Intelligent lighting control system and method based on AI algorithm

Abstract

The invention relates to the technical field of intelligent lighting, in particular to an intelligent lighting control system and method based on an AI algorithm, and the method comprises the following steps: S1, marginalized environment perception: synchronously collecting an unsteady light field intensity gradient of a coverage area, a user gait feature vector and a light pollution overlapping coefficient between devices; s2, pre-judging illumination conflicts: generating a potential spectral interference topological graph of each lamp in the next control period, and marking conflict hot spot areas; s3, dynamic priority distribution: distributing a dynamic adjustment priority weight for each lighting device; s4, privacy constraint collaborative optimization: generating a lamp brightness-color temperature collaborative adjustment matrix; and S5, anti-interference execution verification: verifying the deviation degree between the actual parameter and the expected target based on the light field intensity, and triggering a conflict resolution compensation mechanism. According to the invention, the brightness and color temperature of the lamp are intelligently optimized, the illumination effect is improved, the user privacy is guaranteed, the light pollution interference is effectively reduced, and the user experience is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of intelligent lighting technology, and in particular to an intelligent lighting control system and method based on an AI algorithm. Background Art

[0002] With the rapid development of smart homes and smart cities, intelligent and energy-saving control of lighting systems has become an important research direction. Modern lighting systems are not only required to provide appropriate lighting, but also need to dynamically adjust according to the environment and user needs to achieve efficient energy saving, comfortable experience and reduce light pollution. To achieve these goals, many smart lighting systems rely on sensors, data analysis and AI algorithms to optimize lighting control. However, how to ensure energy saving while reducing environmental interference and protecting user privacy remains a challenge in the field of smart lighting.

[0003] Currently, many smart lighting systems rely on fixed control algorithms or simple sensor data analysis and lack the ability to adapt to dynamic environmental changes in real time. Although existing systems can adjust brightness and color temperature in certain scenarios, most fail to take into account the complexity of user behavior, light pollution, and interference between devices, making it difficult to achieve precise control in high-density or complex environments. In addition, some existing smart lighting systems fail to effectively protect user privacy during data transmission and processing, especially for data involving user gait or other personal information, which poses a high risk of privacy leakage. Summary of the Invention

[0004] The present invention provides an intelligent lighting control system and method based on AI algorithms to achieve precise control of the lighting system. By real-time sensing of light field intensity, user gait, and light pollution overlap coefficient, it can dynamically adjust the brightness and color temperature of lamps to optimize lighting effects, while reducing light pollution and adapting to changes in user behavior, thereby improving user experience and energy efficiency.

[0005] The intelligent lighting control method based on AI algorithm includes the following steps: S1, edge environment perception: Through the embedded sensor array deployed in the lighting device body, the non-steady-state light field intensity gradient, user gait feature vector and light pollution overlap coefficient between devices are synchronously collected in the coverage area; S2, Lighting Conflict Prediction: Input the light pollution overlap coefficient into the spatiotemporal convolutional network to generate a potential spectral interference topology map of each lamp in the next control cycle and identify conflict hotspots; S3, dynamic priority allocation: Based on the identified conflict hotspots and the confidence level of the user's movement intention analyzed by the gait feature vector, dynamically adjust the priority weight for each lighting device; S4, privacy-constrained collaborative optimization: A federated learning framework is built on edge computing nodes to generate a lamp brightness-color temperature collaborative adjustment matrix based on priority weights through encrypted gradient exchange; S5, anti-interference execution verification: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

[0006] Optionally, the marginalized environment perception in S1 includes: S11, light field intensity gradient acquisition: By deploying an embedded sensor array on the lighting device body, non-steady-state light field intensity data within the coverage area is collected in real time. The sensor array includes multiple light sensors for collecting light field intensity at different positions to form a spatial distribution of light field intensity within the area; S12, user gait feature vector collection: using the sensor array to collect user gait features within the coverage area, and generating the user's gait feature vector by calculating the gait motion trajectory, cadence, and stride length; S13, calculation of light pollution overlap coefficient between devices: Based on the collected light field intensity data, calculate the degree of light pollution overlap between devices. By analyzing the overlapping areas of the light field intensity data, determine the light pollution overlap coefficients of different lighting devices, and characterize the cross-effect of light pollution generated in space.

