Load balancing method for intelligent lighting collaborative management and related equipment

Abstract

The invention provides a load balancing method for intelligent lighting collaborative management and related equipment, and the method comprises the steps: collecting the environment parameters of each lighting region in real time, and carrying out the preprocessing of the environment parameters; inputting the preprocessed environment parameters into a preset deep neural network model, and determining a load prediction sequence of each illumination area; judging the load state of each lighting area according to the load prediction sequence; based on the load state and the real-time environment data, generating a multi-region collaborative dimming rate set through a region layering collaborative strategy; and performing illumination control on the illumination equipment in the target illumination area according to the multi-area cooperative dimming rate set. Through the implementation of the scheme of the invention, intelligent dimming control under cross-region, dynamic perception and prediction driving is realized, and the dynamic adaptive capacity and the load control precision of the illumination system are improved.

Description

Technical Field

[0001] The present application relates to the field of optical display, and in particular to a load balancing method and related equipment for intelligent lighting collaborative management. Background Art

[0002] Amid the rapid development of smart cities and intelligent buildings, lighting systems, as a crucial component of building energy consumption, are gradually evolving towards intelligent and adaptive features. This is particularly true in complex, multifunctional spaces such as large office buildings, hospitals, schools, and transportation hubs, where different areas face varying lighting needs, frequent human activity, and the difficulty of controlling energy consumption. For example, a large, integrated office building has meeting areas, open workspaces, corridors, and rest areas distributed across different floors. These areas exhibit significantly varying occupancy densities and usage frequencies throughout the day. Traditional lighting systems often rely on timed control or simple strategies based on single sensors (such as light or occupancy sensors). These strategies struggle to achieve precise dimming based on multi-source environmental data, resulting in frequent over- or under-illumination of localized areas. This not only wastes energy but also impacts users' visual comfort and spatial experience. Summary of the Invention

[0003] The present application provides a load balancing method and related equipment for intelligent lighting collaborative management, which are used to solve the problem of lack of global optimization capability in the case of multi-region load mutation or resource conflict in related technologies.

[0004] In a first aspect, the present application provides a load balancing method for collaborative management of intelligent lighting, the load balancing method for collaborative management of intelligent lighting comprising: Collecting environmental parameters of each lighting area in real time and preprocessing the environmental parameters; Inputting the preprocessed environmental parameters into a preset deep neural network model to determine a load prediction sequence for each lighting area; Determining the load status of each lighting area according to the load prediction sequence; Based on the load status and real-time environmental data, generating a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy; Lighting control is performed on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

[0005] Optionally, in a first implementation of the first aspect of the present application, the step of collecting environmental parameters of each lighting area in real time and preprocessing the environmental parameters includes: Collect environmental parameters of each lighting area in real time and generate a set of original environmental parameters; Traversing the original environmental parameter set through a sliding window mechanism, deleting abnormal values ​​of data offset, and generating an abnormality-filtered data sequence; Generating a time-synchronized multi-data stream by time-aligning the data sequences; The multivariate data stream is normalized, and the processed multivariate data stream is spatially aggregated to generate a standardized data matrix.

[0006] Optionally, in a second implementation of the first aspect of the present application, the step of inputting the preprocessed environmental parameters into a preset deep neural network model to determine the load prediction sequence for each lighting area includes: Inputting the standardized data matrix into a preset deep neural network model to determine the data to be predicted by the deep neural network model; Acquiring historical data of each lighting area within a preset time period through the sliding window mechanism; Inputting the historical data into the deep neural network model, and outputting a load prediction value for each lighting area; performing error correction judgment on the load prediction value; If the prediction error of a preset number of time windows exceeds the error threshold, the load prediction value is dynamically corrected; The load prediction value is overwritten according to the prediction value corresponding to the modified time window to generate a load prediction sequence for each lighting area.

[0007] Optionally, in a third implementation of the first aspect of the present application, the step of determining the load state of each lighting area according to the load prediction sequence includes: Extracting the peak load value and the average load value of each lighting area in a preset future time period from the load prediction sequence; Calculating a dynamic load index based on a difference between the peak load value and the average load value and a historical load fluctuation range; Match dynamic correction factors based on regional function type and current time period type; Performing weighted fusion of the dynamic correction factor and the dynamic load index to generate a corrected load index; Comparing the modified load index with a load benchmark value of the same functional area in the same time period in a historical database; When the correction index continuously exceeds the load reference value, the corresponding lighting area is marked as overloaded, and a load state matrix is ​​generated in combination with the regional topological relationship; A load diffusion constraint is imposed on the overloaded lighting area according to the spatial correlation of the adjacent areas, and a final load state matrix is ​​generated by adjusting the weight values ​​of the adjacent areas.

[0008] Optionally, in a fourth implementation of the first aspect of the present application, the step of generating a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy based on the load status and real-time environmental data includes: generating an initial dimming rate for each of the lighting areas according to the weight values ​​of the final load state matrix and the light intensity distribution in the real-time environmental data; Performing inter-area conflict detection on the initial dimming rate, and generating a conflict-free dimming rate set by adjusting the dimming rate difference of the conflicting areas; Applying a dynamic attenuation factor to the dimming rate of each area in the non-conflicting dimming rate set according to the dynamic change rate of the population density in the real-time environmental data; Performing particle swarm optimization on the attenuated conflict-free dimming rate set based on the regional topological relationship to generate a dynamic weighted dimming rate set; The dynamic weighted dimming rate set is hierarchically integrated according to regional priority and spatial correlation to generate a multi-region coordinated dimming rate set.

