Light control method and system based on AI model analysis

Abstract

The invention relates to a light control method and system based on AI model analysis, and the method comprises the steps: obtaining environment illumination data of a target scene and user interaction input, and carrying out the semantic demand analysis, and obtaining semantic description information; performing illumination demand analysis on the semantic description information through a pre-trained light driving model to obtain illumination demand parameters; performing light field demand analysis on the illumination demand parameters according to the environment illumination data and the user interaction input, and generating light field control information; and acquiring power load data, performing instruction construction on the light field control information, and generating a light control instruction. According to the invention, dynamic self-adaptive regulation and control of the light environment can be realized, and the limitation that the existing method only depends on static compensation and cannot adapt to complex scene changes is overcome.

Description

Technical Field

[0001] The present application relates to the field of lighting control technology, and in particular to a lighting control method and system based on AI model analysis. Background Art

[0002] With the rapid development of artificial intelligence (AI) technology, how can AI models achieve multi-dimensional, precise control of lighting to meet personalized user needs while maintaining energy efficiency? Existing lighting control methods typically rely solely on ambient light sensor data or preset scene modes, lacking in-depth analysis of user semantic needs and failing to fully consider the coordinated optimization of dynamic light fields and power loads. Summary of the Invention

[0003] In order to overcome the problems existing in the related technologies, the present application provides a lighting control method and system based on AI model analysis, which can realize dynamic adaptive regulation of the light environment, overcoming the limitations of existing methods that only rely on static compensation and cannot adapt to complex scene changes.

[0004] This application provides a lighting control method based on AI model analysis, including: Obtain the ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; Performing a lighting demand analysis on the semantic description information through a pre-trained lighting driving model to obtain lighting demand parameters; Performing a light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input to generate light field control information; The power load data is obtained, and the light field control information is constructed into instructions to generate lighting control instructions.

[0005] Preferably, the step of acquiring the ambient lighting data of the target scene and the user interaction input, performing semantic requirement analysis, and obtaining semantic description information includes: Collecting spectral distribution data of the target scene through a preset light sensor to obtain an original light sampling signal; Performing frequency domain filtering and noise suppression on the original light sampling signal to generate ambient light data; Receive user voice input of the target scene through a preset voice interaction device, perform voice recognition, and obtain text interaction data; The environmental lighting data and the text interaction data are combined through semantic coding to output the semantic description information.

[0006] Preferably, the lighting requirement analysis of the semantic description information using a pre-trained lighting driving model to obtain lighting requirement parameters includes: Inputting the semantic description information into the light driving model, and performing context association analysis on the semantic description information through the semantic understanding layer of the light driving model to obtain semantic features; The intention inference layer calculates user requirements based on the semantic features to obtain initial lighting requirement information; Performing color temperature compensation and brightness adjustment on the initial lighting requirement information by combining the ambient lighting data and the user interactive input through a parameter optimization layer to obtain color temperature requirement parameters, brightness requirement parameters and dynamic change parameters; The color temperature requirement parameter, brightness requirement parameter and dynamic change parameter are integrated into multiple dimensions through the output fusion layer to output the lighting requirement parameter.

[0007] Preferably, the parameter optimization layer combines the ambient light data and the user interactive input to perform color temperature compensation and brightness adjustment on the initial lighting requirement information to obtain color temperature requirement parameters, brightness requirement parameters and dynamic change parameters, including: The optical parameter decomposition unit of the parameter optimization layer analyzes the initial lighting requirement information, extracts spectral features, and generates basic color temperature parameters and basic brightness parameters; Calculating the color temperature requirement of the ambient light data for the basic color temperature parameter, and outputting the color temperature requirement parameter; Obtain historical interaction information, perform preference analysis based on the user interaction input, and output user preference information; Performing a brightness demand analysis on the basic brightness parameter according to the user preference information to obtain a brightness demand parameter; The color temperature requirement parameter is dynamically matched with the brightness requirement parameter to generate a dynamically changing parameter.