[0007] Optionally, the user gait feature vector collection in S12 includes: S121, gait motion trajectory collection: Through the sensor array deployed in the area, the user's motion trajectory in the area is collected in real time. Each sensor records the user's location information based on the user's movement. ; S122, Cadence Calculation: Estimate the user's cadence by analyzing position changes in the motion trajectory ; S123, stride calculation: Calculate the stride length of each step based on the starting and ending positions of each step in the user's motion trajectory ; S124, generating a user's gait feature vector: using the calculated motion trajectory, cadence, and stride information to generate the user's gait feature vector .

[0008] Optionally, the illumination conflict prediction in S2 includes: S21, light pollution overlap coefficient input: The light pollution overlap coefficient is used as the input feature of the spatiotemporal convolutional network to analyze the spatiotemporal distribution characteristics of light pollution and predict the potential light interference between lamps; S22, spatiotemporal convolutional network processing: The input light pollution overlap coefficient is processed using a spatiotemporal convolutional network. The spatiotemporal convolutional network extracts features in the spatial and temporal dimensions through convolution operations, and analyzes the potential interference pattern of each lamp in the future control cycle; S23, identifying conflict hotspots: Generate a spectral interference topology map generated by each lamp in the next control cycle through the output of the spatiotemporal convolutional network, indicating the interference relationship and interference intensity between different lamps. Based on the generated spectral interference topology map, identify and mark the conflict hotspots.

[0009] Optionally, the identification conflict hotspot area in S23 includes: S231, generating an interference topology map: using the interference pattern output by the spatiotemporal convolutional network, generating a spectral interference topology map of each lamp in the next control cycle; S232, identifying conflict hotspot areas: Based on the generated spectral interference topology map, identify areas where the interference intensity exceeds the interference threshold and mark them as conflict hotspot areas, which are expressed as: ; in, is a collection of conflict hotspots. is the interference threshold, For lamps and lamps The spectral interference intensity between them.

[0010] Optionally, the dynamic priority allocation in S3 includes: S31, Analyze the user's gait feature vector: Analyze the user's movement intention using the user's gait feature vector, predict the user's movement direction, speed and activity area through gait analysis, and calculate the user's movement intention confidence ; S32, priority allocation: based on conflict hotspot areas and user movement intention confidence , calculate the dynamic priority weight of each lamp ; S33, normalize priority weights: normalize the dynamic priority weights of all lamps.

[0011] Optionally, parsing the user gait feature vector in S31 includes: S311, moving direction prediction: Based on the motion trajectory data in the gait feature vector, the user's moving direction is predicted by calculating the position change between consecutive steps; S312, speed prediction: Using the cadence and stride data, combined with the user's movement trajectory, the user's movement speed is predicted by multiplying the stride length and cadence. ; S313, activity area prediction: predict the user's activity area based on the user's movement direction and movement speed ; S314, calculate the confidence level of the user's movement intention: calculate the confidence level of the user's movement intention by integrating the prediction results of the movement direction, speed and activity area .

[0012] Optionally, the privacy constraint collaborative optimization in S4 includes: S41, build a federated learning framework: At the edge computing node, build a federated learning framework. Each lamp locally calculates its updated gradient and transmits the gradient to the central node for aggregation. S42, Gradient Encryption and Exchange: Each lighting device calculates its local gradient and encrypts it using homomorphic encryption technology. The encrypted gradient exchange is aggregated through the edge computing node; S43, Gradient Aggregation and Model Update: The central node (edge computing node) aggregates the received encrypted gradients, ensures that the synthesized gradients still meet the privacy protection requirements through encryption operations, and updates the global model; S44, adjusting brightness and color temperature based on priority weights: optimizing and adjusting the brightness and color temperature of each lamp according to the dynamic priority weights of each lamp device and the collaborative adjustment matrix output by the global model.

