circuit; the photoresistor circuit is used to monitor the natural light intensity indoors and outdoors; the infrared sensor circuit monitors the thermal radiation of the human body, and when the infrared sensor circuit detects human activity, it provides a trigger signal for the image recognition module; the temperature sensor circuit is used to monitor the temperature of the environment, and when the indoor temperature is greater than 30°C, the lighting control module controls the lighting device to reduce the brightness and reduce the heat load; Image recognition module, which detects the position of people in real time through cameras with a resolution greater than 1080P; Image feature extraction, data analysis and lighting area active object tracking are performed through a regional tracking neural network model, the regional tracking neural network model includes a regional feature extraction module, a personnel detection module, a regional tracking module, an activity recognition module and a data analysis and fusion module; the regional tracking neural network model performs feature extraction and regional tracking based on an adaptive convolutional neural network; the signal output end of the regional feature extraction module is connected to the signal input end of the personnel detection module, the signal output end of the personnel detection module is connected to the signal input end of the regional tracking module, the signal output end of the regional tracking module is connected to the signal input end of the activity recognition module, and the signal output end of the activity recognition module is connected to the signal input end of the data analysis and fusion module; A lighting control module, which adjusts the lighting brightness according to the data information of the sensor integration module and the image recognition module, provides local lighting according to the position of personnel, or adjusts the light intensity for a specific area; A lighting drive module uses energy-saving LED lamps for lighting, and adjusts the switch, angle and light intensity of the lighting device according to the instructions of the lighting control module; The host computer module displays the operating status of the lighting device through the display screen, receives user instructions and adjusts the control strategy, and performs remote control and optimization settings as needed; A communication module, which uses a wireless communication method ZigBee to realize data transmission between the lighting control module and the host computer module; Fault monitoring module, which monitors the operating status of lighting devices in real time through isolated forests; The signal output end of the sensor integrated module is connected to the signal input end of the image recognition module, the signal output end of the sensor integrated module is connected to the signal input end of the lighting control module, the signal output end of the image recognition module is connected to the signal input end of the lighting control module, the signal output end of the lighting control module is connected to the signal input end of the lighting driving module, the lighting control module is bidirectionally connected to the communication module, the communication module is bidirectionally connected to the host computer module, and the fault monitoring module of the lighting control module is bidirectionally connected.

2. A heating energy-saving lighting device according to claim 1, characterized in that: The energy-saving LED lamp includes a focusing module, a reflection cover module and an angle rotation module; the focusing module is used to control and concentrate the light beam of the energy-saving LED lamp, and concentrate the light on the target area by changing the focal length of the convex lens; the reflection cover module reflects or concentrates the light emitted by the lamp to the area that needs to be illuminated, reducing the loss and waste of light; the angle rotation module adjusts the illumination angle according to the data information of the image recognition module to achieve directional lighting and dynamic lighting.

3. A heating energy-saving lighting device according to claim 1, characterized in that: The working principle of the region tracking neural network model is: (1) Collecting video image information in real time through a camera with a resolution greater than 1080P installed in the heating area; preprocessing the collected images, wherein the preprocessing includes at least image cropping, normalization and denoising, so as to improve image quality and reduce the amount of calculation; (2) extracting feature information from an image or video using the regional feature extraction module, and performing data analysis to generate a feature map; (3) Analyzing the image data features according to the feature graph, the person detection module, the area tracking module and the activity recognition module detect and locate the target in the image or video; the person detection module uses a classifier to detect the target, and the classifier is used to determine whether there is a person in the image; the area tracking module uses a classifier to locate the target and determine the specific location of the person; (4) The data analysis and fusion module processes and analyzes the result data information from the personnel detection module, the area tracking module and the activity recognition module, generates a final personnel tracking report, and feeds the tracking report back to the lighting control module. The lighting control module drives the lighting driving module to adjust the light intensity and projection angle of the lighting device according to the movement trajectory and number of personnel, so as to illuminate the dynamic area.

4. A heating energy-saving lighting device according to claim 3, characterized in that: The output of the activation function in the regional feature extraction module is: In formula (1), is the number of convolutional unit layers, For the The activation output of the layer, is the activation function, is the activation output of the previous layer, For the The convolution kernel of the layer, is the bias term, is the offset control factor, m is the convolution random coefficient, For the The linear combination output of the layer; the formula of the activation function is: In formula (2), The output probability result of the personnel detection module is: In formula (3), x is whether human activity is detected, j is the category, The probability of class j is is the original output of the classifier for category j, N is the total number of categories, is the output of the classifier for the nth category.