[0008] Preferably, calculating the color temperature requirement of the ambient light data for the basic color temperature parameter and outputting the color temperature requirement parameter includes: Analyzing spectral components from the ambient light data to extract spectral distribution data; Calculating a color temperature offset for the basic color temperature parameter according to the spectral distribution data to obtain a color temperature compensation value; Performing color temperature superposition calculation on the color temperature compensation value and the basic color temperature parameter to generate a corrected color temperature parameter; The color temperature range constraint is performed on the corrected color temperature parameter to obtain the color temperature requirement parameter.

[0009] Preferably, performing light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input to generate light field control information includes: Extracting light intensity information and color temperature gradient data from the ambient light data; Analyze the intention of the user's interactive input and identify explicit demand information and implicit behavior information; Performing basic compensation calculation according to the light intensity information and the explicit demand information to obtain basic light compensation parameters; Adjusting the color temperature gradient data based on the implicit behavior information to generate a color temperature adjustment coefficient; A light field synthesis calculation is performed on the basic light compensation parameter and the color temperature adjustment coefficient according to the light requirement parameter to obtain the light field control information.

[0010] Preferably, performing basic compensation calculation according to the light intensity information and the explicit demand information to obtain basic light compensation parameters includes: Segmenting the illumination intensity information into intervals based on a preset illumination interval threshold to obtain high light area data and low light area data; For the highlight area data, combining the explicit demand information, performing dynamic area brightness balance, and outputting highlight area balance parameters; For the low-light area data, performing minimum illumination compensation according to the explicit requirement information to generate low-light area compensation parameters; The highlight area balance parameter and the light area compensation parameter are integrated for basic compensation to obtain the basic illumination compensation parameter.

[0011] Preferably, the acquiring of power load data, constructing instructions for the light field control information, and generating lighting control instructions include: Identifying the grid load of the target scenario and monitoring the load in real time to obtain the power load data; assigning priorities to the light field control information based on the power load data to form a light field control sequence; Performing time-division power matching on the light field control sequence to obtain time-division control parameters; Using the time-division control parameters to perform instruction encoding on the light field control information to generate an initial control instruction set; The load conflict of the initial control instruction set is detected and dynamically adjusted, and the lighting control instruction is output.

[0012] This application also provides a lighting control system based on AI model analysis, which is applied to any of the above-mentioned lighting control methods based on AI model analysis, including: A recognition module is used to obtain ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; An analysis module is used to analyze the lighting requirements of the semantic description information using a pre-trained lighting driving model to obtain lighting requirement parameters; a processing module, configured to perform a light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input, and generate light field control information; A construction module is used to obtain power load data, perform instruction construction on the light field control information, and generate lighting control instructions.

[0013] The technical solution provided by this application may have the following beneficial effects: This application solves the problem of lighting control deviation caused by insufficient semantic understanding in traditional systems by converting user interaction input into quantifiable lighting demand parameters, thereby improving the satisfaction of users' personalized needs. By combining ambient lighting data and user interaction input to analyze light field needs, dynamic adaptive control of the light environment is achieved, overcoming the limitations of existing methods that rely solely on static compensation and cannot adapt to complex scene changes, and improving the comfort and consistency of the light environment. By introducing power load data to construct instructions for light field control information, it is ensured that the lighting control strategy meets user needs while taking into account grid load balancing, avoiding energy waste and system overload risks, and improving overall energy efficiency. Based on the pre-trained lighting drive model, it can automatically adjust lighting parameters according to different scenarios and user needs, enhance adaptability and intelligence, and enable it to be widely used in various scenarios such as home, office, and business.

[0014] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other objects, features and advantages of the present application will become more apparent through a more detailed description of exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0016] Figure 1 This is a flow chart of a lighting control method based on AI model analysis shown in this application; Figure 2 This is a structural diagram of a lighting control system based on AI model analysis shown in this application. DETAILED DESCRIPTION

[0017] The preferred embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. Instead, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0018] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0019] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0020] Reference Figure 1 As shown, the present application provides a lighting control method based on AI model analysis, including: Step S1: Obtain the ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; Step S2: Analyze the lighting requirements of the semantic description information using the pre-trained lighting driving model to obtain lighting requirement parameters; Step S3: performing light field demand analysis on the lighting demand parameters according to the ambient lighting data and user interaction input, and generating light field control information; Step S4: Obtain power load data, construct instructions for light field control information, and generate lighting control instructions.