[0013] Optionally, the anti-interference execution verification in S5 includes: S51, issuing a coordinated adjustment matrix: issuing the calculated lamp brightness-color temperature coordinated adjustment matrix to each lamp for execution; S52, collecting light field intensity data after execution: the light field intensity data after the lamp is executed is collected in real time through the sensor array; S53, verify the deviation between the actual parameters and the expected target: verify the actual parameters by calculating the change in the light field intensity gradient and expected goals The deviation between ; S54, triggering the conflict resolution compensation mechanism: when the deviation When the preset threshold (set to 0.5) is exceeded, it indicates that light pollution interference exists, triggering the conflict resolution and compensation mechanism to further adjust the brightness-color temperature coordinated adjustment matrix of the lamp; S55, re-issuing the adjusted coordination matrix: re-issuing the adjusted brightness-color temperature coordination adjustment matrix to achieve illumination control and interference elimination.

[0014] The AI-based intelligent lighting control system is used to implement the above-mentioned AI-based intelligent lighting control method, and includes the following modules: Edge-based environmental perception module: By deploying an embedded sensor array on the lighting device itself, it collects the non-steady-state light field intensity gradient, user gait feature vector, and light pollution overlap coefficient between devices in real time. Lighting conflict prediction module: The collected light pollution overlap coefficient is input into the spatiotemporal convolutional network to generate a potential spectral interference topology map for each lamp in the next control cycle and identify conflict hotspots. Dynamic priority allocation module: Based on the identified conflict hotspots and the analyzed user gait feature vector, the user's movement intention confidence is calculated. Based on this confidence and the priority of the conflicting areas, the module dynamically adjusts the priority weights for each lighting device. Privacy-constrained collaborative optimization module: This module builds a federated learning framework on edge computing nodes and generates a brightness-color temperature collaborative adjustment matrix for lamps based on dynamic priority weights through encrypted gradient exchange. Anti-interference execution verification module: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

[0015] Beneficial effects of the present invention: This invention achieves a high degree of adaptability in intelligent lighting systems through the comprehensive application of AI algorithms, sensor technology, and edge computing. The system can perceive and analyze changes in light field intensity, user behavior, and environmental conditions in real time, accurately predict lighting conflicts, and dynamically adjust the brightness and color temperature of lamps according to different regions and user needs. This intelligent control method based on environmental perception improves the efficiency and response speed of the lighting system.

[0016] This invention, by building a federated learning framework at the edge computing node, can not only utilize the data of distributed devices for optimization, but also ensure the protection of data privacy. All gradient exchanges and calculations are protected by homomorphic encryption, avoiding the risk of sensitive data leakage. At the same time, through encrypted gradient exchange and dynamic priority allocation, the system can achieve fine control of lamp brightness and color temperature without compromising privacy protection, ensuring both optimization effect and privacy protection.

[0017] The present invention precisely controls the interference of light pollution, combines the user's gait characteristics to predict movement intentions, and makes real-time adjustments to different conflict hotspot areas. When it is found that the deviation between the expected target and the actual parameters exceeds the set threshold, the system automatically triggers the conflict resolution and compensation mechanism, and further adjusts the brightness and color temperature of the lamps, thereby minimizing light pollution and interference, ensuring the effectiveness of the system in dynamic environments, and improving the user's lighting experience and comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 Schematic diagram of the control method flow in an embodiment of the present invention; Figure 2 Schematic diagram of system function modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0021] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0022] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0023] like Figure 1 As shown, the intelligent lighting control method based on AI algorithm includes the following steps: S1, edge environment perception: Through the embedded sensor array deployed in the lighting device body, the non-steady-state light field intensity gradient, user gait feature vector and light pollution overlap coefficient between devices are synchronously collected in the coverage area; S2, Lighting Conflict Prediction: Input the light pollution overlap coefficient into the spatiotemporal convolutional network to generate a potential spectral interference topology map of each lamp in the next control cycle and identify conflict hotspots; S3, dynamic priority allocation: Based on the identified conflict hotspots and the confidence level of the user's movement intention analyzed by the gait feature vector, dynamically adjust the priority weight for each lighting device; S4, privacy-constrained collaborative optimization: A federated learning framework is built on edge computing nodes to generate a lamp brightness-color temperature collaborative adjustment matrix based on priority weights through encrypted gradient exchange; S5, anti-interference execution verification: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

