AI-driven personalized customized lighting service method and system

Abstract

The invention relates to the technical field of intelligent lighting, in particular to an AI-driven personalized customized lighting service method and system, and the method comprises the steps: receiving a lighting customization request, containing scene demand information, of a user side, and determining a corresponding sub-lighting model as a target model from an AI lighting model library according to the lighting customization request; obtaining a personalized lighting interface corresponding to the target model, customizing and updating a user side lighting control interface according to the personalized lighting interface, and generating a user customized interface; the method comprises the following steps: receiving environment data and preference data of a user side through a personalized interface, sending the environment data and the preference data to an AI decision server of a target model corresponding to a cloud lighting platform, and splitting and analyzing the data by an environment sensing unit to obtain a plurality of sub-lighting parameter sets; and the illumination suitability of the sub-parameter set is verified through an illumination optimization unit, and a personalized illumination scheme is generated and pushed to a user-side interface for display. According to the method, through AI model driving and multi-data processing, personalized and precise adaptation of the lighting service is realized.

Description

Technical Field

[0001] This invention relates to the field of smart lighting technology, and in particular to an AI-driven method and system for personalized lighting services. Background Technology

[0002] With the development of smart IoT technology, lighting systems are evolving from traditional fixed models towards intelligent and personalized solutions. However, existing lighting services generally suffer from insufficient scene adaptability and delayed response to users' personalized needs. Traditional lighting systems typically provide fixed illumination parameters based on preset scenes, failing to dynamically perceive environmental changes and real-time user preferences. This leads to a disconnect between the lighting experience and actual needs, and may even cause visual fatigue and energy waste due to improper lighting.

[0003] Meanwhile, users' demands for lighting environments are becoming increasingly sophisticated. Different scenarios, time periods, and activities all place differentiated requirements on the brightness, color temperature, and dynamic adjustment modes of lighting. While some existing smart lighting systems achieve basic environmental perception through sensors, they lack the ability to perform deep data analysis and personalized model-driven approaches based on artificial intelligence, making it difficult to construct multi-layered, dynamic lighting adaptation solutions. For example, traditional solutions cannot generate customized lighting curves based on user behavior data, nor can they achieve intelligent linkage of lighting parameters across multiple scenarios through cross-device collaboration.

[0004] Furthermore, the data processing flow of existing lighting systems is relatively simple, lacking mechanisms for integrated analysis and hierarchical optimization of multi-source data. For example, it cannot effectively break down and analyze complex environmental parameters, resulting in insufficient accuracy in generating lighting schemes; at the same time, the lack of a multi-layered verification mechanism for lighting adaptability makes it difficult to ensure a balance between comfort and energy efficiency in practical applications.

[0005] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The main objective of this invention is to provide an AI-driven personalized lighting service method and system, which aims to solve the technical problems of existing intelligent lighting systems, such as insufficient scene adaptability, difficulty in dynamically responding to environmental changes and personalized user needs, and lack of AI-driven multi-source data deep analysis and precise lighting adaptation scheme generation capabilities.

[0007] To achieve the above objectives, the present invention provides an AI-driven method for personalized lighting services, the method comprising: Receive a lighting customization request from the user, the lighting customization request including scene requirement information, and determine the corresponding sub-lighting model in the AI ​​lighting model library as the target lighting model based on the scene requirement information; Obtain the personalized lighting interface corresponding to the target lighting model, and customize and update the lighting control interface on the user end according to the personalized lighting interface to generate a user-customized interface. According to the personalized lighting interface, the environmental data and preference data received from the user terminal are sent to the AI ​​decision server of the cloud lighting platform corresponding to the target lighting model. The environmental perception unit of the AI ​​decision server is called to split and parse the environmental data and preference data to obtain multiple sub-lighting parameter sets. The lighting optimization unit at the AI ​​decision server verifies the lighting adaptability of multiple sub-lighting parameter sets to obtain a personalized lighting scheme, which is then sent to the personalized lighting interface on the user's end for display.

[0008] Optionally, receiving a lighting customization request from a user, the lighting customization request including scene requirement information, and determining a corresponding sub-lighting model in the AI ​​lighting model library as the target lighting model based on the scene requirement information, includes: In response to the lighting customization request, obtain the user requirement tags corresponding to the scene requirement information. The user requirement tags include scene type tags, preference tags, and environment tags. The AI ​​lighting model library includes multiple sub-lighting models, and each sub-lighting model is set with a corresponding model function label; The sub-lighting model corresponding to the model function label of the user requirement label is determined as the target lighting model.

[0009] Optionally, the environmental perception unit at the AI ​​decision server retrieves and analyzes the environmental data and preference data to obtain multiple sub-lighting parameter sets, including: Based on the environmental perception unit, the environmental data and preference data are split and processed to obtain multiple sub-data modules, including a spatial data module, a temporal data module, a user behavior data module, and a natural light data module. Determine the baseline parameter set for the standard lighting scene and the adjustment weights corresponding to each of the sub-data modules; The sub-data modules corresponding to the benchmark parameter set are obtained as benchmark data modules. Standard illumination features are extracted from the benchmark data modules. The parameter extraction intervals in each sub-data module are obtained according to the maximum adjustment threshold of the standard illumination features and the adjustment weight. Parameter points whose parameter values ​​are within the lighting parameter threshold in the parameter extraction interval are identified as valid parameter points, and a set of valid parameter points is generated based on adjacent valid parameter points. Generate a sub-lighting parameter set corresponding to the sub-data module based on the set of valid parameter points; The sub-lighting parameter sets are arranged and analyzed from high to low according to the influence priority of each sub-data module to obtain the sub-lighting parameter sequence.

[0010] Optionally, the step of obtaining the sub-data module corresponding to the benchmark parameter set as the benchmark data module, extracting standard illumination features from the benchmark data module, and obtaining the parameter extraction interval in each sub-data module according to the maximum adjustment threshold of the standard illumination features and the adjustment weight includes: Retrieve the standard scene parameter template corresponding to the lighting customization request. The standard scene parameter template includes a core parameter area. Obtain the standard parameter area in the benchmark data module based on the core parameter area. Based on the lighting parameter threshold, extract the standard illumination features within the standard parameter region, and obtain the maximum adjustment threshold of the standard illumination features; Using the adjustment weight corresponding to each of the sub-data modules as the benchmark weight, the range of parameters in each of the sub-data modules that are positively correlated with the benchmark weight and are far from the maximum adjustment threshold of the benchmark weight is obtained as the division interval; The region corresponding to the baseline weight and the division interval in the sub-data module is obtained as the parameter extraction interval of the corresponding sub-data module.

[0011] Optionally, the step of extracting standard illumination features within the standard parameter region based on the illumination parameter threshold, and obtaining the maximum adjustment threshold of the standard illumination features, includes: Parameter points whose parameter values ​​are within the lighting parameter threshold within the standard parameter region are identified as standard parameter points. A set of standard parameter points is generated based on adjacent standard parameter points, and standard lighting features are generated based on the set of standard parameter points. Extract the standard adjustment curve of the standard illumination feature, and obtain the peak fluctuation amplitude of the standard adjustment curve as the maximum adjustment threshold corresponding to the standard illumination feature. The waveform of the standard adjustment curve is a time-parameter change curve.

[0012] Optionally, the step of verifying the lighting adaptability of multiple sub-lighting parameter sets based on the lighting optimization unit at the AI ​​decision server to obtain a personalized lighting scheme includes: Based on the illumination optimization unit, obtain the cooperative adjustment point corresponding to each sub-illumination parameter set in the sub-illumination parameter sequence; The sub-lighting parameter sets are fused according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set to obtain a primary lighting scheme; Extract the actual adjustment curve of the primary lighting scheme, locate the feature points of the standard adjustment curve based on the feature points of the actual adjustment curve, and superimpose the standard adjustment curve on top of the actual adjustment curve; The parameter points that overlap between the actual adjustment curve and the standard adjustment curve are obtained as matching parameter points. The number of matching parameter points and the total number of parameter points corresponding to the actual adjustment curve are counted. The parameter matching degree is obtained based on the ratio of the number of matching points to the total number. Identify the primary lighting schemes whose parameter matching degree is less than the benchmark matching degree as schemes to be optimized, and mark the set of sub-parameters that need to be adjusted in the schemes to be optimized; A primary lighting scheme whose parameter matching degree is greater than or equal to the baseline matching degree is determined as a secondary optimization scheme; The secondary optimization scheme and the standard illumination characteristics are adapted and verified to obtain the final lighting scheme or the scheme to be adjusted. Based on the final lighting scheme or the scheme to be adjusted, a personalized lighting scheme is obtained.

[0013] Optionally, obtaining the coordinated adjustment point corresponding to each sub-illumination parameter set in the sub-illumination parameter sequence based on the illumination optimization unit includes: The sub-lighting parameter sets are processed in a time sequence to extract multiple adjustment time periods corresponding to each sub-lighting parameter set; Obtain the maximum and minimum parameter values ​​corresponding to multiple adjustment time periods in the sub-lighting parameter set; The adjustment period corresponding to the maximum parameter value is determined as the peak adjustment period, and the adjustment period corresponding to the minimum parameter value is determined as the trough adjustment period; If the peak adjustment period and the trough adjustment period are adjacent, the boundary time of the adjacent periods is determined as the coordinated adjustment point; If the peak adjustment period and the trough adjustment period are not adjacent, then the end time of the peak adjustment period is taken as the first adjustment point, and the start time of the trough adjustment period is taken as the second adjustment point. Connect the first adjustment point and the second adjustment point to obtain the adjustment transition interval, and determine the midpoint of the adjustment transition interval as the corresponding cooperative adjustment point of the sub-lighting parameter set.

[0014] Optionally, the step of fusing the sub-lighting parameter sets according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set to obtain a primary lighting scheme includes: The first sub-lighting parameter set in the sub-lighting parameter sequence is obtained as the first fusion parameter set, and the sub-lighting parameter set adjacent to the first fusion parameter set is obtained as the second fusion parameter set; The point at which the first fusion parameter set and the second fusion parameter set are coordinated is determined as the fusion start point, and the point at which the second fusion parameter set and the first fusion parameter set are coordinated is determined as the fusion end point; Using the fusion starting point as a reference point, the parameters of the fusion ending point are controlled to transition towards the fusion starting point until the parameter change rates of the fusion starting point and the fusion ending point are consistent, at which point the transition of the fusion ending point stops. The fusion processing of the first fusion parameter set and the second fusion parameter set is completed based on the transition processing of the fusion start point and the fusion end point; The second fusion parameter set is updated to the first fusion parameter set, and the adjacent sub-lighting parameter sets of the updated first fusion parameter set are obtained as the next second fusion parameter set; Repeat the above steps of transition processing for the fusion start point and fusion end point, and perform fusion processing on each of the sub-lighting parameter sets to obtain a primary lighting scheme.

[0015] Optionally, the step of performing secondary adaptation verification on the secondary optimization scheme and the standard illumination characteristics to obtain the final illumination scheme or the scheme to be adjusted includes: Extract the real-time user feedback parameters from the secondary optimization scheme and the ideal feedback parameters from the standard illumination features; The feedback similarity verification is performed on the real-time feedback parameters and the ideal feedback parameters to obtain the feedback matching value. The secondary adaptation verification includes feedback similarity verification. The secondary optimization scheme with a feedback matching value greater than the benchmark feedback value is obtained as the adaptation scheme, and the adaptation scheme is determined as the final lighting scheme; The secondary optimization scheme with a feedback matching value less than the benchmark feedback value is identified as the scheme to be adjusted. Parameter correction suggestions are generated based on the difference parameters in the scheme to be adjusted. The secondary optimization scheme is then updated in conjunction with the parameter correction suggestions to obtain the final lighting scheme.

[0016] Furthermore, to achieve the above objectives, the present invention also provides an AI-driven personalized lighting service system, the system comprising: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the steps of the AI-driven personalized lighting service method as described in any of the above.

[0017] This invention provides an AI-driven method for personalized lighting services. The method receives scene requirement information and combines it with pre-set sub-lighting models in an AI lighting model library. Based on precise matching of user requirement tags and model function tags, it can quickly locate the target lighting model. For example, for office-meeting scenarios, the system automatically calls a model adapted for high brightness and cool color temperature, avoiding visual fatigue or distraction caused by excessively dim or warm lighting in traditional fixed modes. Through an environmental perception unit, environmental data and preference data are analyzed to generate a multi-level sub-lighting parameter set, enabling the lighting scheme to respond to environmental changes in real time. For example, the system automatically adjusts artificial lighting brightness based on the natural light intensity in the office, or switches to a warm light mode based on the user's nighttime reading preferences, improving the subtlety of scene adaptation. It generates personalized lighting interfaces based on target lighting models, customizing and updating the user-side control interface. The generated personalized lighting schemes are fed back to the user interface in real time, supporting parameter previews. Users can intuitively view the scheme's effects and adjust them as needed, enhancing the transparency and engagement of the interaction. The environmental perception unit breaks down and organizes complex data, extracting core features using standard scene parameter templates to ensure structured and efficient data processing. The lighting optimization unit ensures the scientific validity and reliability of the scheme through parameter matching degree calculations and secondary adaptation verification. For instance, if the parameter matching degree of a primary lighting scheme is lower than the benchmark value, the system automatically marks the sub-parameter set that needs adjustment and iteratively optimizes it through feedback data, reducing manual debugging costs and improving scheme generation efficiency. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an embodiment of the AI-driven personalized lighting service method of the present invention.

