Intelligent lighting control method and device based on machine learning and electronic equipment

By using machine learning-based lighting control methods in smart buildings, the lighting control network model is trained to decide on the switching state of the lighting equipment, and the problems of high lighting control cost and poor accuracy in the prior art are solved, and refined lighting control is realized and higher control accuracy.

CN119071986APending Publication Date: 2024-12-03QINGDAO HISENSE INTELLIGENT BUILDING TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411265933.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The lighting control schemes in existing smart buildings are costly and have poor accuracy, and the sensor accuracy and sensitivity limit the effectiveness of lighting control.

Method used

Using a smart lighting control method based on machine learning, by determining the translucent side orientation information, time point information, weather information and staff status information of the lighting equipment, train the lighting control network model to decide on the switch state, and perform fine control to avoid The use of sensors.

Benefits of technology

It reduces the lighting control cost, improves the accuracy and refinement of control, and avoids the limitations of sensor accuracy and sensitivity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119071986A_ABST
    Figure CN119071986A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent lighting control method and device based on machine learning and electronic equipment, and the method comprises the steps: determining the light-transmitting side orientation information, time point information and weather information of a room where the current lighting equipment is located, and the working state information of a worker in the room for each piece of lighting equipment in an intelligent building; inputting the information into a trained lighting control network model corresponding to the lighting equipment, and determining on-off state decision information of the current lighting equipment based on the lighting control network model; and performing on-off control on the lighting equipment according to the on-off state decision information. A sensor does not need to be arranged, so that the illumination control cost is reduced; the problem that the accuracy of illumination control is poor due to the precision and sensitivity of the sensor is solved, and therefore the accuracy of illumination control is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of smart building energy-saving technology, and in particular to a smart lighting control method, device and electronic equipment based on machine learning. Background Art

[0002] In the field of smart building energy conservation, it is necessary to realize automatic control of lighting equipment. The current industry solutions are usually divided into three categories: ① Add infrared or radar sensors to the intelligent lighting circuit, and realize the situation where someone turns on the light and no one turns off the light through personnel detection or changes; this solution adds infrared or radar sensors in the area, which increases the cost of lighting control, and is subject to the accuracy and sensitivity of the sensor, making the accuracy of lighting control poor. ② Add light sensors in the area to realize automatic control of lighting equipment by monitoring brightness changes; this solution adds light sensors, which increases the cost of lighting control, and is subject to the accuracy and sensitivity of the sensor, making the accuracy of lighting control poor. ③ Timing control, set a control strategy based on a schedule, and automatically switch on and off at the time. This solution is not only cumbersome to set up, but also has poor accuracy in lighting control. Summary of the invention

[0003] The present application provides a machine learning-based intelligent lighting control method, device and electronic device to solve the problems of high control cost and poor lighting control accuracy of the intelligent lighting control solutions in the prior art.

[0004] In a first aspect, the present application provides a smart lighting control method based on machine learning, the method comprising:

[0005] For each lighting device in the smart building, determine the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located;

[0006] The light-transmitting side orientation information, the time point information, the weather information and the working status information are input into the trained lighting control network model corresponding to the lighting device, and the current switch state decision information of the lighting device is determined based on the lighting control network model; the lighting device is controlled to switch on and off according to the switch state decision information.

[0007] In a second aspect, the present application provides a smart lighting control device based on machine learning, the device comprising:

[0008] A determination module is used to determine, for each lighting device in the smart building, the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located;

[0009] A control module is used to input the light-transmitting side orientation information, the time point information, the weather information and the working status information into a trained lighting control network model corresponding to the lighting device, determine the current switch state decision information of the lighting device based on the lighting control network model; and perform switch control on the lighting device according to the switch state decision information.

[0010] In a third aspect, the present application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0011] Memory, used to store computer programs;

[0012] The processor is used to implement the intelligent lighting control method based on machine learning when executing the program stored in the memory.

[0013] In a fourth aspect, the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the machine learning-based intelligent lighting control method is implemented.

[0014] The present application provides a smart lighting control method, device and electronic device based on machine learning, the method comprising: for each lighting device in a smart building, determining the light-transmitting side orientation information, time point information, weather information and working status information of the room where the lighting device is currently located; inputting the light-transmitting side orientation information, the time point information, the weather information and the working status information into a trained lighting control network model corresponding to the lighting device, determining the current switch state decision information of the lighting device based on the lighting control network model; and performing switch control on the lighting device according to the switch state decision information.

[0015] The above technical solution has the following advantages or beneficial effects:

[0016] In this application, considering that the switch state of the lighting device is correlated with the light-transmitting side orientation information, weather information, time information, and work status information of the staff in the room where the lighting device is located, a lighting control network model for making switch state decisions of the lighting device is trained. For each lighting device in the smart building, after determining the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located, the above information is input into the trained lighting control network model corresponding to the lighting device, and the switch state decision information of the current lighting device is determined based on the lighting control network model; the lighting device is switched on and off according to the switch state decision information. On the first hand, the present application does not need to set up sensors, thereby reducing the cost of lighting control; on the second hand, the present application trains a lighting control network model suitable for each lighting device, and then controls the lighting device based on the lighting control network model, thereby realizing a refined lighting control solution; on the third hand, the present application provides a smart lighting control solution based on machine learning, which avoids the problem of poor accuracy of lighting control due to the accuracy and sensitivity of the sensor, thereby improving the accuracy of lighting control. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of the first intelligent lighting control process based on machine learning provided for this application;

[0019] Figure 2 A schematic diagram of the lighting equipment control process provided for this application;

[0020] Figure 3 A schematic diagram of the second intelligent lighting control process based on machine learning provided for this application;

[0021] Figure 4 A schematic diagram of the third intelligent lighting control process based on machine learning provided for this application;

[0022] Figure 5 A schematic diagram of the training process of the first lighting control network model provided in this application;