[0021] Based on the above steps, the detailed process is as follows: Step S1: The light sensor collects the spectral distribution signal of the target area, including visible light and infrared radiation intensity. The acquisition process eliminates 50Hz power supply interference and uses an adaptive threshold noise reduction algorithm to suppress ambient electromagnetic noise, generating standardized ambient light data. Simultaneously, a directional microphone captures raw speech input, removes reverberation through spectral subtraction, and then segments valid speech segments. The audio segments are converted into textual interaction data, and an encoder-based mapping between dialects and technical terminology is processed. The semantic requirement parsing unit receives both ambient light data and textual interaction data and performs cross-modal feature fusion: It discretizes the real-time light intensity into light level labels and combines them with spatial location keywords (such as "sofa area" and "workbench") extracted through text entity recognition to generate structured semantic description information. This information is encapsulated in JSON-LD format and includes a timestamp, scene partition identifier, light status code, and user intent encoding.

[0022] Step S2: The lighting-driven model utilizes a three-layer cascaded neural network architecture. The first semantic understanding layer uses a Bi-LSTM network to analyze contextual associations within semantic descriptions, extracting feature vectors of spatial location and temporal behavior. For example, when input contains the phrase "lower brightness in reading area," the model identifies the "reading area" entity and associates it with the preset location coordinates. The "lower brightness" action then associates it with a negative brightness adjustment flag.

[0023] The second-level intention inference layer receives feature vectors, classifies user demand types through a fully connected network, and outputs an initial lighting demand information matrix, which includes basic color temperature values, basic brightness values, and expected change trends.

[0024] The third-level parameter optimization layer performs environmental compensation operations: it reads the spectral energy distribution in the ambient light data in real time, calculates the deviation between the current color temperature and the target color temperature, and applies the color adaptation transformation algorithm to generate the color temperature compensation coefficient. At the same time, based on the brightness adjustment records in the user's historical interaction log, it fits the personalized brightness preference curve and corrects the initial brightness value.

[0025] The final output lighting requirement parameters are encapsulated as a four-dimensional vector, which represents the target color temperature (unit: K), target brightness (unit: lux), gradient time (unit: s) and brightness fluctuation tolerance threshold.

[0026] Step S3: The target scene's ambient lighting data is converted to obtain its spectral energy distribution, and an interpolation algorithm is used to generate a three-dimensional color temperature gradient field and illuminance distribution field. The system also simultaneously processes the semantic features of user interaction input, separating explicit operational instructions from implicit behavioral patterns. Explicit instructions are directly mapped to a library of preset lighting strategies (e.g., "Conference Mode" activates a uniform lighting template), while implicit patterns are derived by inferring the user's underlying intentions (e.g., frequent dimming triggers energy-saving mode).

[0027] The light field demand analysis engine integrates three input sources—lighting demand parameters, the ambient gradient field, and user demand classification results—to perform dynamic grid-based zoning calculations. Using the user's location as the origin of the spherical coordinate system, the engine divides the lighting sector by azimuth and calculates the compensation difference between the current illuminance and the target value for each sector. It also divides the vertical lighting levels (work plane layer, decorative atmosphere layer, and safety indicator layer) according to the functional attributes of the space and assigns differentiated brightness weighting coefficients. The resulting light field control information is encapsulated in a topological structure, with data units containing the regional grid number, target color temperature correction, brightness compensation value, and gradient time parameter.

[0028] Step S4: Smart meters collect bus voltage harmonic distortion, real-time load factor, and dynamic parameters of power factor angle to construct a load state matrix. The command optimization engine analyzes the topological structure of the light field control information and performs four-dimensional resource scheduling. The first dimension prioritizes devices according to electrical safety regulations. The second dimension generates command execution time windows based on the load state matrix, automatically delaying commands for non-critical devices during peak grid periods (load factor > 85%). The third dimension uses association rule mining to identify mutually exclusive groups of devices (e.g., a spotlight group and a projector cannot be turned on simultaneously) and construct a conflict avoidance table. The fourth dimension calculates the optimal phase allocation solution.