[0019] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0021] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the AI-driven personalized lighting service method of the present invention, which presents an embodiment of the AI-driven personalized lighting service method of the present invention.

[0022] In one embodiment, the AI-driven personalized lighting service method includes the following steps: Step S100: Receive a lighting customization request from the user. The lighting customization request includes scene requirement information. Based on the scene requirement information, determine the corresponding sub-lighting model in the AI ​​lighting model library as the target lighting model.

[0023] The AI ​​lighting model library can be a structured database composed of multiple pre-trained sub-lighting models. Each model is bound to specific functional labels and scene adaptation rules, enabling rapid matching and invocation of lighting solutions and supporting standardized responses to personalized lighting needs. In this embodiment, the AI ​​lighting model library can generate sub-models and label them with functional tags based on historical user behavior data, environmental feedback data, and expert lighting specifications through clustering and supervised learning. For example, the AI ​​lighting model library may include, but is not limited to, one or more of office scene lighting models, sleep scene lighting models, and reading scene lighting models. The target lighting model can be a single sub-model instance selected from the AI ​​lighting model library based on scene requirement information. It represents the configuration template for the current lighting task and can serve as a benchmark framework for generating lighting solutions, determining the parameter ranges for initial brightness, color temperature, and dynamic modes. The target lighting model can have an inclusion relationship with the AI ​​lighting model library, providing model instances for the target lighting model; it can also collaborate with the environmental perception unit to provide initial constraint boundaries for parameter generation.

[0024] Determining the target lighting model from the AI ​​lighting model library based on scene requirements can be achieved by parsing the semantic tags in the scene requirements information and matching them with the functional tags of each sub-model in the AI ​​lighting model library, selecting the model with the highest matching degree as the target. Furthermore, this operation can be implemented using a retrieval method combining keyword matching and semantic similarity calculation, such as TF-IDF weighted comparison with BERT semantic embedding, or by using a decision tree rule engine to filter sub-models that meet the conditions layer by layer according to scene priority. This allows for non-manual, low-latency localization of lighting solutions, avoiding the failure of generalization based on fixed scene presets.

[0025] Step S200: Obtain the personalized lighting interface corresponding to the target lighting model, and customize and update the lighting control interface on the user end according to the personalized lighting interface to generate a user-customized interface.

[0026] The personalized lighting interface can be a dynamically rendered template of the user-end control interface. Based on the target lighting model, it generates customized interactive elements and parameter visualization controls, which can enhance the user's perception of the lighting scheme's transparency and controllability, supporting preview and fine-tuning. In this embodiment, the personalized lighting interface can extract interface configuration metadata from the target lighting model, load a preset UI component library, and dynamically bind interactive controls such as parameter sliders, color temperature selectors, and mode switching buttons. The personalized lighting interface can be generated driven by the target lighting model, displaying the scheme output by the lighting optimization unit to the user end and transmitting user feedback back to the AI ​​decision server.

[0027] Step S300: Receive environmental and preference data from the user terminal based on the personalized lighting interface, send the environmental and preference data to the AI ​​decision server of the corresponding target lighting model on the cloud lighting platform, and retrieve the environmental perception unit at the AI ​​decision server to split and parse the environmental and preference data to obtain multiple sub-lighting parameter sets.

[0028] The environmental perception unit can be a data parsing module deployed in the AI ​​decision server. It separates and structures multi-source environmental and user preference data, transforming fuzzy perceptual inputs into a computable set of sub-lighting parameters to improve the dynamic response accuracy of the lighting scheme. In this embodiment, the environmental perception unit can receive ambient light intensity, temperature, humidity, timestamps, and user historical preference records from the sensor network, and separate independent parameter dimensions such as brightness, color temperature, and dynamic mode through feature extraction algorithms. For example, the environmental perception unit may include, but is not limited to, a natural light compensation parameter set, a user behavior preference parameter set, and a time-period adaptability parameter set. The environmental perception unit can collaborate with the AI ​​lighting model library, receiving the functional labels of the target lighting model as parsing constraints; it can also collaborate with the lighting optimization unit to output a sub-parameter set for adaptability verification.

[0029] The environmental perception unit at the AI ​​decision server decomposes and analyzes environmental and preference data to obtain multiple sub-lighting parameter sets. This can be achieved by decomposing multi-source input data into independent parameter groups according to preset dimensions, with each group corresponding to a light adjustment variable. Furthermore, this operation can be implemented by using principal component analysis to separate highly correlated variables between ambient light intensity and user preferences, extracting independent adjustment factors, or by using a rule engine to classify data by source, such as classifying sensor data into the natural light compensation set and user history records into the preference adjustment set. This allows for fine-grained decoupling of lighting parameters, supports multi-dimensional dynamic responses, and improves the subtlety of environmental adaptation.

[0030] Step S400: The lighting adaptation of multiple sub-lighting parameter sets is verified by the lighting optimization unit at the AI ​​decision server to obtain a personalized lighting scheme, and the personalized lighting scheme is sent to the personalized lighting interface on the user terminal for display.

[0031] The illumination optimization unit can be a parameter verification and iteration module located in the AI ​​decision server. It performs secondary verification and correction on the sub-illumination parameter set based on a matching degree threshold, ensuring that the generated personalized lighting scheme meets multi-dimensional standards of comfort and energy efficiency, and reducing the probability of invalid output. In this embodiment, the illumination optimization unit can compare the sub-parameter set output by the environmental perception unit with the functional labels of the target lighting model item by item, calculate the matching degree index, and trigger a feedback correction mechanism for subsets below the threshold. For example, the illumination optimization unit may include, but is not limited to, a parameter consistency verification module, a feedback iteration triggering module, and a scheme reliability scoring module. The illumination optimization unit can form a closed loop with the environmental perception unit: receiving its output and providing feedback correction instructions; it can also collaborate with the target lighting model to verify whether the output meets preset functional boundaries.

[0032] The AI ​​decision server's lighting optimization unit verifies the lighting adaptability of multiple sub-lighting parameter sets to obtain a personalized lighting scheme. This can be achieved by comparing each sub-parameter set with the functional labels of the target lighting model, calculating the overall matching degree, and triggering corrective feedback for low-match items. Furthermore, this operation can be implemented using a weighted scoring mechanism, assigning weights to dimensions such as brightness, color temperature, and dynamic rhythm, and comprehensively calculating the matching degree. Alternatively, fuzzy logic rules can be introduced to perform fuzzy matching and tolerance adjustment of boundary parameters. This allows for the construction of a closed-loop optimization path, automatically correcting scheme deviations, and improving the scientific rigor and reliability of the final output.

[0033] Taking an office lunchtime meeting as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the user submits an office-meeting scenario request, and the system matches a sub-model with high brightness, cool color temperature, and no flicker from the AI ​​lighting model library as the target lighting model; the personalized lighting interface is then updated to a customized panel that includes a brightness slider, color temperature selector, and mode lock button; the environmental perception unit receives the natural light intensity increase signal from the window light sensor and the user's recent preference records, and separates the natural light compensation parameter set and the user preference parameter set; the lighting optimization unit compares the two sets of parameters with the cool color temperature and high brightness requirements of the target model, and finds that the brightness value in the natural light compensation parameter is too high, resulting in a decrease in the overall matching degree, automatically reduces the artificial lighting output and marks the subset to be optimized; the corrected solution is pushed to the user interface, and the user can preview the effect and confirm it, and the system completes the closed-loop adjustment.

[0034] This embodiment provides an AI-driven personalized lighting service method. By receiving lighting customization requests from users and matching target lighting models based on scene requirements, the method achieves dynamic customization of the control interface through a personalized lighting interface. An environmental perception unit decouples and structures multi-source data to generate sub-parameter sets, and a lighting optimization unit verifies the matching degree of the sub-parameter sets and provides feedback correction. This achieves a complete process from static template calling, dynamic parameter generation to adaptive verification, breaking through the limitations of traditional lighting systems that rely on presets, have delayed responses, and lack feedback. It achieves a collaborative operation mechanism for personalized, dynamic, and automated lighting services.

[0035] In some embodiments, a lighting customization request from a user terminal is received. The lighting customization request includes scene requirement information. Based on the scene requirement information, a corresponding sub-lighting model in the AI ​​lighting model library is determined as the target lighting model, including: In response to the lighting customization request, obtain the user requirement tags corresponding to the scene requirement information. The user requirement tags include scene type tags, preference tags, and environment tags. The AI ​​lighting model library includes multiple sub-lighting models, and each sub-lighting model is set with a corresponding model function label; The sub-lighting model corresponding to the model function label of the user requirement label is determined as the target lighting model.

[0036] In this embodiment, user demand tags can be structured request descriptors composed of three semantic dimensions: scene type, user preferences, and environmental conditions. These descriptors represent the user's intent regarding lighting services and can transform ambiguous user requests into a computable, standardized set of tags, improving the accuracy of demand understanding and the traceability of system responses. In this embodiment, user demand tags can be collected through the user's input interface, including scene type selection, preference settings, and environmental context options, or aggregated tags can be extracted from historical behavior logs to form triplet structured data. In this embodiment, user demand tags can serve as the matching input source for model function tags, driving the selection of the target lighting model; they also form a mapping relationship with scene demand information, representing its structured expression. For example, user demand tags may include, but are not limited to, scene type tags, preference tags, and environmental tags.

[0037] Model function tags can be semantic attribute identifiers bound to each sub-lighting model, used to describe the supported lighting parameter combinations and applicable conditions. They can serve as index identifiers for models in the AI ​​lighting model library, achieving semantic alignment between user needs and model capabilities. In this embodiment, model function tags can be manually or automatically labeled during the model training phase based on typical scene features covered by its parameter space, such as a high brightness-cool color temperature-flicker-free combination corresponding to a conference mode. In this embodiment, model function tags can form a mapping and matching relationship with user need tags, serving as a direct basis for selecting target lighting models; they also constitute an attribute binding relationship with the AI ​​lighting model library. For example, model function tags can include, but are not limited to, brightness level tags, color temperature preference tags, and dynamic adjustment mode tags.

[0038] The target lighting model can be a single sub-model instance selected from the AI ​​lighting model library based on scene requirements. It represents the configuration template for the current lighting task and can be used as a baseline framework for generating lighting schemes, determining the parameter ranges for initial brightness, color temperature, and dynamic modes. In this embodiment, the target lighting model can be determined by the matching results of user requirement tags and model function tags, and is the output instance of the tag mapping mechanism.

[0039] Obtaining user demand tags corresponding to scene requirements can be achieved by parsing semantic fields from lighting customization requests, extracting three types of information—scene type, preference settings, and environmental context—according to a preset structure, and combining them into a structured tag set. Furthermore, user demand tags corresponding to scene requirements can be generated directly through options in front-end interactive components, such as selecting the preference tag "warm light" and the environment tag "low ambient light" for nighttime reading, or extracting high-frequency combination tags from user historical behavior logs, such as automatically completing the preference tag "warm light" if the user has used warm light for three consecutive nights of reading. This transforms unstructured requests into standardized semantic units, providing computable input for subsequent tag matching and improving the robustness of demand recognition.

[0040] Determining the sub-lighting model corresponding to the model function label of the user demand label as the target lighting model can be achieved by comparing each sub-label in the user demand label with the model function label of each sub-model in the AI ​​lighting model library, selecting the sub-model that fully matches or has the best coverage as the target. Further, determining the sub-lighting model corresponding to the model function label of the user demand label as the target lighting model can be achieved by using an intersection matching algorithm, requiring that the scene type label and preference label must match completely, while allowing partial overlap in the environment label; or by constructing a label weight graph, assigning higher matching weights to preference labels, and prioritizing matching models with high-weight dimensions. This allows for fine-grained model selection based on semantic dimensions, avoiding coarse-grained matching that relies solely on scene names, and improving the personalized accuracy of the initial solution.

[0041] Taking a user's nighttime reading request as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the user selects a nighttime reading scene on a mobile device, and the system automatically extracts scene type tags: reading, preference tags: warm light, and environment tags: low ambient light as user requirement tags; the model function tags of the reading scene lighting model in the AI ​​lighting model library include color temperature preference: warm light, brightness level: medium-low, and dynamic adjustment: slow gradient. These three tags completely match the user requirement tags, and the system selects it as the target lighting model; after the model is loaded, it drives the personalized lighting interface to display warm smooth blocks and brightness fine-tuning controls, providing accurate initial parameter boundaries for the subsequent environmental perception unit.