[0024] The marginalized environmental perception in S1 includes: S11, light field intensity gradient acquisition: By deploying an embedded sensor array on the lighting device body, the non-steady-state light field intensity data in the coverage area is collected in real time. The sensor array includes multiple light sensors for collecting light field intensity at different positions to form the spatial distribution of light field intensity in the area, which is expressed as: ; in, is the light field intensity gradient, which describes the light field intensity in the spatial coordinate The changes in is the measured light field intensity, 、 、 The light intensity is 、 、 The partial derivatives in the direction represent the components of the light field intensity gradient; S12, user gait feature vector collection: using the sensor array to collect user gait features within the coverage area, and generating the user's gait feature vector by calculating the gait motion trajectory, cadence, and stride length; S13, calculation of light pollution overlap coefficient between devices: Based on the collected light field intensity data, the degree of light pollution overlap between devices is calculated. By analyzing the overlapping areas of the light field intensity data, the light pollution overlap coefficients of different lighting devices are determined to characterize the cross-effect of light pollution generated in space, which is expressed as: ; in, is the light pollution overlap coefficient, 、 Two different lighting devices are located at The light field intensity on This is the area where light pollution between devices overlaps. To consider the lighting coverage area of all lighting equipment, It is a tiny volume element of a spatial region.

[0025] The user gait feature vector collection in S12 includes: S121, gait motion trajectory collection: Through the sensor array deployed in the area, the user's motion trajectory in the area is collected in real time. Each sensor records the user's location information based on the user's movement. , expressed as: ; in, Indicates time The user's spatial location at any moment, and For users in The spatial coordinates of the moment, is a time variable, indicating the moment when data was collected; S122, Cadence Calculation: Cadence refers to the number of steps a user takes per unit time. The user's cadence is estimated by analyzing the position changes in the motion trajectory. , expressed as: ; in, is the cadence, which represents the number of steps per unit time, The total number of steps is calculated by detecting the number of steps completed by the user in the motion trajectory. The total walking time of the user, which means the time from the starting point to the end point; S123, stride calculation: Stride refers to the horizontal distance the user walks with each step. The stride of each step is calculated based on the starting and ending positions of each step in the user's motion trajectory. , expressed as: ; in, It is The length of the steps, and It is The starting position coordinates of the step, and It is The coordinates of the end position of the step; S124, generating a user's gait feature vector: using the calculated motion trajectory, cadence, and stride information to generate the user's gait feature vector , expressed as: .

[0026] Lighting conflict prediction in S2 includes: S21, light pollution overlap coefficient input: The light pollution overlap coefficient is used as the input feature of the spatiotemporal convolutional network to analyze the spatiotemporal distribution characteristics of light pollution and predict the potential light interference between lamps. It is expressed as: ; in, is the input vector of light pollution overlap coefficient, In the position The light pollution overlap coefficient measured at is the total number of sensor arrays, indicating the amount of light pollution data collected; S22, spatiotemporal convolutional network processing: The input light pollution overlap coefficient is processed using a spatiotemporal convolutional network. The spatiotemporal convolutional network extracts features in the spatial and temporal dimensions through convolution operations and analyzes the potential interference pattern of each lamp in the future control cycle, which is expressed as: ; in, is the output of the spatial convolution operation, representing the light pollution pattern at each location, is the spatial convolution kernel weight, which is used to extract spatial features from the light pollution overlap coefficient. is the total number of spatial input data, indicating the locations of all collected light pollution data; S23, identifying conflict hotspots: Generate a spectral interference topology map generated by each lamp in the next control cycle through the output of the spatiotemporal convolutional network, indicating the interference relationship and interference intensity between different lamps. Based on the generated spectral interference topology map, identify and mark the conflict hotspots.