[0023] Figure 6 A schematic diagram of a process for determining a loss value of a lighting control network model to be trained provided in the present application;

[0024] Figure 7 A schematic diagram of the training process of the second lighting control network model provided in this application;

[0025] Figure 8 The machine learning-based smart lighting control architecture diagram provided for this application;

[0026] Fig. 9 A cold start flow chart of lighting equipment provided for this application;

[0027] Fig.10 A flow chart of the training and application of the lighting control network model provided for this application;

[0028] Fig.11 A schematic diagram of the structure of the intelligent lighting control device based on machine learning provided in this application;

[0029] Fig.12 This is a schematic diagram of the electronic device structure provided in this application. DETAILED DESCRIPTION

[0030] In order to make the purpose and implementation method of the present application clearer, the exemplary implementation method of the present application will be clearly and completely described below in conjunction with the drawings in the exemplary embodiments of the present application. Obviously, the described exemplary embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0031] It should be noted that the brief description of terms in this application is only for the convenience of understanding the embodiments described below, and is not intended to limit the embodiments of this application. Unless otherwise specified, these terms should be understood according to their common and usual meanings.

[0032] The terms "first", "second", "third", etc. in the specification and claims of this application and the above drawings are used to distinguish similar or similar objects or entities, and do not necessarily mean to limit a specific order or sequence, unless otherwise noted. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances.

[0033] The terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover but not exclude inclusion, for example, a product or device comprising a list of components is not necessarily limited to all the components expressly listed but may include other components not expressly listed or inherent to such product or device.

[0034] The term "module" refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic or combination of hardware and / or software code that is capable of performing the functions associated with that element.

[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0036] For the convenience of explanation, the above description has been made in conjunction with specific embodiments. However, the above exemplary discussion is not intended to be exhaustive or limit the embodiments to the specific forms disclosed above. Based on the above teachings, various modifications and variations can be obtained. The selection and description of the above embodiments are to better explain the principles and practical applications, so that those skilled in the art can better use the embodiments and various different variations of the embodiments suitable for specific use considerations.

[0037] Figure 1 The first schematic diagram of the intelligent lighting control process based on machine learning provided in this application includes the following steps:

[0038] S101: for each lighting device in the smart building, determining the light-transmitting side orientation information, time point information, weather information, and working status information of the staff in the room where the lighting device is currently located;

[0039] S102: Input the light-transmitting side orientation information, the time point information, the weather information and the working status information into the trained lighting control network model corresponding to the lighting device, and determine the current switch state decision information of the lighting device based on the lighting control network model; and perform switch control on the lighting device according to the switch state decision information.

[0040] The machine learning-based intelligent lighting control method provided in the present application is applied to an electronic device, which may be a terminal device such as a PC or a computer, or a server.

[0041] Smart buildings are, for example, office buildings, hospital buildings, school buildings, etc. Various lighting devices are deployed in smart buildings, such as energy-saving lamps, LED lamps and other devices. In this application, various lighting devices that are connected in series and uniformly controlled are referred to as lighting devices of the same lighting circuit, wherein the same lighting circuit contains at least one lighting device. For any lighting device in the same lighting circuit, a lighting control network model is trained for performing lighting control on the lighting device. When performing lighting control on each lighting device in a smart building, a lighting control cycle can be preset, for example, each lighting device is controlled once every 10 minutes, 20 minutes, etc. Controlling a lighting device also controls other lighting devices in the circuit where the lighting device is located with the same strategy.

[0042] When the current moment is determined as the moment for lighting control of each lighting device according to the preset lighting control cycle, for each lighting device in the smart building, the light-transmitting side orientation information of the room where the lighting device is located, the current time point information, the current weather information and the current working status information of the staff in the room are determined at the current moment. Generally speaking, lighting devices of the same lighting circuit are installed in the same room, and the same room refers to the same office, the same conference room, etc. The light-transmitting side orientation information of the room generally refers to the orientation information of the side where the window is installed in the room. For example, if the window of the room faces south, the light-transmitting side orientation information of the room is determined to be south; if the window of the room faces north, the light-transmitting side orientation information of the room is determined to be north. In special cases, if there is no window installed in the room, the light-transmitting side orientation information of the room is determined to be none. If windows are installed on multiple sides of the room, for example, windows are installed on the east side and the south side of the room, the light-transmitting side orientation information of the room is determined to be east and south, or the areas of the windows installed on the east side and the south side can be compared, and the orientation information of the side with the larger window area is selected as the light-transmitting side orientation information of the room. Optionally, a building information model (BIM) model or structural drawings of the smart building is pre-saved, and position information of the windows in each room is marked in the BIM model or structural drawings. According to the position information of the windows in each room marked in the BIM model or structural drawings, the orientation information of the light-transmitting side of the room can be determined.

[0043] In addition, the current time point information can be obtained based on the clock, and the current time point information is, for example, 10:10 am. Based on the weather-related application APP, the current weather information can be obtained, and the weather information includes, but is not limited to, cloudy, rainy, snowy, temperature, wind force, wind direction and other information. The current working status information of the staff in the room can also be determined. Among them, the working status information includes being in a working state and being in a non-working state. Optionally, the work schedule of the staff in the room can be saved in advance, and the work schedule records the working hours of the staff in the room. For example, the work schedule records 9 am to 12 pm and 1 pm to 6 pm as working hours, and the time other than the above working hours is non-working hours. According to the work schedule and the current time point information, it can be determined whether the current time is working time. If so, the current working status information of the staff in the room is determined to be in a working state, and if not, the current working status information of the staff in the room is determined to be in a non-working state.

[0044] The electronic device is deployed with a trained lighting control network model corresponding to the lighting device. When training the lighting control network model, based on the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is located at each moment in the training set, and the switch status label of the corresponding lighting device, the lighting control network model to be trained is input. Each moment is, for example, every 5 minutes, every 10 minutes, etc.