[0009] Optionally, in a fifth implementation of the first aspect of the present application, the step of hierarchically fusing the dynamic weighted dimming rate set according to regional priority and spatial correlation to generate a multi-region coordinated dimming rate set includes: assigning a priority label to each of the lighting areas based on the overload level weight value of the final load state matrix; dividing the set of dynamic weighted dimming rates into subsets of different priorities according to the priority labels; A global dimming rate baseline value set generated for the conflict-free dimming rate set according to the different priority subsets and a local dimming rate compensation value set calculated by the edge node are differentially weighted according to a preset ratio; wherein the edge node is a distributed decision-making unit corresponding to each of the lighting areas; The differentiated allocated global dimming rate and the local dimming rate are superimposed through a layered fusion strategy to generate the multi-region coordinated dimming rate set.

[0010] Optionally, in a sixth implementation of the first aspect of the present application, the step of performing lighting control on the lighting equipment in the target lighting area according to the multi-zone coordinated dimming rate set includes: Acquire lighting area data information according to the multi-area coordinated dimming rate set and the final load state matrix, and generate a control data message; performing multi-layer convolution processing on the control data message according to the global dimming rate baseline value set data stored at the edge node to generate a joint prediction data structure reflecting the relationship between global and local dimming; Performing a hierarchical fusion operation on the global dimming rate baseline and the local dimming rate compensation data within the lighting area according to the joint prediction data structure and the regional topology relationship data, and generating an anti-conflict dimming instruction set; issuing control instructions to the lighting equipment in the target lighting area using the anti-conflict dimming instruction set; Perform a distributed dimming operation on the lighting device according to the control instruction.

[0011] A second aspect of the present application provides a load balancing device for collaborative management of intelligent lighting, the load balancing device for collaborative management of intelligent lighting comprising: A processing module, used to collect environmental parameters of each lighting area in real time and pre-process the environmental parameters; a determination module, configured to input the preprocessed environmental parameters into a preset deep neural network model to determine a load prediction sequence for each of the lighting areas; a judgment module, configured to judge the load status of each lighting area according to the load prediction sequence; A generating module, configured to generate a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy based on the load status and real-time environmental data; A control module is used to perform lighting control on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

[0012] A third aspect of an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the processor is used to execute a computer program stored on the memory. When the processor executes the computer program, it implements each step of the load balancing method for intelligent lighting collaborative management provided in the first aspect of the embodiment of the present application.

[0013] The fourth aspect of the embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the load balancing method for intelligent lighting collaborative management provided in the first aspect of the embodiment of the present application are implemented.

[0014] In summary, according to the load balancing method and related equipment for intelligent lighting collaborative management provided by the present application, the environmental parameters of each lighting area are collected in real time, and the environmental parameters are preprocessed; the preprocessed environmental parameters are input into a preset deep neural network model to determine the load prediction sequence of each lighting area; the load status of each lighting area is judged according to the load prediction sequence; based on the load status and real-time environmental data, a multi-region collaborative dimming rate set is generated through a regional hierarchical collaborative strategy; and lighting control is performed on the lighting equipment in the target lighting area according to the multi-region collaborative dimming rate set. Through the implementation of the present application, intelligent dimming control driven by cross-regional, dynamic perception and prediction is realized, which improves the dynamic adaptability and load control accuracy of the lighting system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart of a load balancing method for intelligent lighting collaborative management provided in an embodiment of the present application; Figure 2 A schematic diagram of the program modules of the load balancing device for intelligent lighting collaborative management provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of this application.

[0017] In order to solve the problem of lack of global optimization capability in the related art in the case of multi-region load mutation or resource conflict, the embodiment of the present application provides a load balancing method for intelligent lighting collaborative management, such as Figure 1 This is a flow chart of a load balancing method for intelligent lighting collaborative management provided in this embodiment. The load balancing method for intelligent lighting collaborative management includes the following steps: Step 110: Collect environmental parameters of each lighting area in real time and pre-process the environmental parameters.

[0018] Specifically, a distributed sensor network is deployed in different functional areas of the building. Light sensors, occupancy sensors, and equipment power monitoring modules are used to obtain real-time data on multi-dimensional parameters such as light intensity, occupancy density, and lighting equipment power. To ensure data quality and consistency, the collected raw data is filtered using a sliding window mechanism to remove outliers, and a time alignment algorithm is used to correct for sampling clock discrepancies between devices to ensure data synchronization. Furthermore, continuous variables such as light intensity and occupancy density are normalized to unify the data scale. Ultimately, a standardized data matrix is ​​constructed for modeling and analysis, providing input for subsequent intelligent prediction models.

[0019] In an optional implementation of this embodiment, the steps of collecting the environmental parameters of each lighting area in real time and preprocessing the environmental parameters include: collecting the environmental parameters of each lighting area in real time to generate an original environmental parameter set; traversing the original environmental parameter set through a sliding window mechanism, deleting abnormal values ​​of data offset, and generating an abnormality-filtered data sequence; generating a time-synchronized multivariate data stream by time-aligning the data sequence; normalizing the multivariate data stream, and spatially aggregating the processed multivariate data stream to generate a standardized data matrix.