[0027] The identification conflict hotspots in S23 include: S231, generate interference topology map: use the interference pattern output by the spatiotemporal convolutional network to generate the spectral interference topology map of each lamp in the next control cycle, expressed as: ; in, Indicates lamps and lamps The spectral interference intensity between Representing lamps in the spatiotemporal convolutional network output and an intermediate node The interference intensity between Representing lamps in the spatiotemporal convolutional network output and an intermediate node The interference intensity between is the total number of nodes in the network (equal to the number of lamps); S232, Identify conflict hotspot areas: Based on the generated spectral interference topology map, identify areas where the interference intensity exceeds the interference threshold and mark them as conflict hotspot areas. These areas represent lamps that need to be adjusted first to reduce light pollution, expressed as: ; in, is the set of conflict hotspot areas, representing the pairs of lamps whose interference intensity exceeds the threshold. is the interference threshold, For lamps and lamps The spectral interference intensity between them.

[0028] Dynamic priority assignment in S3 includes: S31, Analyze the user's gait feature vector: Analyze the user's movement intention using the user's gait feature vector, predict the user's movement direction, speed and activity area through gait analysis, and calculate the user's movement intention confidence ; S32, priority allocation: based on conflict hotspot areas and user movement intention confidence , calculate the dynamic priority weight of each lamp , expressed as: ; in, For lamps Dynamic priority weights, is the weight coefficient of light pollution interference, reflecting the degree of influence of light pollution. Lighting fixtures analyzed for user gait characteristics Confidence of nearby users’ movement intentions, The weight coefficient of the user's movement intention reflects the impact of the user's gait on priority allocation; S33, normalized priority weight: To ensure that the priorities of different lamps can be compared within a uniform range, the dynamic priority weights of all lamps are normalized, expressed as: ; in, Normalized lamps Dynamic priority weights, The total number of lamps.

[0029] The parsed user gait feature vector in S31 includes: S311, moving direction prediction: Based on the motion trajectory data in the gait feature vector, the user's moving direction is predicted by calculating the position change between consecutive steps, which is expressed as: ; in, is the user's moving direction angle, indicating the direction from the current point to the next point. is the inverse tangent function, used to calculate the distance from the origin to the point angle, return from the positive - the angle from the axis to the point, 、 For in time The user's location coordinates at the moment, 、 For in time The user's location coordinates at the moment; S312, speed prediction: Using the cadence and stride data, combined with the user's movement trajectory, the user's movement speed is predicted by multiplying the stride length and cadence. , expressed as: ; in, is the cadence, For the The stride of the step indicates the length of the step, is the total number of stride sequences, indicating the number of steps taken by the user; S313, activity area prediction: predict the user's activity area based on the user's movement direction and movement speed , expressed as: ; in, For users in time Current location at the moment, is the time step of prediction, is the unit vector of the moving direction, which represents the vector of the user's moving direction; Unit vector of the moving direction The normalized result of the direction vector from the user's current location to the next location, including: (1) Calculate the user's moving direction vector: Based on the user's current location and next position Coordinates, calculate the user's moving direction vector , expressed as: ; (2) Calculate the magnitude of the direction vector: Calculate the modulus of the vector , that is, the length of the direction vector, expressed as: ; (3) Normalized direction vector: To obtain a unit vector, the moving direction vector needs to be divided by its modulus to obtain a vector with a length of 1, which represents the user's direction, expressed as: ; ; in, is a directional unit vector, indicating the user's moving direction; S314, calculate the confidence level of the user's movement intention: calculate the confidence level of the user's movement intention by integrating the prediction results of the movement direction, speed and activity area , the higher the confidence, the clearer the user's mobile intention is, which can be expressed as: ; in, 、 、 are the corresponding weight coefficients respectively.