[0042] This embodiment provides an AI-driven personalized lighting service method. By receiving a lighting customization request from a user and obtaining user requirement tags corresponding to scene requirement information, the method determines the target lighting model based on the matching relationship between the user requirement tags and the model function tags in the AI ​​lighting model library. This shifts the initial selection of lighting solutions from a real-time response dependent on environmental sensing to a pre-loading matching based on intent tags. A highly fitting initial solution can be generated even without real-time data, replacing the coarse-grained calls based on fixed scene names or simple thresholds in traditional systems. This transforms lighting configuration from a passive response to an active prediction, providing a high-quality, low-bias starting point for subsequent fine-grained optimization of the environmental sensing unit, and significantly improving the overall personalized response efficiency and solution consistency of the system.

[0043] In some embodiments, the environmental perception unit at the AI ​​decision server is invoked to split and parse the environmental data and preference data to obtain multiple sub-lighting parameter sets, including: the environmental perception unit splits and processes the environmental data and preference data to obtain multiple sub-data modules, including spatial data modules, temporal data modules, user behavior data modules and natural light data modules.

[0044] The environmental perception unit can be a multi-dimensional data structuring module deployed in an AI decision server. It is responsible for decomposing raw environmental and preference data into semantically defined sub-data modules. This allows for the semantic mapping of unstructured perception data to computable lighting parameter units, supporting subsequent parameter range extraction and priority ranking. In this embodiment, the environmental perception unit can receive raw data streams from multiple types of sensors and user interaction logs. Through preset classification rules and feature annotation mechanisms, it divides the data into four sub-modules: spatial, temporal, user behavior, and natural light. In this embodiment, the environmental perception unit can provide an input source for a baseline parameter set of a standard lighting scene; output classified sub-data modules to a set of valid parameter points; and drive the generation of sub-lighting parameter sequences. For example, the environmental perception unit can include, but is not limited to, one or more of spatial data modules, temporal data modules, user behavior data modules, and natural light data modules.

[0045] Sub-data modules can be independent data units output by the environmental perception unit, divided according to data semantic dimensions. Each module corresponds to a type of environmental or user influencing factor, and can be used to achieve isolated processing of multi-source data, avoid parameter cross-interference, and support independent adjustment and priority sorting by dimension. In this embodiment, sub-data modules can be classified and aggregated according to data source and semantic attributes. For example, spatial data comes from multi-point illuminance sensors, temporal data comes from system clock and work-rest model, user behavior data comes from historical preference records, and natural light data comes from environmental spectral sensors. In this embodiment, sub-data modules can serve as reference objects for the baseline parameter set; participate in the construction of parameter extraction intervals; be the direct source of the effective parameter point set; and determine the constituent elements of the sub-illumination parameter sequence. For example, sub-data modules can include, but are not limited to, one or more of spatial data modules, temporal data modules, user behavior data modules, and natural light data modules.

[0046] Determine the baseline parameter set for the standard lighting scene and the adjustment weights corresponding to each sub-data module.

[0047] The baseline parameter set for standard lighting scenarios can be a predefined, standardized set of parameters representing typical lighting scenarios. This serves as a reference baseline for adjusting parameters in sub-data modules, providing contextual constraints for parameter extraction and ensuring that the sub-parameter set fluctuates within a reasonable physiological and energy efficiency range. In this embodiment, the baseline parameter set for standard lighting scenarios can be based on historical best lighting schemes, lighting health standards, and industry specifications. High-frequency parameter combinations are extracted through cluster analysis to form a static baseline template. In this embodiment, the baseline parameter set for standard lighting scenarios can serve as the calculation benchmark for the parameter extraction interval; together with the adjustment weights, it determines the adjustment range boundaries of each sub-data module. For example, the baseline parameter set for standard lighting scenarios can adopt benchmark parameter sets for office scenarios, sleep scenarios, reading scenarios, and meeting scenarios, etc.

[0048] Adjustment weights can be dynamic influence coefficients assigned to each sub-data module, used to quantify their contribution to the final lighting scheme in the current scenario. They can be used to achieve differentiated control of multi-dimensional parameters, enabling the system to prioritize responses to environmental or behavioral factors with greater impact. In this embodiment, adjustment weights can be obtained through machine learning model training, with historical user feedback and energy efficiency data as input, and the output being the partial derivative weights of each sub-data module's comfort score. In this embodiment, adjustment weights can collaboratively define parameter extraction intervals with a baseline parameter set; influence the selection tendency of the effective parameter point set; and determine the sorting criteria for the sub-lighting parameter sequence. For example, adjustment weights can include, but are not limited to, spatial influence weights, temporal influence weights, user behavior influence weights, and natural light influence weights.

[0049] The sub-data modules corresponding to the benchmark parameter set are obtained as benchmark data modules. Standard illumination features are extracted from the benchmark parameter set. The parameter extraction range in each sub-data module is obtained according to the maximum adjustment threshold and adjustment weight of the standard illumination features.

[0050] This process involves extracting standard illumination features from a set of reference parameters. Based on the maximum adjustment threshold and adjustment weight of these features, the parameter extraction range for each sub-data module is obtained. This can be achieved by using the reference parameters as the origin and combining the adjustment weights and maximum allowable offsets of each sub-data module to calculate the adjustable parameter range for each module. Furthermore, this operation can be implemented by multiplying the reference brightness value by the adjustment weight and then expanding it ± the maximum adjustment threshold to form a brightness parameter range; or, for the natural light data module, by using the reference color temperature as a baseline and scaling the ambient spectral variation amplitude according to the weights to generate a color temperature adjustment range. This transforms the static reference into a dynamically adjustable range, ensuring that parameter adjustments remain within physiological and equipment safety limits.

[0051] Parameter points whose parameter values ​​are within the lighting parameter threshold in the parameter extraction interval are identified as valid parameter points, and a set of valid parameter points is generated based on adjacent valid parameter points.

[0052] Specifically, parameter points whose values ​​fall within the lighting parameter thresholds within the parameter extraction interval are identified as valid parameter points. A set of valid parameter points is generated based on adjacent valid parameter points. This can be achieved by sampling within the parameter extraction intervals of each sub-data module, filtering out continuous parameter points that meet the physical and physiological thresholds of the lighting system, and aggregating them into a continuous interval. Furthermore, this operation can be implemented by retaining continuous sampling points within the 300–1500 lux range for luminance parameter points and merging adjacent points to form a valid interval; or by retaining continuous samples within the 2700K–6500K range for color temperature parameter points that meet healthy lighting standards, thus constructing a valid color temperature point set. This allows for the elimination of invalid parameters that exceed the physiological comfort range or the equipment's capabilities, ensuring the feasibility and safety of the output solution.

[0053] Generate the corresponding sub-lighting parameter set for the sub-data module based on the set of valid parameter points.

[0054] The effective parameter point set can be an ordered set of continuous parameter points within the parameter extraction interval that meet the lighting parameter threshold. It represents a legally operable adjustment interval and can be used to filter noise and invalid fluctuations, retaining feasible parameter combinations that meet physiological comfort and device capabilities. In this embodiment, the effective parameter point set can traverse continuous sampling points within the parameter extraction interval of each sub-data module, filter out points located within the lighting parameter threshold boundary, and merge adjacent points to form a continuous interval. In this embodiment, the effective parameter point set can be generated jointly by the parameter extraction interval and the lighting parameter threshold; it serves as a direct constituent unit of the sub-lighting parameter set. For example, the effective parameter point set may include, but is not limited to, effective point sets for luminance, color temperature, dynamic rhythm, and illuminance uniformity.

[0055] The sub-lighting parameter sets are arranged and analyzed from high to low according to the influence priority of each sub-data module to obtain the sub-lighting parameter sequence.

[0056] The sub-lighting parameter sequence can be an ordered sequence of multiple sub-lighting parameter sets arranged from high to low priority, representing the priority order of the system's lighting adjustment. It can be used to achieve coordinated scheduling of multi-dimensional parameters, ensuring that high-priority factors dominate the final output and avoiding parameter conflicts. In this embodiment, the sub-lighting parameter sequence can sort the effective parameter point sets corresponding to each sub-data module according to the adjustment weight, arranging the parameter sets in descending order of weight to form an execution sequence. In this embodiment, the sub-lighting parameter sequence can be generated by sorting the effective parameter point sets according to priority; it serves as the final input for the illumination optimization unit in scheme generation. For example, the sub-lighting parameter sequence may include, but is not limited to, spatially dominant parameter sequences, user behavior dominant parameter sequences, natural light dominant parameter sequences, and time-dominant parameter sequences.

[0057] Taking a family reading scenario in the evening as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: when a user enters the bedroom and turns on the reading mode, the environmental perception unit breaks down the data into: spatial data (illuminance at the bedside lamp position), temporal data (dusk hours after sunset), user behavior data (preferred warm light at night in the past week), and natural light data (intensity of residual light outside the window). The system retrieves the baseline parameter set for the reading scenario and sets the adjustment weights as follows: user behavior 0.4, natural light 0.3, time 0.2, and space 0.1. Based on the baseline color temperature of 3000K and the maximum adjustment threshold ±500K, the system generates a color temperature extraction range of 2500K–3500K. Within this range, continuous parameter points that meet the healthy lighting threshold (2700K–3400K) are selected to form an effective color temperature point set. After sorting by weight, the warm light parameter set dominated by user behavior is ranked first, followed by natural light compensation, generating a sub-lighting parameter sequence. This sequence is sent to the lighting optimization unit to generate a final personalized lighting scheme with warm light, low brightness, and gradual transition.

[0058] This embodiment provides an AI-driven personalized lighting service method. It uses an environmental perception unit to split environmental and preference data into multiple sub-data modules, determining a baseline parameter set for a standard lighting scene and the corresponding adjustment weights for each sub-data module. Standard illumination features from the baseline parameter set are extracted, and a parameter extraction interval is obtained by combining the maximum adjustment threshold and adjustment weights. Continuous parameter points within the parameter extraction interval that meet the lighting parameter thresholds are selected to form a valid parameter point set. Sub-lighting parameter sets are generated based on the valid parameter point set, and then sorted according to their impact priority to generate a sub-lighting parameter sequence. This achieves structured isolation of multi-source data through semantic splitting, constructs a legal adjustment interval through baseline constraints and dynamic weights, retains physiological and equipment-feasible parameters through threshold filtering, and ensures that key factors dominate the output through priority sorting, forming a hierarchical and interpretable processing link from initial perception to an executable solution. This mechanism enables the system to accurately identify complex needs such as user preferences plus reduced natural light, dynamically generating parameter combinations that balance physiological comfort and energy efficiency. It solves the problems of ambiguous response and collaborative failure caused by mixed parameters and lack of priority in traditional systems, significantly improving the refinement, reliability, and interpretability of the solution.

[0059] In one embodiment, the sub-data module corresponding to the benchmark parameter set is obtained as the benchmark data module, the standard illumination features in the benchmark data module are extracted, and the parameter extraction range in each sub-data module is obtained according to the maximum adjustment threshold and adjustment weight of the standard illumination features. This includes: retrieving the standard scene parameter template corresponding to the lighting customization request. The standard scene parameter template includes a core parameter area, and the standard parameter area in the benchmark data module is obtained according to the core parameter area.

[0060] The standard scene parameter template can be a predefined, structured configuration template that stores the core parameters of typical lighting scenarios. It includes an expandable parameter area and adjustment constraint rules, providing a standardized reference framework for parameter extraction and ensuring that personalized solutions do not deviate from basic lighting specifications. In this embodiment, the standard scene parameter template can be generated based on industry lighting standards and historical best-practice clustering. Each template is bound to a scene tag and includes a core parameter area and adjustable parameter boundary descriptions. In this embodiment, the standard scene parameter template can serve as the carrier of the core parameter area, providing contextual basis for the maximum adjustment threshold and adjustment weight, and driving the initial definition of the parameter extraction range. For example, the standard scene parameter template may include one or more of the following: office scene parameter template, sleep scene parameter template, reading scene parameter template, and meeting scene parameter template.

[0061] The core parameter area can be a fixed set of parameters in a standard scene parameter template used to define typical lighting conditions. It represents the baseline lighting characteristics of the scene and can be used as an anchor point for parameter adjustment, ensuring that personalized adjustments are made within the range that conforms to the essence of the scene. In this embodiment, the core parameter area can extract preset key parameter values ​​such as brightness, color temperature, and dynamic mode from the template. These values ​​are derived from expert specifications and user satisfaction verification data. In this embodiment, the core parameter area is the direct source of standard lighting characteristics and, together with the maximum adjustment threshold, constitutes the starting point for calculating the parameter extraction interval. For example, the core parameter area may include a baseline brightness area, a baseline color temperature area, a baseline illuminance uniformity area, and a baseline dynamic frequency area.