[0020] Specifically, the environmental parameters of each lighting area are collected in real time through a distributed sensor network to obtain multi-dimensional data such as light intensity, personnel density, and equipment power. Sensors are electronic devices that can detect physical quantities and convert them into processable signals. The significance of this is to construct a detailed set of original environmental parameters. For example, in an office building, different areas output instantaneous illumination values ​​and personnel density data through photosensors and infrared sensors, respectively, to provide a data basis for subsequent predictions and help the system reflect the on-site situation in real time. Then, a sliding window mechanism is used to traverse the original environmental parameter set. Sliding window is a technology that continuously updates sampled data within a fixed time period to detect data offset caused by acquisition errors or instantaneous interference in the data. This method can dynamically identify and delete outliers, thereby generating an outlier-filtered data sequence. For example, when the collected value suddenly soars due to equipment failure at a certain moment, the mechanism will remove the offset data to ensure the stability and reliability of the data sequence. The filtered data sequence is further time-aligned. Time alignment refers to the correction and matching of data from each sensor on the time axis. Its significance is to overcome the time errors caused by the asynchronous internal clocks of each sensor. Differences are eliminated, allowing multivariate data to be compared and analyzed under the same time reference. For example, after alignment, light data from different floors can accurately reflect the environmental conditions at the same moment, thereby improving the accuracy of subsequent model predictions. The time-synchronized multivariate data streams are then normalized. Normalization involves mapping each data set to a unified numerical range through mathematical transformations. Its importance lies in eliminating inconsistencies in different data dimensions and scales. For example, normalizing light intensity and occupancy density to a range of 0 to 1 ensures that the model treats all input features fairly during training. Finally, the normalized multivariate data streams are integrated through spatial aggregation. Spatial aggregation combines adjacent areas or data of the same type based on the actual physical layout to form a unified data representation. Its purpose is to eliminate the noise caused by individual fluctuations of a single sensor. For example, the light data from each corner of an office can be averaged according to spatial weights and output as a standardized data matrix, providing stable and accurate input for subsequent deep learning models. The entire process ensures that the acquired data not only reflects the true dynamic state of the site, but also eliminates noise and time deviations through processing technology, thereby achieving overall system performance optimization and precise matching of energy consumption control.

[0021] Step 120: Input the pre-processed environmental parameters into a preset deep neural network model to determine the load prediction sequence of each lighting area.

[0022] Specifically, the preprocessed environmental parameters are input into a preset deep neural network model to achieve time-series prediction of lighting loads. This model, composed of multiple neural network layers containing temporal memory units, is capable of identifying time-dependent features in the input data. The model receives a standardized data matrix as input, uses the previous 24 hours of historical environmental conditions in each area as a time window, and utilizes a sliding prediction mechanism to output a sequence of lighting load trends over a preset future duration. A dynamic learning rate adjustment strategy is introduced during training to dynamically optimize network parameters based on prediction errors, improving prediction accuracy and generalization capabilities. This method enables reliable estimates of future load trends to be obtained before actual lighting demand occurs, providing a key basis for proactive adjustments to lighting control.

[0023] In an optional implementation of this embodiment, the preprocessed environmental parameters are input into a preset deep neural network model to determine the load prediction sequence of each lighting area, including: inputting the standardized data matrix into the preset deep neural network model to determine the data to be predicted by the deep neural network model; obtaining the historical data of each lighting area within a preset time length through a sliding window mechanism; inputting the historical data into the deep neural network model to output the load prediction value of each lighting area; performing error correction judgment on the load prediction value; if the prediction error of a preset number of time windows exceeds the error threshold, the load prediction value is dynamically corrected; the load prediction value is overwritten according to the prediction value corresponding to the corrected time window to generate a load prediction sequence for each lighting area.

[0024] Specifically, when inputting the standardized data matrix into the preset deep neural network model, the data matrix obtained by normalization and spatial aggregation is first used as the model input. This matrix has been numerically converted to a unified scale, allowing the network to eliminate inconsistencies between different parameter dimensions, thereby accurately identifying the time series characteristics of the environmental status in each lighting area and determining the model's predicted data, that is, the future lighting load change trend to be predicted. Then, the historical data of each lighting area within a preset time period is obtained through a sliding window mechanism. The sliding window mechanism refers to continuous sampling within a fixed time period and capturing information about data changes over time through continuous windows. The operational significance is to reflect the historical dynamics of the lighting area over a period of time. For example, in an office building, the population density and light intensity data of a certain area in the past 24 hours are integrated into continuous time series data, providing sufficient historical basis for load prediction. The collected historical data is then input into a deep neural network model, which contains memory units and nonlinear transformation modules. The model uses the time correlation contained in the historical data to output the load forecast value of each lighting area, thereby obtaining the lighting load estimation information for a period of time in the future. This process is similar to using neurons to learn historical patterns and obtain the expected load value through feedforward operations. Its significance lies in providing a basic prediction basis for subsequent load control decisions. After the predicted value is output, the load prediction value is subjected to error correction judgment. The error correction judgment is intended to identify the possible deviation between the predicted value and the true value in a continuous time window, and thereby quantify the prediction error. If the prediction error in a preset number of consecutive time windows exceeds the preset error threshold, the dynamic correction mechanism is activated, and the original prediction value is adjusted through the set dynamic correction strategy to improve the prediction accuracy. Its dynamic correction formula can be expressed as: ， in, represents the load forecast value after dynamic correction at time t, is the initial prediction value of the neural network, and Represent the historical actual load value and the predicted value at time j, is the normalization factor of the prediction error, k represents the number of continuous time windows involved in the calculation, and p is the parameter that adjusts the error impact. To correct the scale coefficient, the expression's significance lies in comprehensively considering the magnitude and distribution characteristics of the prediction error over several past moments, and deriving a dynamic correction value through power averaging to ensure that the predicted value is closer to the actual load changes. For example, if a certain area experiences a large prediction deviation over several consecutive time windows, the overall correction calculated in the formula will be large, and thus, by superimposing the correction terms, the final output prediction value will be more reasonable. Finally, the original load prediction value is overwritten with the prediction value corresponding to the corrected time window to form a load prediction sequence for each lighting area. This operation ensures that the final prediction sequence generated through dynamic correction data can better guide the control of lighting equipment in future applications, achieving load balancing and energy consumption optimization. Its practical significance lies in enabling intelligent lighting control systems to have the ability to self-correct and adapt to changing environments, thereby effectively reducing energy consumption and improving lighting comfort in practical applications.

[0025] Step 130: Determine the load status of each lighting area according to the load prediction sequence.

[0026] Specifically, the load status of each lighting area is determined based on the load prediction sequence. Specifically, the peak and average values ​​of each area within the prediction time period are analyzed to construct a dynamic load index that reflects the intensity of lighting usage. Combined with the spatial function type of the area and the current time period, a correction factor is introduced to perform weighted correction on the load index, so as to be closer to the actual lighting demand scenario. The corrected load index is then compared with the benchmark value in the historical database. If the continuous time period exceeds the set threshold, the area is identified as being in a high-load state. At the same time, based on the spatial topological relationship between regions, load diffusion constraints are imposed on adjacent areas to reflect the load conduction effect between regions, and finally a complete load status matrix is ​​formed as the basis for formulating dimming strategies.