5. The heating energy-saving lighting device according to claim 1, characterized in that: The fault monitoring module includes a data processing unit and a collection unit, an analysis and diagnosis unit, a communication unit and an alarm unit connected to the data processing unit; the alarm unit includes a first alarm unit and a second alarm unit, the first alarm unit is located on the lighting device, and the second alarm unit is located in the host computer module. When the analysis and diagnosis unit detects an abnormality, the first alarm unit performs an audible and visual alarm, and the second alarm unit performs an alarm through the display unit in the host computer module; the collection unit collects environmental parameters in real time through the sensor module; the analysis and diagnosis unit determines whether the light intensity exceeds the threshold and whether the angle selection module works normally, and the analysis and diagnosis unit uses an isolation forest for abnormal diagnosis. The working principle of the isolation forest is to extract 256 sub-sample sets from the original data set, and an isolated tree is formed for each sub-sample set, a feature is randomly selected, and a cutting point q is randomly selected within the value range of the feature to generate a hyperplane, and the samples less than q are divided into the left subtree, and the samples greater than or equal to q are divided into the right subtree, and the above steps are recursively performed until only one sample is left; The anomaly score is the ratio of the average length of the data point to the average path length of the normal sample in the tree. If the anomaly score exceeds the threshold, it is judged as an abnormal point, and the first alarm unit and the second alarm unit start to alarm; the signal output end of the acquisition unit is connected to the signal input end of the data processing unit, the signal output end of the data processing unit is connected to the signal input end of the analysis and diagnosis unit, the signal output end of the analysis and diagnosis unit is connected to the signal input end of the communication unit, and the signal output end of the communication unit is connected to the signal input end of the alarm unit.

6. A lighting control method for a lighting device for energy-saving heating, characterized in that: The application of a heating energy-saving lighting device as claimed in any one of claim 15 comprises the following steps: Step 1: Collecting environmental data information through the sensor integration module and the image recognition module; The environmental data information at least includes natural light intensity, presence or absence of human activities, human activity conditions and environmental temperature; the collected data is preprocessed, including at least data cleaning, conversion and standardization operations to eliminate noise and abnormal values ​​in the data; the sensor integration module and the image recognition module transmit the preprocessed data to the lighting control module; Step 2: The lighting control module controls the lighting device through an adaptive PID control model; The adaptive PID control model includes a rule generation module, a rule evaluation module, an adaptive parameter adjustment module, a PID control module and a feedback loop module; the signal output end of the rule generation module is connected to the signal input end of the rule evaluation module, the signal output end of the rule evaluation module is connected to the signal input end of the adaptive parameter adjustment module, the signal output end of the adaptive parameter adjustment module is connected to the signal input end of the PID control module, the signal output end of the PID control module is connected to the signal input end of the feedback loop module, and the signal output end of the feedback loop module is connected to the signal input end of the rule generation module; Step three: the lighting driving module automatically adjusts the light intensity and angle of the energy-saving LED lamp according to the received control signal to meet the lighting needs and achieve energy-saving effects.

7. A lighting control method for a lighting device for energy-saving heating according to claim 6, characterized in that: The working principle of the adaptive PID control model is: (1) obtaining the actual state of the lighting device through the sensor integration module and the image recognition module, wherein the actual state includes at least light intensity, temperature, and whether there is human activity, and comparing the actual state with the expected state to obtain an error; (2) The control quantity is calculated by the PID control module according to the error. The calculation formula of the control quantity is: In formula (4), is the system error, is the output light brightness, is the proportionality coefficient, is the integration time, d is the time, c is the compensation coefficient, k is the number of sampling times, is the proportionality coefficient, is the integration coefficient, is the differential coefficient, is the sampling number, r is the number of people; (3) The adaptive adjustment module adjusts the lighting fixture online according to the changes in the lighting fixture and the uncertainty of the external environment. (4) Adjust the control amount As an output signal, it is transmitted to the lighting driving module; the lighting driving module adjusts the light intensity and lighting angle of the energy-saving LED lamp according to the received control signal; the actual state is continuously detected through the sensor integration module and the image recognition module, and fed back to the lighting control module to form a closed-loop control.

8. The lighting control method of a lighting device for energy-saving heating according to claim 6, characterized in that: The rule generation module generates adjustment rules for PID control parameters according to the operating characteristics, energy-saving targets and external environmental conditions of the lighting device; the rule evaluation module compares the calculated performance index with the preset expected performance through a threshold comparison algorithm to determine whether the current PID parameter adjustment rules meet the requirements; the adaptive parameter adjustment module dynamically adjusts the parameters of the PID controller according to the output of the rule evaluation module and the real-time feedback of the operating status of the lighting device. The PID control module calculates a control signal based on the error between the set target light intensity and the actual light intensity to adjust the brightness of the lighting device; the feedback loop module feeds back the actual light intensity of the lighting device to the PID control module to form a closed-loop control to ensure that the lighting device can accurately track the target light intensity.

Images