[0045] The predicted switch state of the lighting device is determined based on the lighting control network model to be trained, and the loss value of the lighting control network model to be trained is determined according to the predicted switch state of the lighting device and the switch state label of the lighting device; the lighting control network model to be trained is iteratively trained according to the loss value, and when a preset number of iterative training times is reached or the loss value meets the requirements, it is determined that the lighting control network model training is completed. The switch state label of the lighting device is the actual switch state of the lighting device at the current moment.

[0046] The trained lighting control network model has the ability to determine the switch state decision information of the lighting device at the current moment according to the input light-transmitting side orientation information of the room where the current lighting device is located, time point information, weather information and the working status information of the staff in the room. The switch state decision information includes whether the lighting device is in the on state or in the off state.

[0047] In the present application, the light-transmitting side orientation information, the current time point information, the current weather information and the working status information at the current moment are input into the trained lighting control network model corresponding to the lighting device, and the switch state decision information of the lighting device at the current moment is determined based on the lighting control network model; the lighting device is switched on and off according to the switch state decision information. If the switch state decision information indicates that the lighting device is in the on state, the lighting device is controlled to be in the on state; if the switch state decision information indicates that the lighting device is in the off state, the lighting device is controlled to be in the off state.

[0048] In this application, considering that the switch state of the lighting device is correlated with the light-transmitting side orientation information, weather information, time information, and work status information of the staff in the room where the lighting device is located, a lighting control network model for making switch state decisions of the lighting device is trained. For each lighting device in the smart building, after determining the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located, the above information is input into the trained lighting control network model corresponding to the lighting device, and the switch state decision information of the current lighting device is determined based on the lighting control network model; the lighting device is switched on and off according to the switch state decision information. On the first hand, the present application does not need to set up sensors, thereby reducing the cost of lighting control; on the second hand, the present application trains a lighting control network model suitable for each lighting device, and then controls the lighting device based on the lighting control network model, thereby realizing a refined lighting control solution; on the third hand, the present application provides a smart lighting control solution based on machine learning, which avoids the problem of poor accuracy of lighting control due to the accuracy and sensitivity of the sensor, thereby improving the accuracy of lighting control.

[0049] In the present application, the lighting control network model is trained based on the real on / off state of the lighting device at each moment. Before the lighting control network model is trained, the lighting device is generally turned on or off manually by the staff. Before the lighting control network model is trained, that is, when the trained lighting control network model is not deployed in the electronic device, in order to reduce the cost of lighting control while ensuring the lighting effect, the method also includes:

[0050] For each room in the smart building, determine the location information of the room and the location information of the light-transmitting area corresponding to the light-transmitting side of the room;

[0051] Determine the installation position information of each lighting device in the room, and according to the light-transmitting area position information and the installation position information of each lighting device, take the lighting device located in the light-transmitting area as a candidate lighting device;

[0052] Determine the ambient light transmittance at the current time point according to the location information of the room and the current weather information; determine the control ratio information of the candidate lighting device according to the ambient light transmittance;

[0053] A candidate lighting device with the control ratio information is selected as a target lighting device; and the target lighting device is controlled to switch on and off.

[0054] In the present application, for each room in a smart building, the location information of the room and the location information of the light-transmitting area corresponding to the light-transmitting side of the room are determined. The location information of the room is, for example, the longitude and latitude information of the room. In addition, the installation location information of each lighting device in the room is determined. According to the annotation information of the installation position of each lighting device in the pre-saved BIM model or structural drawing of the room, the installation location information of each lighting device in the room can be determined. After determining the light-transmitting side of the room, that is, the side where the window is installed, the location information of the light-transmitting area corresponding to the light-transmitting side of the room can be determined. Among them, the position of the window can be extended to the inside of the room by a preset length, and the area location information obtained after the extension is used as the light-transmitting area location information corresponding to the light-transmitting side. Then, based on the light-transmitting area location information and the installation location information of each lighting device, it can be determined which lighting devices are in the light-transmitting area, and each lighting device in the light-transmitting area is used as each candidate lighting device.

[0055] According to the location information of the room and the current weather information, the ambient light transmittance at the current time is determined; and the control ratio information of the candidate lighting device is determined according to the ambient light transmittance. Optionally, the current sunrise time and sunset time can be determined according to the location information of the room and the current date and month information. During the daytime, the closer to the sunrise time and sunset time, the less ambient light transmittance, and the farther away from the sunrise time and sunset time, the more ambient light transmittance. In addition, the influence of different weather information on the ambient light transmittance can be saved in advance, for example, the ambient light transmittance is less on rainy days and more on sunny days. Through the above analysis, the ambient light transmittance at the current time can be determined according to the location information of the room and the current weather information. The control ratio information of the candidate lighting device is determined according to the ambient light transmittance. It should be noted that the control ratio information can be the closing ratio information or the opening ratio information. When the ambient light transmittance is large, the closing ratio information is large, and the opening ratio information is small. When the ambient light transmittance is small, the closing ratio information is small, and the opening ratio information is large.

[0056] Take the shutdown ratio information as an example for explanation. For example, if the current time point is between sunrise and sunset, that is, it is daytime and the weather is sunny, and the ambient light transmittance is relatively high, then the shutdown ratio information of the candidate lighting device is determined to be 60%. If the current time point is between sunrise and sunset, that is, it is daytime and the weather is cloudy, and the ambient light transmittance is relatively low, then the shutdown ratio information of the candidate lighting device is determined to be 50%. It should be noted that since the lighting devices of the same lighting circuit are uniformly controlled, the shutdown ratio information in this application can be the shutdown ratio information of the candidate lighting circuit.