[0027] In an optional implementation of this embodiment, the step of judging the load status of each lighting area according to the load prediction sequence includes: extracting the peak load value and the average load value of each lighting area in a preset future time period from the load prediction sequence; calculating the dynamic load index based on the difference between the peak load value and the average load value and the historical load fluctuation range; matching the dynamic correction factor according to the regional functional type and the current time period type; weightedly fusing the dynamic correction factor with the dynamic load index to generate a corrected load index; comparing the corrected load index with the load reference value of the same functional area in the same time period in the historical database; when the correction index continuously exceeds the load reference value, marking the corresponding lighting area as overloaded, and generating a load status matrix in combination with the regional topological relationship; applying load diffusion constraints to the overloaded lighting area according to the spatial correlation of adjacent areas, and generating a final load status matrix by adjusting the weight values ​​of adjacent areas.

[0028] Specifically, extracting the peak load and average load of each lighting area within a preset future time period from the load forecast sequence can accurately describe the future load fluctuations in that area. The peak load reflects the highest energy consumption level that the area may encounter, while the average load provides the overall energy consumption level over a period of time. This operation can be obtained through statistical calculations of the forecast data within the future time period. For example, if the forecast data in an office area fluctuates greatly, the peak value can reveal the risk of short-term high load, while the average value reflects the overall trend. Then, based on the difference between the peak value and the average value and combined with the historical load fluctuation range, the dynamic load index can be calculated as follows: , in, Represents the dynamic load index, P represents the peak load in a preset future time period, and A represents the average load in the same period. is the historical load fluctuation range, To avoid small constants with zero denominators, As a coefficient for adjusting sensitivity, this formula can nonlinearly amplify load fluctuations, thereby highlighting the sensitivity of large deviations to the system. For example, when the load in a certain area suddenly exceeds the normal fluctuation range, the index value will increase rapidly, indicating potential risks. Then, according to the regional functional type and the current time period type, a dynamic correction factor is matched through preset rules. This factor reflects the tolerance of different functional areas or different time periods to load fluctuations. For example, the conference area may require stricter load control during peak hours, while the corridor can be moderately relaxed. Matching is achieved through functional relationships to ensure that the load forecast results in the actual environment are closer to the actual situation. Next, the dynamic correction factor is weighted and fused with the dynamic load index calculated above to form a corrected load index. This process multiplies the previously obtained index by the correction factor for adjustment. Its significance is to further compensate for the original index and ensure that the load forecast is universal and accurate under different functions and time periods.

[0029] The generated corrected load index is then compared with the load benchmark value of the same functional area in the corresponding time period in the historical database. The historical load benchmark value is a standard energy consumption level obtained through long-term statistics and is used to measure whether the current load deviates from the expected range. During the comparison process, when the corrected index continuously exceeds the benchmark value, the corresponding area is marked as overloaded. On this basis, combined with the regional topological relationship, the spatial connection between regions is used to construct a load state matrix. Physically adjacent or functionally related regions are aggregated and analyzed to further reflect the load conduction effect between regions. Finally, by introducing the spatial correlation of adjacent regions, load diffusion constraints are imposed on the areas marked as overloaded. By adjusting the weights of adjacent regions, the overall load state is balanced and coordinated. The significance of this is to prevent the overload of a single area from evolving into the risk of imbalance in system performance consumption. For example, in a multi-area office building, adjacent areas can help each other alleviate local overloads under the load diffusion constraint, thus forming the final load state matrix, providing a scientific decision-making basis and implementation guarantee for the intelligent lighting system.

[0030] Step 140: Based on the load status and real-time environmental data, a multi-region collaborative dimming rate set is generated through a regional hierarchical collaborative strategy.

[0031] Specifically, based on load status and real-time environmental data, a regional hierarchical collaborative strategy is used to generate a multi-region collaborative dimming rate set. Within a local area, fuzzy rule matching is used to generate the initial dimming rate based on the current load status and environmental parameters (such as real-time light distribution and occupant density) to ensure that lighting adjustments are adapted to on-site needs. At the global level, the initial dimming rates of multiple regions are jointly optimized through swarm intelligence algorithms such as particle swarm optimization. During the optimization process, search parameters are dynamically adjusted to balance local lighting needs with overall energy-saving goals. Finally, based on regional priority and spatial correlation, the dimming rates are differentiated and fused to generate a global multi-region collaborative dimming rate set, providing an executable parameter set for lighting control.

[0032] In an optional implementation of this embodiment, based on the load status and real-time environmental data, the steps of generating a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy include: generating an initial dimming rate for each lighting area according to the weight value of the final load status matrix and the light intensity distribution in the real-time environmental data; performing inter-regional conflict detection on the initial dimming rate, and generating a conflict-free dimming rate set by adjusting the dimming rate difference of the conflicting area; applying a dynamic attenuation factor to the dimming rate of each area in the conflict-free dimming rate set according to the dynamic change rate of the population density in the real-time environmental data; performing particle swarm optimization on the attenuated conflict-free dimming rate set based on the regional topological relationship to generate a dynamic weighted dimming rate set; and hierarchically fusing the dynamic weighted dimming rate set according to regional priority and spatial correlation to generate a multi-region collaborative dimming rate set.