[0062] Based on the lighting parameter threshold, extract the standard illumination features within the standard parameter region, and obtain the maximum adjustment threshold of the standard illumination features; The maximum adjustment threshold can be the maximum allowable deviation from standard illumination characteristics, used to define the safety boundary for parameter adjustment. This prevents parameters from deviating excessively from the baseline, ensuring the output scheme meets physiological comfort and equipment capability constraints. In this embodiment, the maximum adjustment threshold can be based on human visual comfort experimental data and the upper limit of the physical performance of the lighting equipment, setting a unidirectional or bidirectional maximum adjustable range for each core parameter. In this embodiment, the maximum adjustment threshold and adjustment weight are calculated together to divide the interval, jointly determining the endpoints of the parameter extraction interval with the core parameter area. For example, the maximum adjustment threshold may include one or more of the following: maximum brightness adjustment threshold, maximum color temperature adjustment threshold, maximum dynamic rhythm adjustment threshold, and maximum illuminance uniformity adjustment threshold.

[0063] Using the adjustment weight corresponding to each of the sub-data modules as the benchmark weight, the range of parameters in each of the sub-data modules that are positively correlated with the benchmark weight and are far from the maximum adjustment threshold of the benchmark weight is obtained as the division interval; The adjustment weights can be dynamic influence coefficients assigned to each sub-data module, used to quantify their contribution to the final lighting scheme in the current scenario. They can be used to achieve differentiated control of multi-dimensional parameters, enabling the system to prioritize responses to environmental or behavioral factors with greater impact. In this embodiment, the adjustment weights can be obtained through machine learning model training, with historical user feedback and energy efficiency data as input, and the output being the partial derivative weights of each sub-data module's comfort score. In this embodiment, the adjustment weights serve as benchmark weights in the calculation of the interval division, jointly determining the expansion range of the parameter extraction interval with the maximum adjustment threshold. For example, the adjustment weights can include one or more of the following: spatial influence weights, temporal influence weights, user behavior influence weights, and natural light influence weights.

[0064] The region corresponding to the baseline weight and the division interval in the sub-data module is obtained as the parameter extraction interval of the corresponding sub-data module.

[0065] The parameter extraction interval can be a range of acceptable parameter values ​​within each sub-data module, calculated jointly by a baseline parameter, a maximum adjustment threshold, and an adjustment weight. It can be used to filter invalid or out-of-limit parameters, retaining only valid adjustment intervals that conform to scenario specifications and individual preferences. In this embodiment, the parameter extraction interval can be based on the core parameter area as the origin, scaling the maximum adjustment threshold according to the adjustment weight to form the adjustable parameter interval for each sub-data module. In this embodiment, the parameter extraction interval is generated collaboratively by the core parameter area, the maximum adjustment threshold, and the adjustment weight, serving as the input source for the set of valid parameter points. For example, the parameter extraction interval may include one or more of the following: brightness parameter extraction interval, color temperature parameter extraction interval, dynamic rhythm parameter extraction interval, and illuminance uniformity parameter extraction interval.

[0066] The system retrieves the standard scene parameter template corresponding to the lighting customization request. Based on the core parameter area, it obtains the standard parameter region from the benchmark data module. This can be achieved by parsing the scene tags in the user request, matching them with the corresponding standard scene parameter template, and extracting the benchmark parameter values ​​defined in its core parameter area. Furthermore, retrieving the standard scene parameter template corresponding to the lighting customization request and obtaining the standard parameter region from the benchmark data module can quickly locate the template bound to the scene tag using a hash index, read the pre-stored JSON format core parameter structure, or use a semantic matching algorithm to map ambiguous scene descriptions to the closest standard template implementation. This establishes a connection between personalized requests and standardized benchmarks, ensuring adjustments are based on evidence and avoiding randomness in parameter generation.

[0067] Based on the lighting parameter thresholds, standard illumination characteristics within the standard parameter area are extracted to obtain the maximum adjustment threshold for these standard illumination characteristics. This can be achieved by combining the parameter values ​​in the core parameter area with hard thresholds related to lighting health and equipment capabilities to calculate the maximum allowable deviation of each parameter. Furthermore, obtaining the maximum adjustment threshold for these standard illumination characteristics can be achieved by: for a reference color temperature of 3000K, if the lower limit for physiological comfort is 2700K and the upper limit for equipment is 3500K, then the maximum adjustment threshold is ±300K; or for a reference brightness of 500 lux, if the minimum output of the equipment is 200 lux and the maximum is 1000 lux, then the maximum adjustment threshold is ±300 lux. This transforms subjective comfort standards into calculable physical boundaries, ensuring that parameter adjustments remain within safe and compliant ranges.

[0068] Using the adjustment weights corresponding to each sub-data module as baseline weights, the parameter ranges in each sub-data module that are positively correlated with the baseline weights and are far from the maximum adjustment threshold of the baseline weights are defined as the dividing intervals. This can be achieved by multiplying the adjustment weights as a scaling factor by the maximum adjustment threshold to obtain the parameter extension range of each sub-data module. Furthermore, using the adjustment weights corresponding to each sub-data module as baseline weights, defining the parameter ranges in each sub-data module that are positively correlated with the baseline weights and are far from the maximum adjustment threshold of the baseline weights as dividing intervals can be achieved by, for example, if the user behavior weight is 0.5 and the maximum color temperature adjustment threshold is ±500K, then the dividing interval is ±250K; or if the natural light weight is 0.7, then its brightness dividing interval is 70% of the maximum adjustment threshold. This allows for dynamic weighting of parameter adjustment amplitudes, giving high-influence modules more adjustment space and improving response sensitivity.

[0069] The region corresponding to the baseline weight and the division interval in the sub-data module is used as the parameter extraction range for that sub-data module. This can be achieved by centering on the baseline value of the core parameter area and expanding outwards along the division interval to form the final range of selectable parameters. Furthermore, obtaining the region corresponding to the baseline weight and the division interval in the sub-data module as the parameter extraction range can be achieved by adding a baseline color temperature of 3000K to the division interval ±250K, resulting in a parameter extraction range of 2750K–3250K; or by adding a baseline brightness of 500 lux to the division interval ±210 lux, resulting in a parameter extraction range of 290 lux–710 lux. This generates parameter ranges with physical meaning and user preference guidance, providing a clear input range for subsequent effective parameter point selection.

[0070] Taking nighttime home reading as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: when a user requests nighttime reading, the system retrieves a reading scene parameter template and extracts the core parameter areas as color temperature 3000K and brightness 400 lux; based on physiological comfort thresholds, the maximum adjustment thresholds for color temperature are determined to be ±500K and brightness ±200 lux; according to historical data, the user behavior weight is 0.6 and the natural light weight is 0.4; by scaling according to the weights, the color temperature is divided into ranges of ±300K and brightness is ±80 lux; the final color temperature parameter extraction range is 2700K–3300K and brightness is 320 lux–480 lux; this range will be used for subsequent screening of effective parameter points to ensure that the solution not only meets the reading scene benchmark but also responds to user preferences and environmental changes.

[0071] This embodiment provides an AI-driven personalized lighting service method. By retrieving the standard scene parameter template corresponding to the lighting customization request and extracting the core parameter area to establish the benchmark lighting features, the maximum adjustment threshold of each feature is calculated by combining the lighting parameter threshold. The threshold is dynamically scaled according to the adjustment weight of the sub-data module to form a division interval. The parameter extraction interval is then expanded with the benchmark value as the center. This achieves the technical effect of anchoring typical lighting benchmarks through standard scene parameter templates, locking safety boundaries by combining physiological and equipment constraints, and dynamically allocating adjustment space according to the relative importance of user behavior and environmental factors, thereby generating a parameter adjustment range that combines standardization, safety, and individual adaptability.

[0072] In one embodiment, standard illumination features within a standard parameter region are extracted based on illumination parameter thresholds, and the maximum adjustment threshold of the standard illumination features is obtained, including: Parameter points whose parameter values ​​are within the illumination parameter threshold within the standard parameter area are identified as standard parameter points. A set of standard parameter points is generated based on adjacent standard parameter points, and standard illumination features are generated based on the set of standard parameter points. Extract the standard adjustment curve of the standard illumination feature, obtain the peak fluctuation amplitude of the standard adjustment curve as the maximum adjustment threshold corresponding to the standard illumination feature, and the waveform of the standard adjustment curve is a time-parameter change curve.

[0073] The lighting parameter threshold can be a preset physical and physiological safety boundary that limits the range of illumination parameter values, used to screen valid parameter points. In this embodiment, the lighting parameter threshold is set according to the International Commission on Illumination (CIE) human visual standards, equipment hardware limits, and industry specifications, such as a color temperature of 2700K–6500K and a brightness of 200–1500 lux. The lighting parameter threshold serves as the sole criterion for screening standard parameter points; it is a prerequisite constraint for generating the set of standard parameter points.

[0074] The standard parameter point set can be an ordered set of continuous parameter points within a standard parameter region that meet the illumination parameter threshold, representing a legal illumination state sequence under typical scenarios. In this embodiment, the standard parameter point set is obtained by traversing all parameter points within the standard parameter region, filtering out points located within the illumination parameter threshold boundary, and merging adjacent points to form a continuous time series. The standard parameter point set is a direct input to standard illumination characteristics; it drives the generation of standard adjustment curves; and it provides the original data source for peak fluctuation amplitude. For example, the standard parameter point set may include, but is not limited to, one or more of the following: luminance standard parameter point set, color temperature standard parameter point set, illuminance uniformity standard parameter point set, and dynamic frequency standard parameter point set.

[0075] Standard illumination features are illumination behavior patterns with a temporal structure abstracted from a set of standard parameter points, representing the typical evolution of parameters in a given scene. In this embodiment, standard illumination features are generated by aggregating the standard parameter point set, preserving its temporal order and change trend to form a semantically meaningful description of illumination behavior. Standard illumination features are generated from the standard parameter point set; they are the carrier of the standard adjustment curve; and their peak fluctuation amplitude constitutes the maximum adjustment threshold.

[0076] A standard adjustment curve can be a time-series function curve generated based on a set of standard parameter points, describing the change of illumination parameters over time, with time on the horizontal axis and parameter values ​​on the vertical axis. In this embodiment, the standard adjustment curve is generated by sorting the parameter points in the set of standard parameter points by timestamp and using interpolation or fitting algorithms to create a continuously changing curve, reflecting dynamic trends such as increased morning light and decreased nighttime light. The standard adjustment curve quantifies the natural evolution pattern of illumination parameters under typical scenarios, serving as the basis for calculating the maximum adjustment threshold; it also works in conjunction with the adjustment weights to determine the division intervals of each sub-data module. For example, the standard adjustment curve may include, but is not limited to, brightness time adjustment curves, color temperature time adjustment curves, illuminance uniformity time adjustment curves, and dynamic frequency time adjustment curves.

[0077] Peak fluctuation amplitude can be the difference between the maximum and minimum values ​​on the standard adjustment curve, representing the maximum dynamic range of the illumination characteristic in a typical time series. In this embodiment, the peak fluctuation amplitude is calculated by taking the global maximum difference of the amplitude difference between all local extreme points on the standard adjustment curve as the peak fluctuation amplitude of the standard illumination characteristic. The peak fluctuation amplitude serves as the numerical expression of the maximum adjustment threshold and, in conjunction with the adjustment weight, determines the division interval of each sub-data module. For example, the peak fluctuation amplitude may include, but is not limited to, peak fluctuation amplitude of luminance, peak fluctuation amplitude of color temperature, peak fluctuation amplitude of illuminance uniformity, and peak fluctuation amplitude of dynamic frequency.

[0078] Standard parameter points are identified by identifying parameter values ​​within the lighting parameter threshold within a standard parameter region. A set of standard parameter points is then generated based on adjacent standard parameter points. This can be achieved by checking each parameter value within the standard parameter region to see if it falls within the lighting parameter threshold range, and then aggregating the qualifying points into a continuous set in chronological order. Furthermore, this process of identifying parameter values ​​within the standard parameter region that fall within the lighting parameter threshold and generating a set of standard parameter points based on adjacent standard parameter points can be accomplished by using a sliding window method to threshold-filter parameter points within a continuous time window, retaining a continuous sequence of valid points, or by using a dynamic programming algorithm to identify the longest valid parameter subsequence and eliminate isolated outliers. This approach achieves the technical effect of purifying the original parameter space into a compliant behavior trajectory, eliminating noise and exceeding limits, and constructing a modelable dynamic benchmark.

[0079] Generating standard illumination features from a set of standard parameter points can be achieved by organizing the set of standard parameter points into a structured data sequence along the time dimension, and abstracting it into a semantically meaningful illumination evolution pattern. Furthermore, generating standard illumination features from the set of standard parameter points can be achieved by using spline interpolation to fit the parameter points to generate a continuous and smooth illumination change function, or by using a Hidden Markov Model (HMM) to model typical state transition paths and identify pattern segments. This transforms discrete sampling points into reusable dynamic behavior templates, providing a semantic carrier for adjustment curve extraction.