[0057] Select the candidate lighting device with the control ratio information as the target lighting device; and perform on / off control on the target lighting device. If the control ratio information is the off ratio information, the target lighting device is controlled to be off; if the control ratio information is the on ratio information, the target lighting device is controlled to be on. For example, there are 5 candidate lighting devices in the light-transmitting area, and it is determined that the off ratio information of each candidate lighting device is 20%, then one candidate lighting device is selected from the 5 candidate lighting devices as the target lighting circuit, and the one target lighting device is turned off, and the other 4 candidate lighting devices are turned on.

[0058] Figure 2 The schematic diagram of the lighting device control process provided for this application includes the following steps:

[0059] S201: For each room in the smart building, determine the location information of the room and the location information of the light-transmitting area corresponding to the light-transmitting side of the room;

[0060] S202: Determine installation location information of each lighting device in the room, and select the lighting device located in the light-transmitting area as a candidate lighting device according to the light-transmitting area location information and the installation location information of each lighting device;

[0061] S203: determining the ambient light transmittance at the current time point according to the location information of the room and the current weather information; and determining the control ratio information of the candidate lighting device according to the ambient light transmittance;

[0062] S204: Select the candidate lighting device with the control ratio information as the target lighting device; and perform on / off control on the target lighting device.

[0063] When the trained lighting control network model is not deployed, follow the above steps to control the switch of the lighting equipment. If the user experience is not good, manual control is performed. During the control process of the lighting equipment, the light-transmitting side orientation information, time point information, weather information, and working status information at each moment are gradually collected to form a training set, and then the lighting control network model is trained based on the information in the training set.

[0064] In the present application, the determining of the light-transmitting side orientation information of the room where the lighting device is currently located includes:

[0065] Determine the light-transmitting side orientation information of the room where the lighting device is currently located according to the annotation information in the pre-saved building information model (BIM) model or the structural drawing of the room; or

[0066] Determine the first image information of each side of the room where the lighting device is currently located, input the first image information of each side into a trained light-transmitting side recognition model, determine the light-transmitting side of the room based on the light-transmitting side recognition model, and determine the orientation information of the light-transmitting side based on the position information of the light-transmitting side in the room.

[0067] The present application provides two methods for determining the light-transmitting side orientation information of the room where the lighting device is located at the current moment. Method 1: Determine the light-transmitting side orientation information of the room where the lighting device is located at the current moment based on the annotation information of the windows in the pre-saved BIM model or structural drawing of the room. The process of determining the light-transmitting side orientation information in this method is simple and efficient. However, considering that the weather is hot, the windows on one or several sides of the room have curtains drawn or blinds closed. Although there are windows, after the curtains are drawn or the blinds are closed, they are still considered as light-transmitting sides, which will lead to inaccurate determination of the light-transmitting side orientation information. In order to improve the accuracy of determining the light-transmitting side orientation information, the present application provides a second method. Method 2: Determine the first image information of each side of the room where the current lighting device is located, input the first image information of each side into the trained light-transmitting side recognition model, determine the light-transmitting side of the room based on the light-transmitting side recognition model, and determine the light-transmitting side orientation information based on the position information of the light-transmitting side in the room.

[0068] Considering the difference between the images of the light-transmitting side of the unblocked window of the room and the non-light-transmitting side such as the wall, the light-transmitting side recognition model can be pre-trained. The training process is: collect sample images of the light-transmitting side and sample images of the non-light-transmitting side in the room under various scenes to form a training set, and the sample images of the light-transmitting side have a light-transmitting side label, and the sample images of the non-light-transmitting side have a non-light-transmitting side label. Input the sample images of the light-transmitting side and the sample images of the non-light-transmitting side in the training set, as well as their respective light-transmitting side labels and non-light-transmitting side labels into the light-transmitting side recognition model to be trained; predict whether the sample image is a light-transmitting side image or a non-light-transmitting side image based on the light-transmitting side recognition model to be trained, determine the loss value of the light-transmitting side recognition model to be trained according to the prediction information and the label information, iterate the light-transmitting side recognition model to be trained according to the loss value, and determine that the training of the light-transmitting side recognition model is completed when the preset number of iterative training times is reached or the loss value meets the requirements. The trained light-transmitting side recognition model has the ability to identify whether it is a light-transmitting side image or a non-light-transmitting side image according to the input image.

[0069] Determine the first image information of each side of the room where the lighting device is located at the current moment, input the first image information of each side into the trained light-transmitting side recognition model, determine the first image of the light-transmitting side based on the light-transmitting side recognition model, and then determine the light-transmitting side of the room according to the first image of the light-transmitting side, and then determine the orientation information of the light-transmitting side according to the position information of the light-transmitting side in the room.

[0070] The process of determining the light-transmitting side orientation information by method 2 of the present application can avoid the problem of inaccurate determination of the light-transmitting side orientation information by method 1 due to window obstruction, thereby improving the accuracy of determining the light-transmitting side orientation information and further improving the accuracy of smart lighting control.

[0071] Figure 3 A second schematic diagram of a smart lighting control process based on machine learning provided in this application, the process includes the following steps:

[0072] S301: for each lighting device in the smart building, determine the light-transmitting side orientation information, time point information, weather information and work status information of the staff in the room where the lighting device is currently located; wherein, determine the first image information of each side of the room where the lighting device is currently located, input the first image information of each side into a trained light-transmitting side recognition model, determine the light-transmitting side of the room based on the light-transmitting side recognition model, and determine the light-transmitting side orientation information according to the position information of the light-transmitting side in the room;

[0073] S302: Input the light-transmitting side orientation information, the time point information, the weather information and the working status information into the trained lighting control network model corresponding to the lighting device, determine the current switch state decision information of the lighting device based on the lighting control network model; and perform switch control on the lighting device according to the switch state decision information.

[0074] In this application, determining the working status information of the staff in the current room includes:

[0075] Obtaining a work schedule of the staff in the room, and determining the current work status information of the staff in the room according to the work schedule; or,

[0076] Determine second image information currently in the room, input the second image information into a trained working status recognition model, and determine the working status information of the staff member based on the working status recognition model.