[0033] Specifically, the system generates an initial dimming rate for each lighting zone based on the zone's current load status and ambient lighting conditions. The load status is derived from the final load state matrix generated by the pre-process prediction module, which incorporates the inter-zone topology, overload detection logic, and spatial diffusion constraints. Ambient light data, derived from a real-time light sensor network, reflects the intensity and distribution of natural light. By integrating these two types of information, the system calculates a preliminary dimming rate for each lighting zone. This rate represents a preliminary balance between illumination demand and energy consumption. For example, in areas near large windows with abundant natural light, the initial dimming rate may be lower to minimize energy waste. In inner areas with dense crowds and high loads, the initial dimming rate may be higher. Subsequently, to avoid uneven lighting or sudden changes in light intensity at zone boundaries caused by varying dimming rates across zones, the system performs inter-zone dimming rate conflict detection. A conflict occurs when the dimming rate difference between two spatially adjacent zones is too large, causing user perception abruptness or visual fatigue. To resolve such conflicts, the system identifies conflicting areas and iteratively adjusts their dimming rates, mitigating any discrepancies. Ultimately, it generates a dimming set with minimal global dimming rate fluctuations. After conflict resolution, the system further applies a damping factor based on the dynamic rate of change in occupancy density, improving the dimming system's adaptability to changes in occupancy. Changes in occupancy density not only affect local illumination requirements but also the overall system's load regulation strategy. Therefore, when the crowd size rapidly increases or decreases, the dimming rates in corresponding areas should be dynamically adjusted. Specifically, dimming rates in densely populated areas will be appropriately maintained or increased to ensure lighting quality. In vacant or low-occupancy areas, a certain damping effect will be applied to the dimming rate to further minimize unnecessary energy consumption. To achieve a globally optimal dimming effect across all areas, the system uses a particle swarm optimization algorithm to deeply optimize the conflict-free and damping-corrected dimming set. Particle swarm optimization (PSO) is an optimization method that simulates the collaborative evolution of swarm intelligence. It simulates the iterative movement of multiple "dimming solution particles" in a search space. During each iteration, the particles gradually adjust their solutions by referring to their own historical optimal values ​​and the global optimal value of the swarm. This ensures that the dimming rate of each zone is not only influenced by its local state but also guided by the best solutions from the entire system, demonstrating the intelligence of global collaborative dimming. After the PSO completes optimization, the system performs hierarchical fusion based on zone priority (e.g., meeting areas are prioritized over aisles) and spatial topology (e.g., the density of connectivity between zones). This fusion process adjusts the contribution and response of each zone's dimming rate based on its importance within the system, ultimately generating a set of coordinated dimming rates for multiple zones. High-priority zones have greater room for dimming rate adjustment, while low-priority zones serve more as a buffer to adapt to the overall lighting strategy. Furthermore, spatial connectivity influences the fusion weights, with closely connected zones favoring synchronized and coordinated dimming behavior to avoid local discontinuities.Ultimately, the system outputs a set of coordinated, multi-zone dimming rates, representing the lighting system's optimal response to global energy efficiency, comfort, and dynamic adaptability, taking into account factors such as load forecasting, lighting conditions, occupant flow, and spatial structure. This control output is not only highly real-time and adaptive, but also effectively supports the needs of intelligent lighting control across multiple zones and in dynamic environments within large architectural spaces, demonstrating the deep integration of intelligent perception and coordinated control.

[0034] Furthermore, the dynamic weighted dimming rate set is hierarchically fused according to regional priority and spatial correlation to generate a multi-region collaborative dimming rate set, including: assigning a priority label to each lighting area based on the overload level weight value of the final load state matrix; dividing the dynamic weighted dimming rate set into different priority subsets according to the priority label; performing differentiated weight allocation on the global dimming rate baseline value set generated by the conflict-free dimming rate set and the local dimming rate compensation value set calculated by the edge node according to the different priority subsets according to a preset ratio; and superimposing the differentiatedly allocated global dimming rate and local dimming rate through a hierarchical fusion strategy to generate a multi-region collaborative dimming rate set.

[0035] Specifically, the edge node is a distributed decision-making unit corresponding to each lighting area. In this embodiment, the overload level weight value of each lighting area reflected in the final load state matrix is ​​used as the key indicator of the regional energy consumption load. A priority label is assigned to each area according to the weight value through a preset mapping rule. The priority label describes the importance and urgency of the area in the system dimming. For example, in the same building, the meeting area or temporary activity area may be given a higher priority, while the corridor or secondary functional area is given a lower priority, thereby providing a basic criterion for dimming control, and then the dynamic weight dimming rate set obtained by particle swarm optimization is divided into different priority subsets according to the priority label, so that the areas in each priority subset follow similar dimming logic in energy allocation; then, the system targets each The priority subset extracts a global dimming rate baseline value set generated by the conflict-free dimming rate set. This baseline value set is derived from large-scale data analysis and reflects the overall energy consumption target and lighting needs of the region. At the same time, the local dimming rate compensation value set calculated by the edge node uses the real-time detection data of the distributed decision-making unit in the local environment to reflect the local characteristics and real-time response requirements. Here, the edge node acts as a distributed controller for each lighting area, and through the synergy of local data and global strategies, more refined dimming control is achieved; furthermore, the global dimming rate baseline value and the local dimming rate compensation value are differentially weighted according to the preset ratio, so that the allocated dimming rate can maintain the overall energy consumption balance and meet the uneven characteristics of local lighting needs, thereby enabling the system to respond promptly to special situations in different regions. The formula for generating multi-region collaborative dimming rates through a layered fusion strategy can be expressed as: , in, represents the final multi-zone coordinated dimming rate of zone i, represents the global dimming rate baseline value of area i, represents the local dimming rate compensation value calculated by the edge node of area i, represents the priority label of region i, It is a priority mapping function whose value fluctuates between 0 and 1, and is used to adjust the contribution ratio of global and local dimming rates. The purpose of designing this function is to make high-priority areas more dependent on the global dimming rate baseline, while low-priority areas are more focused on local real-time correction. In the implementation process, for example, an office area is in a high-energy consumption state and has a high priority, so its priority mapping function value is large, so that the final dimming rate is mainly determined by the global strategy. In aisles with less traffic, the mapping function value is low, and local compensation plays a greater role. This fusion strategy ensures that the entire lighting system achieves the dual goals of energy optimization and visual comfort in a multi-zone collaborative situation through hierarchical and differentiated weight allocation, combined with global planning and local intelligent response.