[0080] Extracting the standard adjustment curve of the standard illumination feature and obtaining the peak fluctuation amplitude of the standard adjustment curve as the maximum adjustment threshold corresponding to the standard illumination feature can be achieved by performing time-series analysis on the standard illumination feature, extracting the curve of its parameters changing over time, and calculating the global extreme value difference of the curve as the maximum adjustment threshold. Furthermore, extracting the standard adjustment curve of the standard illumination feature and obtaining the peak fluctuation amplitude of the standard adjustment curve as the maximum adjustment threshold corresponding to the standard illumination feature can be achieved by using Fourier transform to analyze the periodic fluctuations of the curve and extracting the dominant frequency amplitude as the peak fluctuation amplitude, or by using a local extremum detection algorithm to identify the rising and falling inflection points of the curve and calculating the maximum peak-to-valley difference. This allows the static threshold to be upgraded to a dynamic boundary based on historical behavior data, enabling the adjustment range to adapt to the actual evolution of illumination.

[0081] Taking morning wake-up lighting as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the system retrieves the standard scene parameter area for morning wake-up and extracts the parameter sequence in which the brightness gradually increases from 50 lux to 800 lux within 30 minutes; the lighting parameter threshold limits the brightness to 100–1000 lux, and selects continuous parameter points that meet the range to form a standard parameter point set; a standard brightness adjustment curve is generated by spline fitting, and its peak fluctuation range is found to be 750 lux (800-50); this value is used as the maximum adjustment threshold for range expansion when users make personalized adjustments in the future; when the user turns on the reading mode in the morning, the system allows the brightness to fluctuate by ±375 lux on the basis of the benchmark 500 lux, ensuring that the adjustment range always fits the gradual change mode of natural light and avoids sudden stimulation.

[0082] This embodiment provides an AI-driven personalized lighting service method. It filters standard parameter points by setting lighting parameter thresholds and constructs a set of standard parameter points, generating standard illumination characteristics with a temporal structure. Then, it extracts the standard adjustment curve and uses the peak fluctuation amplitude as the maximum adjustment threshold, achieving a paradigm shift from static thresholds to dynamic waveform boundaries. This mechanism enables the system to quantify the natural fluctuation range of typical adjustment modes, such as gradual brightening at dawn or attenuation at night, based on historical real-world behavioral data. This automatically constrains parameter changes during personalized adjustments to ensure they do not deviate from physiological comfort and environmental evolution patterns. This overcomes the limitations of traditional systems that rely on manually set static boundaries, achieving an intelligent upgrade with adaptive adjustment boundaries, reproducible behavioral patterns, and interpretable scheme generation, significantly improving the scientific rigor, stability, and user experience consistency of personalized lighting.

[0083] In one embodiment, the illumination adaptability of multiple sub-illumination parameter sets is verified by the illumination optimization unit at the AI ​​decision server to obtain a personalized lighting scheme, including: The coordinated adjustment point corresponding to each sub-lighting parameter set in the sub-lighting parameter sequence is obtained based on the illumination optimization unit.

[0084] The illumination optimization unit can be a multi-level lighting scheme verification and optimization module deployed in the AI ​​decision server. It is responsible for fusing sub-lighting parameter sets, calculating matching degrees, and performing secondary verification. This allows for objective evaluation and closed-loop correction of personalized lighting schemes, ensuring that the output conforms to standard behavioral patterns and physiological energy efficiency constraints. In this embodiment, the illumination optimization unit can receive sub-lighting parameter sequences, execute collaborative adjustment point extraction, curve overlay, matching degree calculation, and secondary adaptation verification processes, and output the final scheme or adjustment markers. The illumination optimization unit can be a collaborative hub driving the extraction and fusion of collaborative adjustment points, calling standard adjustment curves for matching comparison, triggering secondary adaptation verification, and outputting parameter matching degree results. The illumination optimization unit may include, but is not limited to, a parameter fusion module, a curve alignment module, a matching degree calculation module, and a secondary verification module.

[0085] Based on the illumination optimization unit, the coordinated adjustment points corresponding to each sub-illumination parameter set in the sub-illumination parameter sequence can be obtained by analyzing the temporal distribution, slope of change, and peak position of each parameter set in the sub-illumination parameter sequence to identify key parameter points with logical correlation or synchronous changes. Furthermore, the coordinated adjustment points obtained based on the illumination optimization unit can be achieved by using cross-correlation analysis to detect similarity peak points of brightness and color temperature parameter sets in the time domain, or by using dynamic time warping algorithms to align the change trajectories of different parameter sets and extracting the alignment points as coordinated adjustment points. This enables structured coordinated modeling of multi-dimensional parameters, avoiding abrupt changes or conflicts in illumination caused by independent adjustments.

[0086] The sub-lighting parameter sets are fused according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set to obtain a primary lighting scheme.

[0087] In this context, the coordinated adjustment point can be a key parameter point where sub-lighting parameters are dependent on other parameter sets in terms of time or function. It guides multi-parameter coordinated fusion, identifies temporal coupling relationships between parameters, avoids conflicts or inconsistencies caused by independent adjustments, and improves the overall consistency of the solution. In this embodiment, the coordinated adjustment point can analyze the start and end times, slope of change, and peak positions of each parameter set in the sub-lighting parameter sequence to identify parameter points that change synchronously or are logically related on the time axis. The coordinated adjustment point can serve as the basis for fusion processing, influence the shape of the actual adjustment curve, and determine the distribution structure of matching parameter points. Coordinated adjustment points may include, but are not limited to, brightness-color temperature synchronization points, dynamic rhythm start points, natural light compensation inflection points, and user preference trigger points.

[0088] By fusing the sub-lighting parameter sets according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set, a primary lighting scheme is obtained. This can be achieved by merging the parameter values ​​of each sub-parameter set according to priority, using the coordinated adjustment points as anchor points, to generate a continuous single actual adjustment curve. Furthermore, by fusing the sub-lighting parameter sets according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set, a primary lighting scheme can be obtained. This can be achieved by using weighted interpolation to fuse parameters such as brightness and color temperature at the coordinated adjustment points, generating a smooth transition curve, or by using a priority queue mechanism to ensure that high-priority parameter sets cover the conflict points of low-priority parameter sets, ensuring that the dominant factor dominates the output. This allows the dispersed parameter sets to be integrated into a unified temporal lighting scheme, forming a verifiable and complete adjustment trajectory.

[0089] Extract the actual adjustment curve of the primary lighting scheme, locate the feature points of the standard adjustment curve based on the feature points of the actual adjustment curve, and superimpose the standard adjustment curve on top of the actual adjustment curve.

[0090] The standard adjustment curve can be a time-series function curve describing the evolution of ideal illumination parameters over time, generated based on historical typical scenarios. It serves as a verification benchmark, providing a comparable reference mode for optimal physiological comfort and energy efficiency, and supporting the quantitative calculation of the matching degree. In this embodiment, the standard adjustment curve can be generated by interpolation fitting of a set of standard parameter points, and its feature points include key morphological nodes such as rising edges, peak values, and attenuation segments. The standard adjustment curve can be superimposed on the actual adjustment curve for matching analysis; its feature points are used to locate overlapping areas and serve as a reference template for secondary adaptation verification. The standard adjustment curve may include, but is not limited to, luminance standard adjustment curves, color temperature standard adjustment curves, illuminance uniformity standard adjustment curves, and dynamic frequency standard adjustment curves.

[0091] The actual adjustment curve of the primary lighting scheme is extracted. Based on the feature points of the actual adjustment curve, the feature points of the standard adjustment curve are located. The standard adjustment curve is then superimposed on the actual adjustment curve. Alternatively, key morphological points of the actual adjustment curve can be extracted and spatiotemporally aligned with corresponding feature points of the standard adjustment curve before being superimposed. Furthermore, extracting the actual adjustment curve of the primary lighting scheme and locating the feature points of the standard adjustment curve based on its feature points, then superimposing the standard adjustment curve on top of the actual adjustment curve, can be achieved by using a feature point matching algorithm to locate corresponding nodes of the two curves on the time axis, or by using the least squares method to fit the offset of the two curves to achieve optimal overlap and alignment. This establishes a comparable visual and mathematical reference framework, providing a spatial alignment basis for matching degree calculation.

[0092] The parameter points where the actual adjustment curve overlaps with the standard adjustment curve are obtained as matching parameter points. The number of matching parameter points and the total number of parameter points corresponding to the actual adjustment curve are counted. The parameter matching degree is obtained based on the ratio of the number of matching points to the total number.

[0093] The parameter matching degree can be defined as the percentage of overlapping parameter points between the actual adjustment curve and the standard adjustment curve after feature point alignment. It quantifies the degree of conformity between the scheme and the ideal model, and can be used to transform subjective comfort into a calculable objective indicator, enabling measurable evaluation of scheme quality. In this embodiment, the parameter matching degree can be calculated by counting the number of overlapping parameter points of the two curves on the time axis, dividing it by the total number of sampling points of the actual adjustment curve, and obtaining a matching ratio between 0 and 1. The parameter matching degree can be calculated from the superposition result of the actual adjustment curve and the standard adjustment curve, determining whether the scheme needs to undergo secondary verification or requires adjustment.

[0094] The parameter points where the actual adjustment curve overlaps with the standard adjustment curve are identified as matching parameter points. The ratio of the number of matching points to the total number of matching points yields the parameter matching degree. This can be achieved by checking whether the parameter values ​​are consistent within a preset tolerance range point by point on the two aligned curves, and then calculating the percentage of matching points. Furthermore, the parameter matching degree can be determined by setting a ±5% parameter tolerance range to identify overlapping points as matching parameter points, or by using a fuzzy membership function to assign partial matching weights based on the degree of deviation to calculate a weighted matching degree. This transforms subjective comfort into a quantifiable and comparable numerical indicator, enabling an objective assessment of the scheme's quality.

[0095] Primary lighting schemes with parameter matching degree less than the baseline matching degree are identified as schemes to be optimized, and the sub-parameter sets that need to be adjusted in the schemes to be optimized are marked; primary lighting schemes with parameter matching degree greater than or equal to the baseline matching degree are identified as secondary optimization schemes; secondary adaptation verification is performed on the secondary optimization schemes and standard illumination characteristics to obtain the final lighting scheme or the scheme to be adjusted, and a personalized lighting scheme is obtained based on the final lighting scheme or the scheme to be adjusted.

[0096] Secondary adaptation verification can be a process of re-verifying the consistency of a highly matched primary lighting scheme with standard illumination characteristics in multiple dimensions. This prevents the neglect of global characteristics due to localized matching, ensuring that the scheme meets standards in terms of physiological comfort, energy efficiency balance, and behavioral patterns. In this embodiment, secondary adaptation verification can perform feature point matching, fluctuation amplitude comparison, and energy distribution analysis on the actual adjustment curve of the primary lighting scheme to determine whether it is completely consistent with standard illumination characteristics. Secondary adaptation verification can use parameter matching degree as a trigger condition, rely on the standard adjustment curve and standard illumination characteristics as verification basis, and output the final scheme or a mark to be adjusted. Secondary adaptation verification may include, but is not limited to, feature point consistency verification, fluctuation amplitude compliance verification, energy distribution matching verification, and time offset tolerance verification.

[0097] The final lighting scheme or the scheme to be adjusted is obtained by performing secondary adaptation verification on the secondary optimization scheme and standard illumination characteristics. This can be a primary scheme that meets the parameter matching standards, and its fluctuation amplitude, energy distribution and consistency with standard illumination characteristics are further examined to determine whether it is the final scheme. Furthermore, the final lighting scheme or the scheme to be adjusted can be obtained by calculating the deviation rate between the integral energy of the actual adjustment curve and the standard curve to determine energy efficiency consistency, or by detecting whether the frequency domain characteristics of the curve match the dominant frequency component of the standard mode. This can prevent local matching from masking global mismatch and ensure that the final scheme meets the standards in terms of behavioral patterns, physiological comfort and energy efficiency.

[0098] Taking midday meeting lighting as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the system generates a primary lighting scheme: brightness gradually increases from 400 lux to 900 lux, color temperature increases from 3000K to 5000K, and the dynamic rhythm is in a stable mode; the illumination optimization unit extracts the collaborative adjustment point, that is, the increase in color temperature and the increase in brightness are synchronized 5 minutes before the start of the meeting; after fusion, an actual adjustment curve is generated; it is aligned and superimposed with the standard adjustment curve for office meetings, and it is found that the matching degree of the brightness increase segment is 92%, but the slope of the color temperature increase deviates from the standard curve; the parameter matching degree is 88%, which is higher than the benchmark of 85%, triggering a secondary adaptation verification; the system further verifies its energy distribution and main frequency components, confirming that it meets the standard illumination characteristics; the final output is a qualified scheme, which is pushed to the user interface; if the matching degree is lower than 85%, the color temperature parameter set is marked as needing adjustment, and feedback is given to the environmental perception unit to recalculate the range.

[0099] This embodiment provides an AI-driven personalized lighting service method. It extracts the coordinated adjustment points of sub-lighting parameter sets through a lighting optimization unit and achieves structured fusion, generating actual adjustment curves with consistent temporal patterns. By aligning and superimposing feature points with standard adjustment curves, a quantifiable evaluation index of parameter matching degree is constructed. Secondary adaptation verification is performed on high-matching solutions to ensure complete consistency with standard characteristics in fluctuation patterns and energy distribution. This elevates the evaluation of lighting solutions from the parameter combination level to the curve morphology level, achieving a paradigm shift from whether parameters meet standards to whether behaviors are similar. This enables the system to possess interpretable, measurable, and iterative scientific decision-making capabilities, completely solving the inefficiencies of blind generation and manual debugging in traditional systems. It achieves a systematic balance between comfort, reliability, and automation in personalized lighting.