[0077] The present application provides two methods for determining the working status of the staff at the current moment. Method 1: Obtain the work schedule of the staff in the room, and determine the current working status information of the staff in the room at the current moment based on the work schedule. The process of determining the working status of the staff in this method is simple and efficient. However, considering the situation that the staff works overtime during rest time, or takes a rest during working time, the accuracy of determining the working status of the staff in method 1 will be poor. In order to improve the accuracy of determining the working status and rest of the staff, the present application provides a second method. Method 2: Determine the second image information in the current room of the building, input the second image information into the trained working status recognition model, and determine the current working status information of the staff in the room based on the working status recognition model.

[0078] Considering that there are differences in the postures of workers in working state or non-working state, a working state recognition model can be pre-trained. The training process is: sample images of rooms containing workers in working state and sample images of rooms containing workers in non-working state in various scenarios are collected to form a training set, and the sample images of rooms containing workers in working state have working state labels, and the sample images of rooms containing workers in non-working state have non-working state labels. The sample images of rooms containing workers in working state and the sample images of rooms containing workers in non-working state in the training set, as well as their respective working state labels and non-working state labels are input into the working state recognition model to be trained; based on the working state recognition model to be trained, it is predicted whether the sample image contains workers in working state or workers in non-working state, and the loss value of the working state recognition model to be trained is determined according to the prediction information and the label information, and the working state recognition model to be trained is iteratively trained according to the loss value. When the preset number of iterative training times is reached or the loss value meets the requirements, it is determined that the training of the working state recognition model is completed. The trained working state recognition model has the ability to identify whether the input room image contains workers in working state or workers in non-working state. Determine the second image information of the current room, input the second image information into the trained working status recognition model, and determine whether the second image is an image of a staff member in a working state or an image of a staff member in a non-working state based on the working status recognition model. If it is determined that the second image is an image of a staff member in a working state, determine the current working status information of the staff member as being in a working state; if it is determined that the second image is an image of a staff member in a non-working state, determine the current working status information of the staff member as being in a non-working state.

[0079] The process of determining the working status information in the second application method can improve the accuracy of the working status information of the staff, thereby improving the accuracy of the smart lighting control.

[0080] Figure 4 A third schematic diagram of a smart lighting control process based on machine learning provided in this application, the process includes the following steps:

[0081] S401: for each lighting device in the smart building, determine the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located; wherein, determine the second image information in the room, input the second image information into the trained work status recognition model, and determine the work status information of the staff based on the work status recognition model;

[0082] S402: Input the light-transmitting side orientation information, the time point information, the weather information and the working status information into the trained lighting control network model corresponding to the lighting device, determine the current switch state decision information of the lighting device based on the lighting control network model; and perform switch control on the lighting device according to the switch state decision information.

[0083] For smart lighting devices, in addition to the controllable on and off state, the brightness information can also be adjusted. In different environments, the brightness of the lighting device in the on state can be controlled to be adjusted to the brightness that suits the user, thereby improving the user experience.

[0084] Based on the above considerations, in the present application, the decision information of the current switch state of the lighting device determined based on the lighting control network model includes:

[0085] Determining the switch state decision information of the current lighting device and the target brightness information of the lighting device currently in the on state based on the lighting control network model;

[0086] The performing switch control on the lighting device according to the switch state decision information comprises:

[0087] The lighting device is switched on and off according to the switch state decision information, and the brightness information of the lighting device currently in the on state is controlled to be adjusted to the target brightness information.

[0088] In the present application, when training the lighting control network model, at each moment, in addition to collecting the light-transmitting side orientation information of the room where the lighting device is located, the current time point information, the current weather information, the current working status information of the staff in the room, and the corresponding switch status label of the lighting device, the brightness information label of the lighting device in the on state is also collected. Based on the light-transmitting side orientation information of the room where the lighting device is located, the current time point information, the current weather information, the current working status information of the staff in the room, and the corresponding switch status label of the lighting device and the brightness information label of the lighting device in the on state, the lighting control network model is trained. The lighting control network model trained in this way has the ability to determine the switch state of the lighting device and determine the brightness information of the lighting device in the on state.

[0089] Furthermore, it is possible to determine the switch state decision information of the lighting device at the current moment and the target brightness information of the lighting device in the on state based on the lighting control network model; then control the switch of the lighting device according to the switch state decision information; and control the brightness information of the lighting device in the on state to be adjusted to the target brightness information.

[0090] Figure 5 The training process diagram of the first lighting control network model provided in this application includes the following steps:

[0091] S501: Input the light-transmitting side orientation information of the room where the lighting device is located, time point information, weather information, and working status information of the staff in the room at each moment in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state, into the lighting control network model to be trained;

[0092] S502: Determine the predicted on / off state of the lighting device and predicted brightness information of the lighting device in the on state based on the lighting control network model to be trained;

[0093] S503: Determine the loss value of the lighting control network model to be trained based on the predicted switch state of the lighting device, the predicted brightness information of the lighting device in the on state, the switch state label of the lighting device, and the brightness information label of the lighting device in the on state; iteratively train the lighting control network model to be trained based on the loss value, and determine that the lighting control network model training is completed when a preset number of iterative training times is reached or the loss value meets the requirements.

[0094] Considering that the lighting control network model is trained to determine the switch state of the lighting device and the brightness information of the lighting device in the on state at the same time, in order to make the lighting control network model training process more accurate, Figure 6 A schematic diagram of a process for determining a loss value of a lighting control network model to be trained provided in the present application, wherein the loss value of the lighting control network model to be trained is determined according to the predicted switch state of the lighting device, the predicted brightness information of the lighting device in the on state, the switch state label of the lighting device, and the brightness information label of the lighting device in the on state, including the following steps:

[0095] S601: Determine a first loss value of the lighting control network model to be trained according to the predicted on / off state of the lighting device and the on / off state label of the lighting device;

[0096] S602: Determine a second loss value of the lighting control network model to be trained according to the predicted brightness information of the lighting device in the on state and the brightness information label of the lighting device in the on state;

[0097] S603: Determine the loss value of the lighting control network model to be trained according to the first loss value and the second loss value.