[0036] Step 150: Perform lighting control on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

[0037] Specifically, the system controls the lighting devices in the target lighting area based on a multi-zone coordinated dimming rate set. The system converts the dimming rate value for each zone into a specific device control command and sends it to the lighting controller in each zone via an asynchronous communication protocol. These control commands include not only the target brightness value but also an embedded gradient rate parameter to control the smoothness of the lighting intensity change process.

[0038] In an optional implementation of this embodiment, the step of performing lighting control on the lighting equipment in the target lighting area according to the multi-area collaborative dimming rate set includes: obtaining lighting area data information according to the multi-area collaborative dimming rate set and the final load state matrix, and generating a control data message; performing multi-layer convolution processing on the control data message according to the global dimming rate baseline value set data stored at the edge node, and generating a joint prediction data structure reflecting the relationship between global and local dimming; performing a layered fusion operation on the global dimming rate baseline and the local dimming rate compensation data in the lighting area according to the joint prediction data structure and the regional topology relationship data, and generating an anti-conflict dimming instruction set; issuing control instructions to the lighting equipment in the target lighting area through the anti-conflict dimming instruction set; and performing distributed dimming operations on the lighting equipment according to the control instructions.

[0039] Specifically, based on the multi-area collaborative dimming rate set and the final load status matrix, the real-time dimming rate, load status, historical energy consumption records, area identity and spatial location of each lighting area are obtained from the system through the data interface, and a data message is constructed according to the predefined data format. The data message contains the dimming rate recommendation and load status identification of the lighting equipment, which is used to transmit to the distributed control module. The construction of the data message can provide unified data input for subsequent modules. For example, in a large office building, the real-time dimming data, load fluctuation, environmental parameters, etc. of each office, conference room, and corridor are packaged and transmitted according to a predetermined protocol to ensure data format standardization and information integrity. Next, the global dimming rate baseline value set stored in the edge node is called. This set is the result of long-term data collection and global load control analysis, and the control data message is sent to the multi-layer convolution processing module. Multi-layer convolution processing refers to the extraction, dimensionality reduction and fusion of features in the data message through a series of convolution filtering operations. The convolution layer in the convolutional neural network can capture the local and global correlation between the dimming rate and load status of each area. A joint prediction data structure is obtained through continuous filtering and activation function output. This data structure is saved in matrix form. The relationship between global and local dimming is clearly reflected in the joint prediction data. For example, comparing the similarities and differences in the dimming rates of conference rooms and corridors in adjacent areas provides a more detailed prediction basis. Subsequently, the joint prediction data structure and regional topological relationship data - regional topological relationship data describes the spatial connection, proximity and functional relevance between areas within the building - are input into the hierarchical fusion module. The hierarchical fusion operation performs a hierarchical weighted integration of the global dimming rate baseline and the local dimming rate compensation data. By setting the weights of each level, the data fusion results can reflect the hierarchical relationship between the overall lighting balance of the building and the compensation of local demand. The generated anti-conflict dimming instruction set contains detailed parameters for the dimming operation of each lighting area in the distributed network, such as target brightness, dimming change rate, dimming time window and other parameters. This ensures that the dimming rates of different areas will not suddenly change or conflict in value. For example, when there is incoordination between adjacent areas, the system can make the dimming rate transition smoothly through fusion adjustment. Finally, the anti-conflict dimming instruction set is transmitted to the lighting equipment in the target lighting area via the network communication module. After receiving the instruction, the lighting equipment performs distributed dimming operations according to the parameters contained in the instruction. The control unit on the lighting equipment adjusts the output brightness and power according to the instruction to achieve gradual changes in indoor lighting. This ensures that the entire lighting system can accurately transmit dimming instructions in a distributed control environment based on coordinated, layered data fusion and prediction, and complete load balancing and light environment control.

[0040] According to the load balancing method for intelligent lighting collaborative management provided by the present application, the environmental parameters of each lighting area are collected in real time and preprocessed; the preprocessed environmental parameters are input into a preset deep neural network model to determine the load prediction sequence of each lighting area; the load status of each lighting area is determined based on the load prediction sequence; based on the load status and real-time environmental data, a multi-region collaborative dimming rate set is generated through a regional hierarchical collaborative strategy; and lighting equipment in the target lighting area is controlled based on the multi-region collaborative dimming rate set. Through the implementation of the present application, intelligent dimming control driven by cross-regional dynamic perception and prediction is realized, which improves the dynamic adaptability and load control accuracy of the lighting system.

[0041] Figure 2 The embodiment of the present application provides a load balancing device for intelligent lighting collaborative management, which can be used to implement the load balancing method for intelligent lighting collaborative management in the above embodiment. Figure 2 As shown, the load balancing device for intelligent lighting collaborative management mainly includes: The processing module 10 is used to collect the environmental parameters of each lighting area in real time and pre-process the environmental parameters; A determination module 20 is configured to input the preprocessed environmental parameters into a preset deep neural network model to determine a load prediction sequence for each lighting area; A judgment module 30 is used to judge the load status of each lighting area according to the load prediction sequence; A generating module 40 is configured to generate a multi-region coordinated dimming rate set through a regional hierarchical coordination strategy based on load status and real-time environmental data; The control module 50 is configured to perform lighting control on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

[0042] In an optional implementation of this embodiment, the processing module is specifically used to: collect the environmental parameters of each lighting area in real time to generate an original environmental parameter set; traverse the original environmental parameter set through a sliding window mechanism, delete abnormal values ​​of data offset, and generate an abnormality-filtered data sequence; generate a time-synchronized multivariate data stream by time-aligning the data sequence; normalize the multivariate data stream, and spatially aggregate the processed multivariate data stream to generate a standardized data matrix.