[0030] Privacy-constrained collaborative optimization in S4 includes: S41, build a federated learning framework: Build a federated learning framework at the edge computing node, allowing different lighting devices to perform model training locally without centralizing data to the cloud. Each lighting device calculates its updated gradient locally and transmits the gradient to the central node for aggregation. The loss function is expressed as: ; in, It is a device The loss function on It is a device The weights of the local model, It is a device The number of local data points, It is a local sample The loss function on ; S42, Gradient Encryption and Exchange: Each lighting device calculates its local gradient and encrypts the gradient using homomorphic encryption technology to ensure data privacy during transmission. The encrypted gradient exchange is aggregated through the edge computing node and expressed as: ; in, For equipment The encrypted gradient on For equipment Compute the local gradient, is a homomorphic encryption operation, For equipment The encryption key used; S43, Gradient Aggregation and Model Update: The central node (edge computing node) aggregates the received encrypted gradients, ensures that the synthesized gradients still meet the privacy protection requirements through encryption operations, and updates the global model, which is expressed as: ; in, is the weight vector of the global model, representing the model obtained by federated learning aggregation, For equipment The weights of the local model, For equipment The amount of local data, is the number of devices participating in the federated learning process, that is, the total number of lighting devices trained on the edge computing node, The total amount of data for all devices; S44, adjust brightness and color temperature based on priority weight: Based on the dynamic priority weight of each lamp device and the coordinated adjustment matrix output by the global model, optimize the brightness and color temperature of each lamp. The priority weight reflects the importance of the lamp in the current lighting control system. Lamps with higher priority will be adjusted first, expressed as: ; in, For lamps Brightness-color temperature collaborative adjustment matrix, 、 are weight coefficients for adjusting brightness and color temperature respectively. For lamps Dynamic priority weights, For lamps The brightness adjustment coefficient, For lamps Color temperature adjustment factor.

[0031] The anti-interference execution verification in S5 includes: S51, issuing the coordinated adjustment matrix: issuing the calculated lamp brightness-color temperature coordinated adjustment matrix to each lamp for execution, which is expressed as: ; in, For lamps The actual adjustment value after execution; S52, collecting light field intensity data after execution: The light field intensity data after the lamp is executed is collected in real time through the sensor array, and is expressed as: ; in, is in position The actual measured light field intensity at After execution Brightness or color temperature adjustment value of each lamp, It's a lamp The weight of the contribution to the light field intensity; S53, verify the deviation between the actual parameters and the expected target: verify the actual parameters by calculating the change in the light field intensity gradient and expected goals The deviation between , expressed as: ; in, is the deviation between the actual light field intensity and the expected target, is the light field intensity of the intended target; S54, triggering the conflict resolution compensation mechanism: when the deviation When the preset threshold (set to 0.5) is exceeded, it indicates that light pollution interference exists, triggering the conflict resolution and compensation mechanism to further adjust the brightness-color temperature coordinated adjustment matrix of the lamp to reduce interference, which is expressed as: ; in, For lamps The adjustment value of is the adjustment factor; S55, re-issuing the adjusted coordination matrix: re-issuing the adjusted brightness-color temperature coordination adjustment matrix to achieve illumination control and interference elimination.

[0032] like Figure 2 As shown, the intelligent lighting control system based on the AI algorithm is used to implement the above-mentioned intelligent lighting control method based on the AI algorithm, and includes the following modules: Edge-based environmental perception module: By deploying an embedded sensor array on the lighting device itself, it collects the non-steady-state light field intensity gradient, user gait feature vector, and light pollution overlap coefficient between devices in real time. Lighting conflict prediction module: The collected light pollution overlap coefficient is input into the spatiotemporal convolutional network to generate a potential spectral interference topology map for each lamp in the next control cycle and identify conflict hotspots. Dynamic priority allocation module: Based on the identified conflict hotspots and the analyzed user gait feature vector, the user's movement intention confidence is calculated. Based on this confidence and the priority of the conflicting areas, the module dynamically adjusts the priority weights for each lighting device. Privacy-constrained collaborative optimization module: This module builds a federated learning framework on edge computing nodes and generates a brightness-color temperature collaborative adjustment matrix for lamps based on dynamic priority weights through encrypted gradient exchange. Anti-interference execution verification module: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