[0098] It should be noted that when determining the loss value of the lighting control network model to be trained according to the first loss value and the second loss value, the sum of the first loss value and the second loss value can be directly used as the loss value of the lighting control network model to be trained, or different weights can be configured for the first loss value and the second loss value, respectively, and the loss value of the lighting control network model to be trained is obtained by weighted summing the first loss value and the second loss value. Then, the lighting control network model to be trained is iteratively trained according to the loss value, and when the preset number of iterative training times is reached or the loss value meets the requirements, it is determined that the lighting control network model training is completed.

[0099] In this application, in order to regularly optimize the lighting control network model, the method further includes:

[0100] According to a preset model update cycle, the light-transmitting side orientation information, time point information, weather information, and staff work status information in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state are updated;

[0101] The trained lighting control network model is trained again according to the updated training set.

[0102] The preset model update cycle is, for example, one month, two months, etc. For example, if the preset model update cycle is one month, the information in the original training set is deleted, and the light-transmitting side orientation information, time point information, weather information, and staff work status information obtained at each moment in the last month, as well as the corresponding switch status label of the lighting device, and the brightness information label of the lighting device in the on state are used as the updated information in the training set, and the trained lighting control network model is retrained according to the updated training set, so as to obtain the optimized lighting control network model. For another example, if the preset model update cycle is one month, the information in the original training set is kept unchanged, and the light-transmitting side orientation information, time point information, weather information, and staff work status information obtained at each moment in the last month, as well as the corresponding switch status label of the lighting device, and the brightness information label of the lighting device in the on state are added to the training set, so as to enrich the information in the training set and obtain the updated training set, and then the trained lighting control network model is retrained according to the updated training set, so as to obtain the optimized lighting control network model.

[0103] Figure 7 A schematic diagram of the training process of the second lighting control network model provided in this application includes the following steps:

[0104] S701: Input the light-transmitting side orientation information of the room where the lighting device is located, time point information, weather information, and work status information of the staff in the room at each moment in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state, into the lighting control network model to be trained;

[0105] S702: Determine the predicted on / off state of the lighting device and predicted brightness information of the lighting device in the on state based on the lighting control network model to be trained;

[0106] S703: Determine the loss value of the lighting control network model to be trained according to the predicted switch state of the lighting device, the predicted brightness information of the lighting device in the on state, the switch state label of the lighting device, and the brightness information label of the lighting device in the on state; iteratively train the lighting control network model to be trained according to the loss value, and determine that the lighting control network model training is completed when a preset number of iterative training times is reached or the loss value meets the requirements;

[0107] S704: According to the preset model update cycle, the light-transmitting side orientation information, time point information, weather information and staff work status information in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state are updated; based on the updated training set, the trained lighting control network model is re-trained.

[0108] Figure 8 The machine learning-based smart lighting control architecture diagram provided for this application includes: a basic information collection module 81, a real-time information perception module 82, a cold start module 83 based on energy-saving rules, a model training module 84, a self-learning optimization module 85 and a model update module 86.

[0109] The basic information collection module mainly binds the lighting equipment information with the building space information based on CAD drawings, BIM models and other related information. The real-time information perception module collects environmental information in real time, such as time, typical weather, and work schedule information in the building space. The cold start module based on energy-saving rules formulates customized intelligent lighting energy-saving control rules for different spaces and scenes such as office buildings, hospitals, and schools before model-based control. The model training module builds the association relationship (time + weather + orientation + working status + switch status + brightness mode) for each lighting circuit based on the LSTM model, and trains based on historical data. The self-learning optimization module, based on the trained model, inputs the influencing factors for each circuit, derives the control strategy and issues it, monitors the status update of manual control in real time, and feeds back the operation data to the model training module. The model update module regularly updates the model and continuously optimizes it.

[0110] Fig. 9 The lighting equipment cold start flow chart provided for this application is as follows: Fig. 9 As shown, the building type of the smart building is obtained. The building types are divided into office buildings, schools, hospitals, etc. If it is a school, hospital, etc., the process ends. If it is an office building, the date and time are obtained, and whether it is working time is determined according to the date and time. If it is working time, the sunset time is determined according to the latitude and longitude of the smart building; if the current time is less than the set time length from the sunset time, no control is performed and the process ends. If the current time is not less than the set time length from the sunset time, the current weather and lighting equipment spatial relationship table are obtained again. If it is sunny, start from the window to turn off the first proportion of lighting equipment. If it is cloudy, start from the window to turn off the second proportion of lighting equipment; wherein the first proportion is greater than the second proportion.

[0111] Working hours: On sunny days, turn off the lighting equipment of the lighting circuit near window X (X defaults to 50%); on cloudy days, turn off the lighting equipment of the lighting circuit near window Y (Y defaults to 25%). On sunny days, the control action will not be executed N minutes before sunset (N defaults to 30) on cloudy days, and the control circuit will not be executed M minutes before sunset (M defaults to 60) on cloudy days; X\Y\N\M can be updated by self-learning.

[0112] Non-working hours: On sunny days, turn off the lighting equipment of the lighting circuit near window X1 (X1 defaults to 75%); on cloudy days, turn off the lighting equipment of the lighting circuit near window Y1 (Y1 defaults to 50%). On sunny days, it will not be executed N minutes before sunset (N defaults to 30); on cloudy days, it will not be executed M minutes before sunset (M defaults to 60); X1\Y1\N\M can be updated by self-learning.