[0043] Furthermore, in an optional implementation of the present embodiment, the determination module is specifically used to: input the standardized data matrix into a preset deep neural network model to determine the data to be predicted by the deep neural network model; obtain historical data of each lighting area within a preset time length through a sliding window mechanism; input the historical data into the deep neural network model to output the load prediction value of each lighting area; perform error correction judgment on the load prediction value; if the prediction error of a preset number of time windows exceeds the error threshold, dynamically correct the load prediction value; overwrite the load prediction value according to the prediction value corresponding to the corrected time window to generate a load prediction sequence for each lighting area.

[0044] In an optional implementation of this embodiment, the judgment module is specifically used to: extract the peak load value and the average load value of each lighting area in a preset future time period from the load prediction sequence; calculate the dynamic load index based on the difference between the peak load value and the average load value and the historical load fluctuation range; match the dynamic correction factor according to the regional functional type and the current time period type; weightedly fuse the dynamic correction factor with the dynamic load index to generate a corrected load index; compare the corrected load index with the load reference value of the same functional area in the same time period in the historical database; when the correction index continuously exceeds the load reference value, mark the corresponding lighting area as overloaded, and generate a load state matrix based on the regional topological relationship; impose load diffusion constraints on the overloaded lighting area according to the spatial correlation of adjacent areas, and generate the final load state matrix by adjusting the weight values ​​of adjacent areas.

[0045] Furthermore, in an optional implementation of the present embodiment, the generation module is specifically used to: generate an initial dimming rate for each lighting area based on the weight value of the final load state matrix and the light intensity distribution in the real-time environmental data; perform inter-regional conflict detection on the initial dimming rate, and generate a conflict-free dimming rate set by adjusting the dimming rate difference of the conflicting area; apply a dynamic attenuation factor to the dimming rate of each area in the conflict-free dimming rate set based on the dynamic change rate of the population density in the real-time environmental data; perform particle swarm optimization on the attenuated conflict-free dimming rate set based on the regional topological relationship to generate a dynamic weighted dimming rate set; and layer and fuse the dynamic weighted dimming rate set according to regional priority and spatial correlation to generate a multi-region collaborative dimming rate set.

[0046] Furthermore, in an optional implementation of the present embodiment, when the generation module performs the function of hierarchically fusing the dynamic weighted dimming rate set according to regional priority and spatial correlation to generate a multi-region collaborative dimming rate set, it is specifically used to: assign a priority label to each lighting area based on the overload level weight value of the final load state matrix; divide the dynamic weighted dimming rate set into different priority subsets according to the priority label; perform differentiated weight allocation on the global dimming rate baseline value set generated for the conflict-free dimming rate set and the local dimming rate compensation value set calculated by the edge node according to the different priority subsets according to a preset ratio; wherein the edge node is a distributed decision-making unit corresponding to each lighting area; and superimpose the differentiated allocated global dimming rate and the local dimming rate through a hierarchical fusion strategy to generate a multi-region collaborative dimming rate set.

[0047] In an optional implementation of this embodiment, the control module is used to: obtain lighting area data information based on the multi-area collaborative dimming rate set and the final load state matrix, and generate a control data message; perform multi-layer convolution processing on the control data message based on the global dimming rate baseline value set data stored at the edge node, and generate a joint prediction data structure reflecting the relationship between global and local dimming; perform a layered fusion operation on the global dimming rate baseline and local dimming rate compensation data in the lighting area based on the joint prediction data structure and the regional topology relationship data, and generate an anti-conflict dimming instruction set; issue control instructions to the lighting equipment in the target lighting area through the anti-conflict dimming instruction set; and perform distributed dimming operations on the lighting equipment according to the control instructions.

[0048] According to the load balancing device for intelligent lighting collaborative management provided by the present application, the environmental parameters of each lighting area are collected in real time and preprocessed; the preprocessed environmental parameters are input into a preset deep neural network model to determine the load prediction sequence of each lighting area; the load status of each lighting area is determined based on the load prediction sequence; based on the load status and real-time environmental data, a multi-region collaborative dimming rate set is generated through a regional hierarchical collaborative strategy; and lighting equipment in the target lighting area is controlled based on the multi-region collaborative dimming rate set. Through the implementation of the present application, intelligent dimming control driven by cross-regional, dynamic perception and prediction is realized, which improves the dynamic adaptability and load control accuracy of the lighting system.

[0049] According to the application plan Figure 3 An electronic device provided in an embodiment of the present application can be used to implement the load balancing method for intelligent lighting collaborative management in the aforementioned embodiment, mainly comprising: Memory 301, processor 302, and computer program 303 stored in memory 301 and executable on processor 302. Memory 301 and processor 302 are connected via communication. When processor 302 executes computer program 303, the load balancing method for intelligent lighting collaborative management described in the aforementioned embodiment is implemented. The number of processors can be one or more.

[0050] The memory 301 can be a high-speed random access memory (RAM) memory or a non-volatile memory such as a disk drive. The memory 301 is used to store executable program code. The processor 302 is coupled to the memory 301 .

[0051] Furthermore, the embodiment of the present application also provides a computer-readable storage medium, which can be provided in the electronic device in the above embodiments. The computer-readable storage medium can be the above Figure 3 Memory in the illustrated embodiment.

[0052] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the load balancing method for intelligent lighting collaborative management in the aforementioned embodiment. Furthermore, the computer-readable storage medium may be a USB flash drive, a mobile hard drive, a read-only memory (ROM), RAM, a magnetic disk, or an optical disk, among other media capable of storing program code.