[0100] In one embodiment, obtaining the coordinated adjustment points corresponding to each sub-lighting parameter set in the sub-lighting parameter sequence based on the illumination optimization unit includes: performing time-series processing on each sub-lighting parameter set and extracting multiple adjustment time periods corresponding to each sub-lighting parameter set.

[0101] The adjustment period can be a continuous time interval in which parameter values ​​in the sub-illumination parameter set remain in a high or low state. It can be used to transform discrete parameter points into semantically meaningful time units, providing structured input for subsequent peak and trough identification. In this embodiment, the adjustment period can be implemented based on the continuity characteristics of parameter values ​​in the time series, using a sliding window and threshold comparison method or a rate of change detection algorithm. When a parameter value continuously exceeds a preset sampling point threshold or the rate of change is lower than a set value and maintained for a sufficient duration, it is identified as a stable adjustment period. Furthermore, the adjustment period can include, but is not limited to, one or more of the following: brightness adjustment period, color temperature adjustment period, illuminance uniformity adjustment period, and dynamic frequency adjustment period.

[0102] The time-series processing of each sub-lighting parameter set extracts multiple adjustment periods corresponding to each sub-lighting parameter set. This can be achieved by sorting the sub-lighting parameter sets by timestamps and identifying continuous intervals where parameter values ​​remain consistently high or low. Further, the time-series processing of each sub-lighting parameter set can be implemented using a sliding window and threshold comparison method. When a parameter value is continuously higher than the mean plus standard deviation for more than N sampling points, it is marked as a peak period. Alternatively, a rate of change detection algorithm can be used; when the parameter rate of change is lower than a threshold and remains below it for more than T seconds, it is considered a stable adjustment period. This transforms discrete parameter points into semantically meaningful time-series units, providing structured input for subsequent peak / valley identification.

[0103] Obtain the maximum and minimum parameter values ​​corresponding to multiple adjustment periods in the sub-lighting parameter set; determine the adjustment period corresponding to the maximum parameter value as the peak adjustment period, and determine the adjustment period corresponding to the minimum parameter value as the off-peak adjustment period.

[0104] The peak adjustment period can be a continuous time interval in which the parameter values ​​in the sub-illumination parameter set reach their maximum range, representing the stage where the parameter dominates enhancement. It can be used to identify the high-influence state of the parameter within a specific time period and to locate the peak region of the adjustment behavior. In this embodiment, the peak adjustment period can be generated based on the scanning results of the time series of the sub-illumination parameter set, identifying continuous time periods where parameter values ​​are continuously higher than a preset threshold and reach a local maximum. Furthermore, the peak adjustment period can include, but is not limited to, one or more of the following: peak brightness period, peak color temperature period, peak illuminance uniformity period, and peak dynamic frequency period.

[0105] A low-influence adjustment period can be a continuous time interval in which the parameter value in the sub-illumination parameter set is at its minimum range, representing a stage where the parameter's dominance weakens. It can be used to identify the low-influence state of a parameter within a specific time period and to locate the low-influence area of ​​the adjustment behavior. In this embodiment, the low-influence adjustment period can be generated based on the scanning results of the time series of the sub-illumination parameter set, identifying continuous time periods where the parameter value is continuously below a preset threshold and reaches a local minimum. Furthermore, the low-influence adjustment period can include, but is not limited to, one or more of the following: brightness low-influence periods, color temperature low-influence periods, illuminance uniformity low-influence periods, and dynamic frequency low-influence periods.

[0106] Determining the adjustment period corresponding to the maximum parameter value as the peak adjustment period and the adjustment period corresponding to the minimum parameter value as the trough adjustment period can be achieved by identifying the intervals with the maximum and minimum parameter values ​​from multiple extracted adjustment periods. Further, this determination process can be implemented by calculating the average parameter value for each period and selecting the highest and lowest values, or by using a peak detection algorithm to identify local maximum / minimum value intervals and verifying whether their duration meets the minimum period threshold. This allows for semantic classification of parameter states, distinguishing between dominant enhancement and dominant weakening adjustment phases, and supporting collaborative logic judgment.

[0107] If the peak adjustment period and the trough adjustment period are adjacent, the boundary time between the adjacent periods is determined as the coordinated adjustment point.

[0108] The coordinated adjustment point can be a timing marker used to identify key positions of coordinated parameter changes during the temporal adjustment of a sub-lighting parameter set. It determines the trigger or midpoint of parameter transitions, guiding multiple parameter sets to achieve structured coordination on the time axis and avoiding abrupt changes or discontinuities caused by independent adjustments. In this embodiment, the coordinated adjustment point can be generated based on the spatiotemporal relationship between peak and trough adjustment periods, determined by the boundary moment or calculated from the midpoint of the transition interval. Furthermore, the coordinated adjustment point can include, but is not limited to, one or more of the following: instantaneous switching coordinated point, gradual transition coordinated point, multi-parameter synchronous trigger point, and dominant parameter guiding point.

[0109] If peak and off-peak adjustment periods are adjacent, the boundary time between these periods is determined as the coordinated adjustment point. This can be achieved when the end of the peak adjustment period is equal to or the start of the off-peak adjustment period has zero time difference. Furthermore, this determination process can be configured with a time tolerance ε. If |end of peak - start of off-peak| ≤ ε, the periods are considered adjacent, and the boundary point is the coordinated adjustment point. Alternatively, an instantaneous parameter switching command can be inserted at the boundary point to trigger a synchronous response from the lighting equipment. This allows for precise marking of the switching timing in scenarios where parameter states switch directly, ensuring synchronous response of multiple parameters and avoiding timing misalignment.

[0110] If the peak adjustment period and the off-peak adjustment period are not adjacent, the end time of the peak adjustment period is taken as the first adjustment point, and the start time of the off-peak adjustment period is taken as the second adjustment point; the first adjustment point and the second adjustment point are connected to obtain the adjustment transition interval, and the midpoint of the adjustment transition interval is determined as the corresponding coordinated adjustment point of the corresponding sub-lighting parameter set.

[0111] The adjustment transition interval can be an intermediate transition time interval constructed to achieve smooth parameter evolution when the peak adjustment period and the trough adjustment period are not adjacent. It can be used to introduce a gradual buffer in the discontinuous region of parameter state switching to avoid visual discomfort caused by abrupt changes. In this embodiment, the adjustment transition interval can be formed by connecting the end time of the peak adjustment period and the beginning time of the trough adjustment period, and its length is set by system default or user preference. Furthermore, the adjustment transition interval can include, but is not limited to, one or more of the following: brightness transition interval, color temperature transition interval, illuminance uniformity transition interval, and dynamic rhythm transition interval.

[0112] If the peak adjustment period and the trough adjustment period are not adjacent, the end time of the peak adjustment period is taken as the first adjustment point, and the start time of the trough adjustment period is taken as the second adjustment point. Connecting the first and second adjustment points yields the adjustment transition interval. The midpoint of the adjustment transition interval is determined as the coordinated adjustment point corresponding to the respective sub-lighting parameter set. This can be achieved by calculating the time difference between the end of the peak and the start of the trough, and taking the midpoint as the coordinated adjustment point of the transition interval. Furthermore, this process can employ linear interpolation to evenly distribute parameter value changes over time between the first and second adjustment points, or dynamically adjust the transition duration based on user comfort preferences, so that the midpoint position is determined according to a weighted time ratio. This allows for the construction of a smooth transition channel in discontinuous adjustment scenarios, ensuring that parameter changes conform to the physiological adaptation rhythm and eliminating abruptness.

[0113] Taking the transition from dusk to nighttime reading as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: The system detects that the peak adjustment period for the daytime office sub-parameter set is 17:00–18:30 (high brightness, cool color temperature), and there is no immediate trough after it ends; while the trough adjustment period for the nighttime reading sub-parameter set is 19:30–20:00 (low brightness, warm color temperature). Since the two are not adjacent, the system extracts the end time of the peak, 18:30, as the first adjustment point and the start time of the trough, 19:30, as the second adjustment point, constructing a 1-hour adjustment transition interval; taking the midpoint, 19:00, as the collaborative adjustment point, at this time, the transition command of gradually warming the color temperature and gradually decreasing the brightness is initiated; the lighting equipment smoothly transitions from 5000K, 3000K, 900lux, and 400lux according to linear interpolation, and the user does not have obvious perceptible abrupt change before and after 19:00, realizing a seamless switch from daytime mode to nighttime mode.

[0114] This embodiment provides an AI-driven personalized lighting service method. By processing the sub-lighting parameter set temporally to extract adjustment periods, peak and trough adjustment periods are identified based on parameter extreme values. The collaborative adjustment point is determined as the boundary moment or the midpoint of the transition interval based on their spatiotemporal relationship, and an adjustment transition interval is constructed to achieve smooth evolution. This can transform the sub-lighting parameter set from discrete point control to continuous modeling of time period-transition-collaboration, enabling complex scenarios such as the transition from daytime to nighttime, or the switch from meeting end to rest mode, to achieve a smooth evolution of physiological adaptation while maintaining functional intent. This completely eliminates problems such as brightness jumps, abrupt color temperature changes, and dynamic discontinuities caused by independent parameter adjustment in traditional systems, significantly improving the experience continuity, visual comfort, and behavioral naturalness of personalized lighting solutions.

[0115] In one embodiment, the sub-lighting parameter sets are fused according to the sub-lighting parameter sequence and the coordinated adjustment point of each sub-lighting parameter set to obtain a primary lighting scheme, including: The first set of sub-lighting parameters in the sub-lighting parameter sequence is obtained as the first fusion parameter set, and the set of sub-lighting parameters adjacent to the first fusion parameter set is obtained as the second fusion parameter set; The point at which the first fusion parameter set and the second fusion parameter set are coordinated is determined as the fusion start point, and the point at which the second fusion parameter set and the first fusion parameter set are coordinated is determined as the fusion end point; Using the fusion starting point as a reference point, the parameters of the fusion ending point are controlled to transition towards the fusion starting point until the parameter change rates of the fusion starting point and the fusion ending point are consistent, at which point the transition of the fusion ending point stops. The fusion processing of the first fusion parameter set and the second fusion parameter set is completed based on the transition processing of the fusion start point and the fusion end point; The second fusion parameter set is updated to the first fusion parameter set, and the adjacent sub-lighting parameter sets of the updated first fusion parameter set are obtained as the next second fusion parameter set; Repeat the above steps of transition processing for the fusion start point and fusion end point, and perform fusion processing on each of the sub-lighting parameter sets to obtain a primary lighting scheme.

[0116] The coordination adjustment point can be a key intersection point where sub-illumination parameter sets have a logical relationship in terms of time or parameter change trends, used to identify the start and end positions of parameter transitions. In this embodiment, the coordination adjustment point can serve as an alignment benchmark for parameter fusion, ensuring that adjacent parameter sets have a physically consistent and connectable change rhythm. For example, the coordination adjustment point can identify parameter points that overlap in time or have logical dependencies by analyzing the start and end points, slope changes, and peak positions of each sub-parameter set. The coordination adjustment point can define the positions of the fusion start point and fusion end point; serve as a reference benchmark for comparing parameter change rates; and drive the triggering timing of the fusion process. The coordination adjustment point can include, but is not limited to, one or more of the following: brightness-color temperature synchronization point, natural light compensation trigger point, user preference switching point, and time period transition response point.

[0117] The fusion starting point can be the last parameter point of a higher priority or already stable parameter set within two adjacent sub-illumination parameter sets, serving as a transition reference. In this embodiment, the fusion starting point can provide a stable reference for parameter transition, ensuring that the fusion process does not disrupt existing reasonable adjustment states. For example, the fusion starting point can be the last valid parameter point of the first parameter set currently being processed in the sub-illumination parameter sequence. The fusion starting point can serve as a reference coordinate for parameter transition; compare the rate of change with the fusion endpoint; and determine the transition direction and termination conditions. The fusion starting point can include, but is not limited to, one or more of the following: luminance fusion starting point, color temperature fusion starting point, dynamic rhythm fusion starting point, and illuminance uniformity fusion starting point.

[0118] The fusion endpoint can be a starting parameter point in two adjacent sub-illumination parameter sets, which is to be aligned with the fusion starting point and has a lower priority or is about to be covered. In this embodiment, the fusion endpoint can play an adjustment role in parameter transition, eliminating abrupt boundaries between parameter sets through gradual alignment. For example, the fusion endpoint can be the first valid parameter point of the current secondary parameter set to be fused in the sub-illumination parameter sequence. The fusion endpoint can transition parameters towards the fusion starting point; together with the fusion starting point, it constitutes a transition interval; its rate of change determines whether the transition terminates. The fusion endpoint can be one or more of the following, including but not limited to luminance fusion endpoint, color temperature fusion endpoint, dynamic rhythm fusion endpoint, and illuminance uniformity fusion endpoint.