[0113] Fig.10 The training and application flow chart of the lighting control network model provided for this application is as follows: Fig.10 As shown, a historical database is constructed: date, lighting equipment ID, building space, longitude and latitude, orientation, weather type, energy-saving closing time, and manual opening time. 70% of the historical database is used as a training set and 30% as a test set. Based on the training set and the test set, the lighting control network model corresponding to each of the N lighting circuits is trained, and the training is completed with an accuracy rate of more than 95%. The lighting control network model can be an LSTM model, and N LSTM models are obtained, namely LSTM_ID1, LSTM_ID2, LSTM_ID3, ..., LSTM_IDN. For each lighting circuit, after obtaining the building type, date and time, longitude and latitude, real-time weather and equipment space relationship table, the corresponding LSTM model is input, one circuit and one strategy, and the switch state decision information of each lighting circuit is obtained, such as LSTM_ID1: off, LSTM_ID2: off, LSTM_ID3: on, ..., LSTM_IDN: on / off. Monitor and record the manual control time, perform statistical analysis, and record the operation data to update the historical database. X: shutdown ratio during sunny working hours; Y: shutdown ratio during cloudy working hours; X1: shutdown ratio during sunny non-working hours; Y1: shutdown ratio during cloudy non-working hours; N: judgment of sunset time on sunny days; M: judgment of sunset time on cloudy days.

[0114] Build a historical database, including information of all dimensions of both control factors and control factors. Train a prediction model with an accuracy of more than 95% for each loop according to the ratio of 7:3, 70% of which is based on LSTM model training and 30% is tested and verified. Transfer real-time data to the prediction model for judgment, give instructions for opening or closing each loop, and issue controls. Continue to monitor changes in the manual control state of each loop. Import actual operation data into the historical database. Statistically calculate the key judgment thresholds based on historical data: X, Y, X1, Y1, N, M.

[0115] In this application, the lighting control network model can select any machine learning model such as LSTM, SVM, or linear regression. The choice of model only affects the difficulty of training and has no substantial impact on the control scheme. This application makes a detailed description of the cold start rules and self-learning process for office buildings, and is applicable to various types of buildings such as hospitals, schools, libraries, museums, industrial parks, etc. In this application, whether the dimension selection of basic information, the dimension selection of real-time information, or the dimension of the historical database can be increased or decreased according to the actual operation effect, deformation is allowed. For the selection of intelligent lighting, this application can be applied to various types of intelligent lighting such as indoor, outdoor, and landscape, and does not distinguish between specific intelligent lighting manufacturers; it can also be combined with intelligent lighting control systems with auxiliary functions such as infrared and light sensing. For changes in working hours, in this application, the working hours can be adjusted according to the actual business format, for example, it can be 8:30-17:30; it can also be 9:00-18:00, it can also be a flexible working system, and it can also be assisted according to winter and summer vacations and national statutory holidays.

[0116] Fig.11 The schematic diagram of the structure of the intelligent lighting control device based on machine learning provided for this application includes:

[0117] The determination module 111 is used to determine, for each lighting device in the smart building, the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located;

[0118] The control module 112 is used to input the light-transmitting side orientation information, the time point information, the weather information and the working status information into the trained lighting control network model corresponding to the lighting device, determine the current switch state decision information of the lighting device based on the lighting control network model; and control the switch of the lighting device according to the switch state decision information.

[0119] The control module 112 is also used to determine, for each room in the smart building, the location information of the room and the location information of the light-transmitting area corresponding to the light-transmitting side of the room; determine the installation location information of each lighting device in the room, and according to the light-transmitting area location information and the installation location information of each lighting device, use the lighting device located in the light-transmitting area as a candidate lighting device; determine the ambient light transmittance at the current time point according to the location information of the room and the current weather information; determine the control ratio information of the candidate lighting device according to the ambient light transmittance; select the candidate lighting device with the control ratio information as the target lighting device; and perform on-off control on the target lighting device.

[0120] The determination module 111 is specifically used to determine the light-transmitting side orientation information of the room where the lighting device is currently located according to the annotation information in the pre-saved building information model (BIM) model or structural drawing of the room; or, to determine the first image information of each side of the room where the lighting device is currently located, input the first image information of each side into a trained light-transmitting side recognition model, determine the light-transmitting side of the room based on the light-transmitting side recognition model, and determine the light-transmitting side orientation information according to the position information of the light-transmitting side in the room.

[0121] The determination module 111 is specifically used to obtain the work schedule of the staff in the room, and determine the work status information of the staff in the room according to the work schedule; or, determine the second image information in the room, input the second image information into the trained work status recognition model, and determine the work status information of the staff based on the work status recognition model.

[0122] The control module 112 is specifically used to determine the switch state decision information of the current lighting device and the target brightness information of the lighting device that is currently in the on state based on the lighting control network model; to control the switch of the lighting device according to the switch state decision information, and to control the brightness information of the lighting device that is currently in the on state to be adjusted to the target brightness information.

[0123] The device also includes:

[0124] The training module 113 is used to input the light-transmitting side orientation information of the room where the lighting device is located, the time point information, the weather information and the working status information of the staff in the room, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state at each moment in the training set into the lighting control network model to be trained; determine the predicted switch state of the lighting device and the predicted brightness information of the lighting device in the on state based on the lighting control network model to be trained; determine the loss value of the lighting control network model to be trained according to the predicted switch state of the lighting device, the predicted brightness information of the lighting device in the on state, the switch status label of the lighting device and the brightness information label of the lighting device in the on state; iteratively train the lighting control network model to be trained according to the loss value, and determine that the training of the lighting control network model is completed when a preset number of iterative training times is reached or the loss value meets the requirements.

[0125] The training module 113 is specifically used to determine the first loss value of the lighting control network model to be trained based on the predicted on / off state of the lighting device and the on / off state label of the lighting device; determine the second loss value of the lighting control network model to be trained based on the predicted brightness information of the lighting device in the on state and the brightness information label of the lighting device in the on state; and determine the loss value of the lighting control network model to be trained based on the first loss value and the second loss value.