[0053] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0054] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0055] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A load balancing method for intelligent lighting collaborative management, characterized in that: include: Collecting environmental parameters of each lighting area in real time and preprocessing the environmental parameters; Inputting the preprocessed environmental parameters into a preset deep neural network model to determine a load prediction sequence for each lighting area; Determining the load status of each lighting area according to the load prediction sequence; Based on the load status and real-time environmental data, generating a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy; Lighting control is performed on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

2. The load balancing method for intelligent lighting collaborative management according to claim 1, characterized in that: The step of collecting the environmental parameters of each lighting area in real time and preprocessing the environmental parameters includes: Collect environmental parameters of each lighting area in real time and generate a set of original environmental parameters; Traversing the original environmental parameter set through a sliding window mechanism, deleting abnormal values ​​of data offset, and generating an abnormality-filtered data sequence; Generating a time-synchronized multi-data stream by time-aligning the data sequences; The multivariate data stream is normalized, and the processed multivariate data stream is spatially aggregated to generate a standardized data matrix.

3. The load balancing method for intelligent lighting collaborative management according to claim 2, characterized in that: The step of inputting the pre-processed environmental parameters into a preset deep neural network model to determine the load prediction sequence of each lighting area includes: Inputting the standardized data matrix into a preset deep neural network model to determine the data to be predicted by the deep neural network model; Acquiring historical data of each lighting area within a preset time period through the sliding window mechanism; Inputting the historical data into the deep neural network model, and outputting a load prediction value for each lighting area; performing error correction judgment on the load prediction value; If the prediction error of a preset number of time windows exceeds the error threshold, the load prediction value is dynamically corrected; The load prediction value is overwritten according to the prediction value corresponding to the modified time window to generate a load prediction sequence for each lighting area.

4. The load balancing method for intelligent lighting collaborative management according to claim 1, characterized in that: The step of determining the load status of each lighting area according to the load prediction sequence includes: Extracting the peak load value and the average load value of each lighting area in a preset future time period from the load prediction sequence; Calculating a dynamic load index based on a difference between the peak load value and the average load value and a historical load fluctuation range; Match dynamic correction factors based on regional function type and current time period type; Performing weighted fusion of the dynamic correction factor and the dynamic load index to generate a corrected load index; Comparing the modified load index with a load benchmark value of the same functional area in the same time period in a historical database; When the correction index continuously exceeds the load reference value, the corresponding lighting area is marked as overloaded, and a load state matrix is ​​generated in combination with the regional topological relationship; A load diffusion constraint is imposed on the overloaded lighting area according to the spatial correlation of the adjacent areas, and a final load state matrix is ​​generated by adjusting the weight values ​​of the adjacent areas.

5. The load balancing method for intelligent lighting collaborative management according to claim 4, characterized in that: The step of generating a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy based on the load status and real-time environmental data includes: generating an initial dimming rate for each of the lighting areas according to the weight values ​​of the final load state matrix and the light intensity distribution in the real-time environmental data; Performing inter-area conflict detection on the initial dimming rate, and generating a conflict-free dimming rate set by adjusting the dimming rate difference of the conflicting areas; Applying a dynamic attenuation factor to the dimming rate of each area in the non-conflicting dimming rate set according to the dynamic change rate of the population density in the real-time environmental data; Performing particle swarm optimization on the attenuated conflict-free dimming rate set based on the regional topological relationship to generate a dynamic weighted dimming rate set; The dynamic weighted dimming rate set is hierarchically integrated according to regional priority and spatial correlation to generate a multi-region coordinated dimming rate set.

6. The load balancing method for intelligent lighting collaborative management according to claim 5, characterized in that: The step of hierarchically fusing the dynamic weighted dimming rate set according to regional priority and spatial correlation to generate a multi-region coordinated dimming rate set includes: assigning a priority label to each of the lighting areas based on the overload level weight value of the final load state matrix; dividing the set of dynamic weighted dimming rates into subsets of different priorities according to the priority labels; A global dimming rate baseline value set generated for the conflict-free dimming rate set according to the different priority subsets and a local dimming rate compensation value set calculated by the edge node are differentially weighted according to a preset ratio; wherein the edge node is a distributed decision-making unit corresponding to each of the lighting areas; The differentiated allocated global dimming rate and the local dimming rate are superimposed through a layered fusion strategy to generate the multi-region coordinated dimming rate set.

7. The load balancing method for intelligent lighting collaborative management according to claim 6, characterized in that: The step of performing lighting control on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set includes: Acquire lighting area data information according to the multi-area coordinated dimming rate set and the final load state matrix, and generate a control data message; performing multi-layer convolution processing on the control data message according to the global dimming rate baseline value set data stored at the edge node to generate a joint prediction data structure reflecting the relationship between global and local dimming; Performing a hierarchical fusion operation on the global dimming rate baseline and the local dimming rate compensation data within the lighting area according to the joint prediction data structure and the regional topology relationship data, and generating an anti-conflict dimming instruction set; issuing control instructions to the lighting equipment in the target lighting area using the anti-conflict dimming instruction set; Perform a distributed dimming operation on the lighting device according to the control instruction.

8. A load balancing device for intelligent lighting collaborative management, characterized in that: The load balancing device for intelligent lighting collaborative management includes: A processing module, used to collect environmental parameters of each lighting area in real time and pre-process the environmental parameters; a determination module, configured to input the preprocessed environmental parameters into a preset deep neural network model to determine a load prediction sequence for each of the lighting areas; a judgment module, configured to judge the load status of each lighting area according to the load prediction sequence; A generating module, configured to generate a multi-region collaborative dimming rate set through a regional hierarchical collaborative strategy based on the load status and real-time environmental data; A control module is used to perform lighting control on the lighting equipment in the target lighting area according to the multi-area coordinated dimming rate set.

9. An electronic device, characterized in that: Comprising a memory and a processor, wherein: The processor is configured to execute a computer program stored in the memory; When the processor executes the computer program, the steps in the load balancing method for intelligent lighting collaborative management described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the load balancing method for intelligent lighting collaborative management described in any one of claims 1 to 7 are implemented.

Images