[0119] The parameter change rate can be the magnitude of parameter value change within a unit time or unit sampling interval, characterizing the dynamic evolution speed of the parameter over time. In this embodiment, the parameter change rate can serve as a convergence criterion for parameter transition, ensuring that the parameter change rhythm at both ends is consistent during the fusion process and avoiding visual abrupt changes. For example, the parameter change rate can be calculated by the difference between adjacent sampling points, such as the ratio of brightness difference to time interval, or the ratio of color temperature difference to sampling period. The parameter change rate can be the basis for the termination condition of the transition at the fusion endpoint; together with the fusion start point and fusion endpoint, it constitutes a transition control closed loop. The parameter change rate can include, but is not limited to, one or more of the following: brightness change rate, color temperature change rate, dynamic frequency change rate, and illuminance uniformity change rate.

[0120] Determining the fusion start point as the coordinated adjustment point between the first and second fusion parameter sets, and the fusion end point as the coordinated adjustment point between the second and first fusion parameter sets, can be achieved by selecting two adjacent parameter sets currently being processed within the sub-illumination parameter sequence, extracting their endpoint values ​​at the coordinated adjustment point, with the former as the fusion start point and the latter as the fusion end point. In this embodiment, this operation can establish a bidirectional anchoring structure for parameter fusion, clearly defining the reference and adjustment ends of the transition, providing a spatial positioning basis for a smooth transition. Furthermore, this operation can be achieved by using a sliding window scanning sequence, truncating the end point of the previous parameter set as the start point and the beginning point of the next parameter set as the end point at the coordinated adjustment point, thereby enabling the fusion process to have temporal alignment accuracy.

[0121] Using the fusion starting point as a reference, the parameter transition from the fusion ending point to the fusion starting point is controlled until the parameter change rates of the fusion starting point and the fusion ending point are consistent, at which point the transition to the fusion ending point stops. This can be achieved by continuously interpolating and adjusting the parameter values ​​of the fusion ending point, gradually bringing its change rate closer to that of the fusion starting point, until the difference between the two is lower than a tolerance threshold. In this embodiment, this operation can achieve physical consistency in parameter transition, eliminate abrupt switching, and simulate the natural adaptation process of the human eye to changes in light. Furthermore, this operation can be achieved by using linear interpolation combined with a gradient descent algorithm to iteratively correct the parameter values ​​of the fusion ending point and minimize the difference in the rate of change, thereby making the transition process present a continuous and gradual characteristic at the perception level.

[0122] The second fusion parameter set is updated to the first fusion parameter set, and the adjacent sub-lighting parameter sets of the updated first fusion parameter set are then obtained as the next second fusion parameter set. This can be achieved by completing the fusion of the current two parameter sets, using the fused new parameter set as the new first fusion parameter set, and then taking the next unprocessed parameter set in the sequence as the second fusion parameter set. In this embodiment, this operation can construct a chained fusion mechanism, allowing multiple sub-parameter sets to be sequentially and without omission integrated into a unified output. Furthermore, this operation can be implemented by using a queue structure to manage the sub-parameter set sequence. After each fusion, the first element is dequeued, the fusion result is enqueued, and the first and second elements of the queue are processed, thus enabling the fusion process to have scalability and closed-loop characteristics.

[0123] Repeating the transition processing steps for the fusion start and end points described above, the fusion processing of each sub-lighting parameter set is performed to obtain a primary lighting scheme. This can be achieved by iteratively executing the parameter alignment process between the fusion start and end points according to the sequence of sub-lighting parameters until all sub-parameter sets are fused. In this embodiment, this operation can achieve continuous integration of multi-dimensional and multi-stage parameter sets, outputting a single lighting adjustment curve without breaks or jitter. Furthermore, this operation can use a loop iterator to update the fusion window and generate intermediate transition curves in each iteration, ultimately stitching them together to form a complete scheme, thereby enabling the lighting scheme to have global consistency and temporal coherence.

[0124] Taking the transition from dusk to nighttime reading as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the system arranges a sequence of sub-parameter sets according to priority: the first is the natural light attenuation set (brightness from 600 lux to 300 lux), the second is the user-preferred warm light set (color temperature from 4000K to 3000K), and the third is the nighttime reading stability set (brightness maintained at 280 lux); the fusion starting point is the end point of the natural light set (300 lux), and the fusion ending point is the beginning point of the warm light set (4000K); the system uses the change rate of the natural light set (-15 lux / min) as a benchmark to gradually... The slope of color temperature change in the warm light set is reduced until its rate of change is equivalent to the rate of change of natural light brightness at the physiological perception level (i.e., the rate of color temperature decrease and the rate of brightness decrease form a coordinated perceptual rhythm). After fusion, the brightness-color temperature transition set of the new combination is updated to the first parameter set and continues to be fused with the stable set for nighttime reading. Since the latter is a stable mode with a rate of change of 0, the system slowly decays the brightness at the transition endpoint from 300 lux to 280 lux and matches its rate of change of 0 to complete the final fusion. A continuous, non-abrupt brightness and color temperature coordinated change curve is output to simulate the gradual transition of natural light.

[0125] This embodiment provides an AI-driven personalized lighting service method. By acquiring adjacent parameter sets in the sub-lighting parameter sequence and extracting their collaborative adjustment points to determine the fusion start and fusion end points, the method controls the parameter transition of the fusion end point towards the fusion start point until the rate of change is consistent. By updating the fused parameter set to the next processing unit and iteratively executing the above process, the method can achieve the goal of locating the fusion start and fusion end points through collaborative adjustment points, using the parameter change rate as the convergence criterion, driving the fusion end point to gradually align with the fusion start point, and realizing a dynamic transition between sub-parameter sets based on physiological perception consistency. Through a chain-iteration mechanism, discrete multi-source parameter sets are integrated into a continuous, stepless primary lighting scheme. This process simulates the natural adaptation mechanism of the human eye to changes in the light environment, avoiding visual discomfort and experience breaks caused by parameter splicing in traditional systems. It upgrades personalized lighting from multi-instruction superposition to physiologically friendly continuous control, significantly improving the naturalness, comfort, and temporal continuity of the lighting scheme.

[0126] In some embodiments, a second adaptation verification is performed on the secondary optimization scheme and standard illumination features to obtain the final lighting scheme or the scheme to be adjusted, including: extracting the real-time user feedback parameters in the secondary optimization scheme and the ideal feedback parameters in the standard illumination features.

[0127] Real-time feedback parameters can be operation records or interaction data actively generated by the user during the use of a personalized lighting solution, reflecting their immediate preferences and comfort perception. It is understood that real-time feedback parameters can be collected through the user interface from manual adjustment behaviors, such as changes in the brightness slider position, color temperature selector switching records, preview confirmation / cancellation operations, and other time-series data. In this embodiment, real-time feedback parameters are compared with ideal feedback parameters as input for feedback similarity verification; this determines the level of the feedback matching value; and it affects the content of parameter correction suggestions. For example, real-time feedback parameters may include, but are not limited to, one or more of the following: brightness fine-tuning amount, color temperature offset value, number of dynamic mode switching times, and solution rejection trigger points.

[0128] Ideal feedback parameters can be feedback pattern templates representing the optimal user experience in typical scenarios, constructed based on user behavior clustering or physiological experiments. It can be understood that ideal feedback parameters are derived by collecting data on the fine-tuning behaviors of a large number of users in standard scenarios over a long period, extracting high-frequency response patterns, such as lowering the color temperature by an average of 600K while maintaining stable brightness during nighttime reading. In this embodiment, ideal feedback parameters serve as a benchmark for feedback similarity verification; together with real-time feedback parameters, they constitute the input for matching calculations; and they support the generation of parameter correction suggestions. For example, ideal feedback parameters may include, but are not limited to, one or more of the following: ideal brightness response in office scenarios, ideal color temperature response in sleep scenarios, ideal dynamic rhythm response in meeting scenarios, and ideal switching delay threshold in reading scenarios.

[0129] Extracting real-time user feedback parameters and ideal feedback parameters from standard lighting features in the secondary optimization scheme can be achieved by reading user fine-tuning behaviors during the preview or usage phase from user interaction logs, and simultaneously loading ideal feedback templates for the corresponding scene from the standard lighting feature library. Furthermore, extracting these parameters can be done by parsing the UI event flow of the user interface, extracting slider displacement, button click timestamps, and parameter value change sequences, or by retrieving the ideal feedback prototype implementation that best matches the current scene from a historical clustering model. This allows for the establishment of comparable data pairs between user behavior and ideal patterns, providing structured input for subsequent similarity calculations.

[0130] Feedback similarity verification is performed on the real-time feedback parameters and the ideal feedback parameters to obtain the feedback matching value. The secondary adaptation verification includes feedback similarity verification.

[0131] Feedback similarity verification can be a calculation process that semantically aligns and numerically compares real-time user feedback parameters with ideal feedback parameters in standard lighting features, used to evaluate the consistency between the proposed solution and the user's actual experience. It can be understood that feedback similarity verification aligns user operation logs with ideal feedback templates through feature mapping, and uses cosine similarity or dynamic time warping (DTW) algorithms to calculate the degree of matching. In this embodiment, feedback similarity verification is the basis for calculating feedback matching values; it drives the classification of adapted solutions and solutions to be adjusted; and it triggers the generation of parameter correction suggestions. For example, feedback similarity verification may include, but is not limited to, brightness preference similarity, color temperature preference similarity, dynamic rhythm acceptability, and switching response delay matching degree.

[0132] To obtain a feedback matching value, feedback similarity verification is performed on real-time feedback parameters and ideal feedback parameters. This can be achieved by mapping the real-time and ideal feedback parameters to the same feature space and calculating their vector similarity as the matching value. Furthermore, feedback similarity verification can be performed to obtain a matching value by using cosine similarity to calculate the angle between the trajectory changes of the two sets of parameters, outputting a matching score in the range of 0–1, or by using the Dynamic Time Warping (DTW) algorithm to align feedback sequences with inconsistent time axes and calculating the normalized similarity of the cumulative path distance. This allows for the objective quantification of subjective experience, making user preferences an assessable and comparable decision-making dimension.

[0133] The secondary optimization scheme with a feedback matching value greater than the baseline feedback value is the adaptation scheme, and the adaptation scheme is determined as the final lighting scheme.

[0134] The feedback matching value can be a numerical indicator representing the similarity between real-time feedback parameters and ideal feedback parameters, calculated through feedback similarity verification. It can be understood that the feedback matching value calculates the similarity after normalizing the feature vectors of the real-time feedback parameters and the ideal feedback parameters, outputting a matching score between 0 and 1. In this embodiment, the feedback matching value is directly output by the feedback similarity verification; it determines whether the scheme is adapted or needs adjustment; and it triggers the generation path of the final lighting scheme. For example, the feedback matching value may include, but is not limited to, brightness feedback matching value, color temperature feedback matching value, dynamic response feedback matching value, and interaction smoothness matching value.

[0135] The secondary optimization scheme with a matching feedback value greater than the baseline feedback value is selected as the adapted scheme, and the adapted scheme is determined as the final lighting scheme. Alternatively, if the matching feedback value exceeds a preset threshold, the scheme is deemed to meet the user experience standard and is directly confirmed as the final lighting scheme. Furthermore, this process of selecting the secondary optimization scheme with a matching feedback value greater than the baseline feedback value and determining it as the final lighting scheme can be achieved by setting the baseline feedback value to 0.85, automatically pushing the solution to the user and closing the optimization process when the matching value is ≥0.85, or by using an adaptive threshold mechanism to dynamically adjust the baseline value based on the stability of historical user feedback. This allows for rapid closed-loop optimization of highly matched schemes, reduces redundant optimization, and improves system response efficiency and user satisfaction.

[0136] The secondary optimization scheme with a feedback matching value less than the baseline feedback value is identified as the scheme to be adjusted. Parameter correction suggestions are generated based on the difference parameters in the scheme to be adjusted. The secondary optimization scheme is then updated in combination with the parameter correction suggestions to obtain the final lighting scheme.

[0137] The parameter correction suggestions can be automatically generated executable parameter adjustment instructions based on the differences between real-time feedback parameters and ideal feedback parameters in the scheme to be adjusted. It can be understood that the parameter correction suggestions analyze the direction and magnitude of the difference parameters, combine them with the adjustment rules of standard illumination characteristics, and generate structured instructions such as reducing the color temperature by 400K or increasing the brightness by 15%. In this embodiment, the parameter correction suggestions are generated from the difference parameters of the scheme to be adjusted; they serve as input for updating the secondary optimization scheme; and are ultimately transformed into a new round of parameter extraction ranges. For example, parameter correction suggestions may include, but are not limited to, color temperature correction suggestions, brightness correction suggestions, dynamic mode correction suggestions, and transition time correction suggestions.