[0126] The training module 113 is also used to update the light-transmitting side orientation information, time point information, weather information and staff work status information in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state according to a preset model update cycle; and re-train the trained lighting control network model based on the updated training set.

[0127] According to the preset model update cycle, the light-transmitting side orientation information, time point information, weather information and staff work status information in the training set, as well as the switch status label of the lighting equipment of the corresponding lighting circuit and the brightness information label of the lighting equipment of the lighting circuit in the on state are updated; based on the updated training set, the trained lighting control network model is trained again.

[0128] The present application also provides an electronic device, such as Fig.12 As shown, it includes: a processor 121, a communication interface 122, a memory 123 and a communication bus 124, wherein the processor 121, the communication interface 122, and the memory 123 communicate with each other through the communication bus 124;

[0129] The memory 123 stores a computer program, and when the program is executed by the processor 121, the processor 121 executes any one of the above method steps.

[0130] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0131] The communication interface 122 is used for communication between the electronic device and other devices.

[0132] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0133] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0134] The present application also provides a computer storage readable storage medium, wherein the computer readable storage medium stores a computer program executable by an electronic device, and when the program runs on the electronic device, the electronic device implements any of the above method steps when executing.

[0135] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0136] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A smart lighting control method based on machine learning, characterized in that: The method comprises: For each lighting device in the smart building, determine the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located; The light-transmitting side orientation information, the time point information, the weather information and the working status information are input into the trained lighting control network model corresponding to the lighting device, and the current switch state decision information of the lighting device is determined based on the lighting control network model; the lighting device is controlled to switch on and off according to the switch state decision information.

2. The method according to claim 1, characterized in that If the lighting control network model is not deployed, the method further includes: For each room in the smart building, determine the location information of the room and the location information of the light-transmitting area corresponding to the light-transmitting side of the room; Determine the installation position information of each lighting device in the room, and according to the light-transmitting area position information and the installation position information of each lighting device, take the lighting device located in the light-transmitting area as a candidate lighting device; Determine the ambient light transmittance at the current time point according to the location information of the room and the current weather information; determine the control ratio information of the candidate lighting device according to the ambient light transmittance; A candidate lighting device with the control ratio information is selected as a target lighting device; and the target lighting device is controlled to switch on and off.

3. The method according to claim 1, characterized in that The determining of the light-transmitting side orientation information of the room where the lighting device is currently located comprises: Determine the light-transmitting side orientation information of the room where the lighting device is currently located according to the annotation information in the pre-saved building information model (BIM) model or the structural drawing of the room; or Determine the first image information of each side of the room where the lighting device is currently located, input the first image information of each side into a trained light-transmitting side recognition model, determine the light-transmitting side of the room based on the light-transmitting side recognition model, and determine the orientation information of the light-transmitting side based on the position information of the light-transmitting side in the room.

4. The method according to claim 1, characterized in that Determine the working status information of the staff in the current room including:. Obtaining a work schedule of the staff in the room, and determining the current work status information of the staff in the room according to the work schedule; or, Determine second image information currently in the room, input the second image information into a trained working status recognition model, and determine the working status information of the staff member based on the working status recognition model.

5. The method according to claim 1, characterized in that The determining of the current switch state decision information of the lighting device based on the lighting control network model includes: Determining the switch state decision information of the current lighting device and the target brightness information of the lighting device currently in the on state based on the lighting control network model; The performing switch control on the lighting device according to the switch state decision information comprises: The lighting device is switched on and off according to the switch state decision information, and the brightness information of the lighting device currently in the on state is controlled to be adjusted to the target brightness information.

6. The method according to claim 5, characterized in that The training process of the lighting control network model includes: Input the light-transmitting side orientation information of the room where the lighting device is located, time point information, weather information, and work status information of the staff in the room at each moment in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state, into the lighting control network model to be trained; Determining the predicted on / off state of the lighting device and the predicted brightness information of the lighting device in the on state based on the lighting control network model to be trained; Determine the loss value of the lighting control network model to be trained based on the predicted on / off state of the lighting device, the predicted brightness information of the lighting device in the on state, the on / off state label of the lighting device, and the brightness information label of the lighting device in the on state; iteratively train the lighting control network model to be trained based on the loss value, and determine that the training of the lighting control network model is completed when a preset number of iterative training times is reached or the loss value meets the requirements.

7. The method according to claim 6, characterized in that Determining the loss value of the lighting control network model to be trained according to the predicted switch state of the lighting device, the predicted brightness information of the lighting device in the on state, the switch state label of the lighting device, and the brightness information label of the lighting device in the on state includes: Determining a first loss value of the lighting control network model to be trained according to the predicted on / off state of the lighting device and the on / off state label of the lighting device; Determining a second loss value of the lighting control network model to be trained according to the predicted brightness information of the lighting device in the on state and the brightness information label of the lighting device in the on state; The loss value of the lighting control network model to be trained is determined according to the first loss value and the second loss value.

8. The method according to claim 6, characterized in that The method further comprises: According to a preset model update cycle, the light-transmitting side orientation information, time point information, weather information, and staff work status information in the training set, as well as the switch status label of the lighting device and the brightness information label of the lighting device in the on state are updated; The trained lighting control network model is trained again according to the updated training set.

9. A smart lighting control device based on machine learning, characterized in that: The device comprises: A determination module is used to determine, for each lighting device in the smart building, the light-transmitting side orientation information, time point information, weather information, and work status information of the staff in the room where the lighting device is currently located; A control module is used to input the light-transmitting side orientation information, the time point information, the weather information and the working status information into a trained lighting control network model corresponding to the lighting device, determine the current switch state decision information of the lighting device based on the lighting control network model; and perform switch control on the lighting device according to the switch state decision information.

10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor is used to implement the intelligent lighting control method based on machine learning as described in any one of claims 1-8 when executing the program stored in the memory.