[0033] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0034] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. An intelligent lighting control method based on AI algorithm, characterized in that: The following steps are involved: S1, edge environment perception: Through the embedded sensor array deployed in the lighting device body, the non-steady-state light field intensity gradient, user gait feature vector and light pollution overlap coefficient between devices are synchronously collected in the coverage area; S2, lighting conflict prediction: The light pollution overlap coefficient is input into the spatiotemporal convolutional network to generate the potential spectral interference topology map of each lamp in the next control cycle and identify the conflict hotspot areas; S3, dynamic priority allocation: Based on the identified conflict hotspots and the confidence level of the user's movement intention analyzed by the gait feature vector, dynamically adjust the priority weight for each lighting device; S4, privacy-constrained collaborative optimization: A federated learning framework is built on edge computing nodes to generate a lamp brightness-color temperature collaborative adjustment matrix based on priority weights through encrypted gradient exchange; S5, anti-interference execution verification: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

2. The intelligent lighting control method based on AI algorithm according to claim 1 is characterized in that: The marginalized environmental perception in S1 includes: S11, light field intensity gradient acquisition: By deploying an embedded sensor array on the lighting device body, non-steady-state light field intensity data within the coverage area is collected in real time. The sensor array includes multiple light sensors for collecting light field intensity at different positions to form a spatial distribution of light field intensity within the area; S12, user gait feature vector collection: using the sensor array to collect user gait features within the coverage area, and generating the user's gait feature vector by calculating the gait motion trajectory, cadence, and stride length; S13, calculation of light pollution overlap coefficient between devices: Based on the collected light field intensity data, calculate the degree of light pollution overlap between devices. By analyzing the overlapping areas of the light field intensity data, determine the light pollution overlap coefficients of different lighting devices, and characterize the cross-effect of light pollution generated in space.

3. The intelligent lighting control method based on AI algorithm according to claim 2, characterized in that: The user gait feature vector collection in S12 includes: S121, gait motion trajectory collection: Through the sensor array deployed in the area, the user's motion trajectory in the area is collected in real time. Each sensor records the user's location information based on the user's movement. ; S122, Cadence Calculation: Estimate the user's cadence by analyzing position changes in the motion trajectory ; S123, stride calculation: Calculate the stride length of each step based on the starting and ending positions of each step in the user's motion trajectory ; S124, generating a user's gait feature vector: using the calculated motion trajectory, cadence, and stride information to generate the user's gait feature vector .

4. The intelligent lighting control method based on AI algorithm according to claim 1, characterized in that: The illumination conflict prediction in S2 includes: S21, light pollution overlap coefficient input: The light pollution overlap coefficient is used as the input feature of the spatiotemporal convolutional network to analyze the spatiotemporal distribution characteristics of light pollution and predict the potential light interference between lamps; S22, spatiotemporal convolutional network processing: The input light pollution overlap coefficient is processed using a spatiotemporal convolutional network. The spatiotemporal convolutional network extracts features in the spatial and temporal dimensions through convolution operations, and analyzes the potential interference pattern of each lamp in the future control cycle; S23, identifying conflict hotspots: Generate a spectral interference topology map generated by each lamp in the next control cycle through the output of the spatiotemporal convolutional network, indicating the interference relationship and interference intensity between different lamps. Based on the generated spectral interference topology map, identify and mark the conflict hotspots.

5. The intelligent lighting control method based on AI algorithm according to claim 4 is characterized in that: The identification conflict hotspot areas in S23 include: S231, generating an interference topology map: using the interference pattern output by the spatiotemporal convolutional network, generating a spectral interference topology map of each lamp in the next control cycle; S232, identifying conflict hotspot areas: Based on the generated spectral interference topology map, identify areas where the interference intensity exceeds the interference threshold and mark them as conflict hotspot areas, which are expressed as: ； in, is a collection of conflict hotspots. is the interference threshold, For lamps and lamps The spectral interference intensity between them.