[0138] The secondary optimization scheme with a feedback matching value less than the baseline feedback value is identified as the scheme to be adjusted. Based on the difference parameters in the scheme to be adjusted, parameter correction suggestions are generated. The secondary optimization scheme is then updated with these suggestions to obtain the final lighting scheme. This can be achieved by identifying the parameter dimension with the largest difference when the feedback matching value is below a threshold, generating a structured correction instruction, and writing it back to the secondary optimization scheme to regenerate the parameter set. Furthermore, the process of identifying the secondary optimization scheme with a feedback matching value less than the baseline feedback value as the scheme to be adjusted, generating parameter correction suggestions based on the difference parameters in the scheme to be adjusted, and updating the secondary optimization scheme with these suggestions to obtain the final lighting scheme can be implemented by generating a 500K color temperature reduction correction suggestion if the user frequently lowers the color temperature but the system outputs cool light, triggering the environmental perception unit to recalculate the parameter extraction range; or by automatically extending the transition time and reducing the brightness change slope if the user cancels the scheme multiple times, generating a dynamic slowdown correction instruction. This allows for the construction of a self-learning closed loop of user feedback, difference recognition, instruction generation, and scheme iteration, enabling the system to continuously evolve.

[0139] Taking nighttime reading mode as an example, the AI-driven personalized lighting service method in this embodiment can be as follows: the system generates a primary lighting scheme and outputs it after secondary optimization: brightness 500 lux, color temperature 3200K, no dynamic changes; the user manually lowers the color temperature to 2800K while keeping the brightness unchanged during preview; the feedback similarity verification module extracts the real-time feedback parameters as color temperature -400K and no brightness change, and performs DTW comparison with the ideal feedback template of nighttime reading with an average color temperature reduction of 350-450K, resulting in a feedback matching value of 0.92, which is higher than the benchmark of 0.85; the system determines that it is an adapted scheme and immediately pushes it as the final lighting scheme; if the user feedback is that the color temperature is only reduced by 200K but the mode is frequently switched, the matching value is 0.72. The system identifies two differences: insufficient color temperature reduction and dynamic rhythm interference, and generates two correction suggestions: reducing the color temperature to 2700K and disabling the dynamic mode. After writing back to the secondary optimization scheme, the parameter set is regenerated, and after passing the verification again, the final scheme is output.

[0140] This embodiment provides an AI-driven personalized lighting service method. By extracting real-time user feedback parameters from secondary optimization schemes and ideal feedback parameters from standard illumination features, the method verifies the similarity between the real-time and ideal feedback parameters to obtain a feedback matching value. Secondary optimization schemes with feedback matching values ​​greater than the benchmark feedback value are selected as adaptation schemes and determined as the final lighting scheme. Secondary optimization schemes with feedback matching values ​​less than the benchmark feedback value are identified as schemes to be adjusted, and parameter correction suggestions are generated based on the difference parameters to update the scheme. This achieves objective quantification of subjective experience, rapid closure of high-matching schemes, and precise iterative correction of low-matching schemes. It can semantically align real-time user interaction behavior with ideal feedback templates, enabling the system to shift from relying on preset models to continuously learning real user behavior. This promotes the evolution of lighting schemes from static adaptation to dynamic co-evolution, fundamentally solving the industry pain point of parameters meeting standards but poor experience. It significantly improves the accuracy of personalized services, user trust, and the long-term adaptability of the system.

[0141] Furthermore, this embodiment of the invention also proposes a storage medium storing an AI-driven personalized lighting service program, which, when executed by a processor, implements the steps of the AI-driven personalized lighting service method described above.

[0142] Furthermore, this invention also proposes an AI-driven personalized lighting service system, the system comprising: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the steps of the AI-driven personalized lighting service method as described in any of the above.

[0143] Other embodiments or specific implementations of the AI-driven personalized lighting service system described in this invention can be found in the above-described method embodiments, and will not be repeated here.

[0144] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An AI-driven method for providing personalized lighting services, characterized in that, The method includes: Receive a lighting customization request from the user, the lighting customization request including scene requirement information, and determine the corresponding sub-lighting model in the AI ​​lighting model library as the target lighting model based on the scene requirement information; Obtain the personalized lighting interface corresponding to the target lighting model, and customize and update the lighting control interface on the user end according to the personalized lighting interface to generate a user-customized interface. According to the personalized lighting interface, the environmental data and preference data received from the user terminal are sent to the AI ​​decision server of the cloud lighting platform corresponding to the target lighting model. The environmental perception unit of the AI ​​decision server is called to split and parse the environmental data and preference data to obtain multiple sub-lighting parameter sets. The lighting optimization unit at the AI ​​decision server verifies the lighting adaptability of multiple sub-lighting parameter sets to obtain a personalized lighting scheme, which is then sent to the personalized lighting interface on the user's end for display.

2. The AI-driven personalized lighting service method as described in claim 1, characterized in that, The process of receiving a lighting customization request from a user, the lighting customization request including scene requirement information, and determining a corresponding sub-lighting model from the AI ​​lighting model library as the target lighting model based on the scene requirement information, includes: In response to the lighting customization request, obtain the user requirement tags corresponding to the scene requirement information. The user requirement tags include scene type tags, preference tags, and environment tags. The AI ​​lighting model library includes multiple sub-lighting models, and each sub-lighting model is set with a corresponding model function label; The sub-lighting model corresponding to the model function label of the user requirement label is determined as the target lighting model.

3. The AI-driven personalized lighting service method as described in claim 1, characterized in that, The environmental perception unit at the AI ​​decision server is invoked to split and parse the environmental data and preference data to obtain multiple sub-lighting parameter sets, including: Based on the environmental perception unit, the environmental data and preference data are split and processed to obtain multiple sub-data modules, including a spatial data module, a temporal data module, a user behavior data module, and a natural light data module. Determine the baseline parameter set for the standard lighting scene and the adjustment weights corresponding to each of the sub-data modules; The sub-data modules corresponding to the benchmark parameter set are obtained as benchmark data modules. Standard illumination features are extracted from the benchmark data modules. The parameter extraction intervals in each sub-data module are obtained according to the maximum adjustment threshold of the standard illumination features and the adjustment weight. Parameter points whose parameter values ​​are within the lighting parameter threshold in the parameter extraction interval are identified as valid parameter points, and a set of valid parameter points is generated based on adjacent valid parameter points. Generate a sub-lighting parameter set corresponding to the sub-data module based on the set of valid parameter points; The sub-lighting parameter sets are arranged and analyzed from high to low according to the influence priority of each sub-data module to obtain the sub-lighting parameter sequence.

4. The AI-driven personalized lighting service method as described in claim 3, characterized in that, The step of obtaining the sub-data module corresponding to the benchmark parameter set as the benchmark data module, extracting standard illumination features from the benchmark data module, and obtaining the parameter extraction interval in each sub-data module according to the maximum adjustment threshold of the standard illumination features and the adjustment weight includes: Retrieve the standard scene parameter template corresponding to the lighting customization request. The standard scene parameter template includes a core parameter area. Obtain the standard parameter area in the benchmark data module based on the core parameter area. Based on the lighting parameter threshold, extract the standard illumination features within the standard parameter region, and obtain the maximum adjustment threshold of the standard illumination features; Using the adjustment weight corresponding to each of the sub-data modules as the benchmark weight, the range of parameters in each of the sub-data modules that are positively correlated with the benchmark weight and are far from the maximum adjustment threshold of the benchmark weight is obtained as the division interval; The region corresponding to the baseline weight and the division interval in the sub-data module is obtained as the parameter extraction interval of the corresponding sub-data module.

5. The AI-driven personalized lighting service method as described in claim 4, characterized in that, The step of extracting standard illumination features within the standard parameter region based on the illumination parameter threshold, and obtaining the maximum adjustment threshold of the standard illumination features, includes: Parameter points whose parameter values ​​within the standard parameter region are within the lighting parameter threshold are identified as standard parameter points. A set of standard parameter points is generated based on adjacent standard parameter points, and a standard illumination feature is generated based on the set of standard parameter points. Extract the standard adjustment curve of the standard illumination feature, and obtain the peak fluctuation amplitude of the standard adjustment curve as the maximum adjustment threshold corresponding to the standard illumination feature. The waveform of the standard adjustment curve is a time-parameter change curve.

6. The AI-driven personalized lighting service method as described in claim 5, characterized in that, The step of verifying the lighting adaptability of multiple sub-lighting parameter sets based on the lighting optimization unit at the AI ​​decision server to obtain a personalized lighting scheme includes: Based on the illumination optimization unit, obtain the cooperative adjustment point corresponding to each sub-illumination parameter set in the sub-illumination parameter sequence; The sub-lighting parameter sets are fused according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set to obtain a primary lighting scheme; Extract the actual adjustment curve of the primary lighting scheme, locate the feature points of the standard adjustment curve based on the feature points of the actual adjustment curve, and superimpose the standard adjustment curve on top of the actual adjustment curve; The parameter points that overlap between the actual adjustment curve and the standard adjustment curve are obtained as matching parameter points. The number of matching parameter points and the total number of parameter points corresponding to the actual adjustment curve are counted. The parameter matching degree is obtained based on the ratio of the number of matching points to the total number. Identify the primary lighting schemes whose parameter matching degree is less than the benchmark matching degree as schemes to be optimized, and mark the set of sub-parameters that need to be adjusted in the schemes to be optimized; A primary lighting scheme whose parameter matching degree is greater than or equal to the baseline matching degree is determined as a secondary optimization scheme; The secondary optimization scheme and the standard illumination characteristics are adapted and verified to obtain the final lighting scheme or the scheme to be adjusted. Based on the final lighting scheme or the scheme to be adjusted, a personalized lighting scheme is obtained.

7. The AI-driven personalized lighting service method as described in claim 6, characterized in that, The step of obtaining the coordinated adjustment point corresponding to each sub-illumination parameter set in the sub-illumination parameter sequence based on the illumination optimization unit includes: The sub-lighting parameter sets are processed in a time sequence to extract multiple adjustment time periods corresponding to each sub-lighting parameter set; Obtain the maximum and minimum parameter values ​​corresponding to multiple adjustment time periods in the sub-lighting parameter set; The adjustment period corresponding to the maximum parameter value is determined as the peak adjustment period, and the adjustment period corresponding to the minimum parameter value is determined as the trough adjustment period; If the peak adjustment period and the trough adjustment period are adjacent, the boundary time of the adjacent periods is determined as the coordinated adjustment point; If the peak adjustment period and the trough adjustment period are not adjacent, then the end time of the peak adjustment period is taken as the first adjustment point, and the start time of the trough adjustment period is taken as the second adjustment point. Connect the first adjustment point and the second adjustment point to obtain the adjustment transition interval, and determine the midpoint of the adjustment transition interval as the corresponding cooperative adjustment point of the sub-lighting parameter set.

8. The AI-driven personalized lighting service method as described in claim 6, characterized in that, The step of fusing the sub-lighting parameter sets according to the sub-lighting parameter sequence and the coordinated adjustment points of each sub-lighting parameter set to obtain a primary lighting scheme includes: The first sub-lighting parameter set in the sub-lighting parameter sequence is obtained as the first fusion parameter set, and the sub-lighting parameter set adjacent to the first fusion parameter set is obtained as the second fusion parameter set; The point at which the first fusion parameter set and the second fusion parameter set are coordinated is determined as the fusion start point, and the point at which the second fusion parameter set and the first fusion parameter set are coordinated is determined as the fusion end point; Using the fusion starting point as a reference point, the parameters of the fusion ending point are controlled to transition towards the fusion starting point until the parameter change rates of the fusion starting point and the fusion ending point are consistent, at which point the transition of the fusion ending point stops. The fusion processing of the first fusion parameter set and the second fusion parameter set is completed based on the transition processing of the fusion start point and the fusion end point; The second fusion parameter set is updated to the first fusion parameter set, and the adjacent sub-lighting parameter sets of the updated first fusion parameter set are obtained as the next second fusion parameter set; Repeat the above steps of transition processing for the fusion start point and fusion end point, and perform fusion processing on each of the sub-lighting parameter sets to obtain a primary lighting scheme.

9. The AI-driven personalized lighting service method as described in claim 6, characterized in that, The process of performing secondary adaptation and verification of the secondary optimization scheme and the standard illumination characteristics to obtain the final illumination scheme or the scheme to be adjusted includes: Extract the real-time user feedback parameters from the secondary optimization scheme and the ideal feedback parameters from the standard illumination features; The feedback similarity verification is performed on the real-time feedback parameters and the ideal feedback parameters to obtain the feedback matching value. The secondary adaptation verification includes feedback similarity verification. The secondary optimization scheme with a feedback matching value greater than the benchmark feedback value is obtained as the adaptation scheme, and the adaptation scheme is determined as the final lighting scheme; The secondary optimization scheme with a feedback matching value less than the benchmark feedback value is identified as the scheme to be adjusted. Parameter correction suggestions are generated based on the difference parameters in the scheme to be adjusted. The secondary optimization scheme is then updated in conjunction with the parameter correction suggestions to obtain the final lighting scheme.

10. An AI-driven personalized lighting service system, characterized in that, The system includes: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the steps of the AI-driven personalized lighting service method as described in any one of claims 1-9.

Images