6. The intelligent lighting control method based on AI algorithm according to claim 1, characterized in that: The dynamic priority allocation in S3 includes: S31, Analyze the user's gait feature vector: Analyze the user's movement intention using the user's gait feature vector, predict the user's movement direction, speed and activity area through gait analysis, and calculate the user's movement intention confidence ; S32, priority allocation: based on conflict hotspot areas and user movement intention confidence , calculate the dynamic priority weight of each lamp ; S33, normalize priority weights: normalize the dynamic priority weights of all lamps.

7. The intelligent lighting control method based on AI algorithm according to claim 6, characterized in that: The parsing of the user gait feature vector in S31 includes: S311, moving direction prediction: predicting the user's moving direction by calculating the position change between consecutive steps based on the motion trajectory data in the gait feature vector; S312, speed prediction: Using the cadence and stride data, combined with the user's movement trajectory, the user's movement speed is predicted by multiplying the stride length and cadence. ; S313, activity area prediction: predict the user's activity area based on the user's movement direction and movement speed ; S314, calculate the confidence level of the user's movement intention: calculate the confidence level of the user's movement intention by integrating the prediction results of the movement direction, speed and activity area .

8. The intelligent lighting control method based on AI algorithm according to claim 7, characterized in that: The privacy-constrained collaborative optimization in S4 includes: S41, build a federated learning framework: At the edge computing node, build a federated learning framework. Each lamp locally calculates its updated gradient and transmits the gradient to the central node for aggregation. S42, Gradient Encryption and Exchange: Each lighting device calculates its local gradient and encrypts it using homomorphic encryption technology. The encrypted gradient exchange is aggregated through the edge computing node; S43, Gradient Aggregation and Model Update: The central node aggregates the received encrypted gradients, ensures that the synthesized gradients still meet the privacy protection requirements through encryption operations, and updates the global model; S44, adjusting brightness and color temperature based on priority weights: optimizing and adjusting the brightness and color temperature of each lamp according to the dynamic priority weights of each lamp device and the collaborative adjustment matrix output by the global model.

9. The intelligent lighting control method based on AI algorithm according to claim 8, characterized in that: The anti-interference execution verification in S5 includes: S51, issuing a coordinated adjustment matrix: issuing the calculated lamp brightness-color temperature coordinated adjustment matrix to each lamp for execution; S52, collecting light field intensity data after execution: the light field intensity data after the lamp is executed is collected in real time through the sensor array; S53, verify the deviation between the actual parameters and the expected target: verify the actual parameters by calculating the change in the light field intensity gradient and expected goals The deviation between ; S54, triggering the conflict resolution compensation mechanism: when the deviation When the preset threshold is exceeded, it indicates that there is light pollution interference, triggering the conflict resolution and compensation mechanism to further adjust the brightness-color temperature coordinated adjustment matrix of the lamp; S55, re-issuing the adjusted coordination matrix: re-issuing the adjusted brightness-color temperature coordination adjustment matrix to achieve illumination control and interference elimination.

10. An intelligent lighting control system based on an AI algorithm, used to implement the intelligent lighting control method based on an AI algorithm according to any one of claims 1 to 9, characterized in that: Includes the following modules: Edge-based environmental perception module: By deploying an embedded sensor array on the lighting device itself, it collects the non-steady-state light field intensity gradient, user gait feature vector, and light pollution overlap coefficient between devices in real time. Lighting conflict prediction module: The collected light pollution overlap coefficient is input into the spatiotemporal convolutional network to generate a potential spectral interference topology map for each lamp in the next control cycle and identify conflict hotspots. Dynamic priority allocation module: Based on the identified conflict hotspots and the analyzed user gait feature vector, the user's movement intention confidence is calculated. Based on this confidence and the priority of the conflicting areas, the module dynamically adjusts the priority weights for each lighting device. Privacy-constrained collaborative optimization module: This module builds a federated learning framework on edge computing nodes and generates a brightness-color temperature collaborative adjustment matrix for lamps based on dynamic priority weights through encrypted gradient exchange. Anti-interference execution verification module: After the lamp brightness-color temperature coordinated adjustment matrix is sent to each lamp for execution, the deviation between the actual parameters and the expected target is verified based on the light field intensity, triggering the conflict resolution and compensation mechanism.

Images