[0029] The reconstructed lighting control instructions are encapsulated in a timestamp-driven DALI-2 protocol frame. Each data packet contains: a 16-bit device address, an 8-bit dimming instruction (high 4-bit brightness value / low 4-bit color temperature group), a 4-bit fault flag code, and a 24-bit cyclic redundancy check code.

[0030] The present application provides a lighting control method based on AI model analysis. By converting user interaction input into quantifiable lighting demand parameters, it solves the problem of lighting control deviation caused by insufficient semantic understanding in traditional systems and improves the satisfaction of users' personalized needs. By combining ambient lighting data and user interaction input to analyze light field demand, dynamic adaptive control of the light environment is achieved, overcoming the limitation of existing methods that rely only on static compensation and cannot adapt to changes in complex scenes, and improving the comfort and consistency of the light environment. By introducing power load data to construct instructions for light field control information, it ensures that the lighting control strategy meets user needs while taking into account the load balance of the power grid, avoids energy waste and system overload risks, and improves the overall energy efficiency ratio. Based on the pre-trained lighting drive model, it can automatically adjust lighting parameters according to different scenarios and user needs, enhance adaptability and intelligence, and enable it to be widely used in various scenarios such as home, office, and business.

[0031] Furthermore, in one embodiment, the ambient lighting data of the target scene and the user interaction input are obtained, and semantic requirements are parsed to obtain semantic description information, including: Composite spectral sensors covering visible and near-infrared wavelengths were deployed within the target scene. All sensor nodes were positioned and installed according to spatial layout specifications, ensuring a vertical height of 0.75 meters above a ground reference plane and a 30-degree pitch angle. The sensor array's hardware circuitry was activated, activating synchronous sampling to capture radiance data across the 380-1100 nm spectral range at a rate of 200 frames per second.

[0032] Each frame of sampling synchronously triggers the position marking module to generate a three-dimensional coordinate identifier, achieving millimeter-level positioning resolution. The raw electrical signal output is accompanied by a timestamp and device serial number, and is encapsulated as a multidimensional data matrix via an industrial bus protocol. The horizontal dimension of this matrix corresponds to discrete measurement points in physical space, while the vertical dimension stores the measured instantaneous radiant flux of each spectral channel, measured in watts per square meter per solid angle. Immediately after deployment, an on-site baseline calibration process is performed, using a standard blackbody cavity to eliminate dark current noise floor and simultaneously performing device response linearity correction to obtain the raw illumination sampling signal.

[0033] Multi-stage joint processing is performed on the original light sampling signal. In the initial processing stage, three characteristic frequency bands are divided to implement differentiated filtering operations: power frequency interference is suppressed in the low-frequency band of 0 Hz to 5 Hz, retaining the effective light signal below the microvolt level; the core characteristics of the natural waveform of the ambient light are maintained in the medium-frequency band of 5 Hz to 50 Hz; and switching circuit noise elimination is performed on the high-frequency band above 50 Hz.

[0034] The signal denoising module uses an adaptive wavelet threshold algorithm to complete the wavelet decomposition operation, dynamically matching the local signal-to-noise ratio of the signal to select the optimal wavelet basis function. A fixed 1-second window is set to perform outlier detection on consecutive sampling points, and the detected abnormal data value is automatically replaced with the median of the measurement values ​​of two adjacent frames. The reconstructed ambient light data is output and mapped to a two-dimensional coordinate grid. Each unit records the tristimulus value, the correlated color temperature parameter, and the measured illuminance value. The spatial resolution is set to 0.1 m x 0.1 m. The output data is verified by the quality control system to improve the signal-to-noise ratio before and after frequency domain filtering. After packaging, a standardized light environment distribution map is generated.

[0035] Multiple sets of directional microphones are installed within the target scene space. Each microphone unit is positioned 1.5 meters above the ground according to standard acoustic measurement positions, with a 120-degree angle distribution to create an omnidirectional pickup coverage area. Upon startup, the voice interaction device automatically runs a sound field calibration program, loading preset impulse response parameters to compensate for ambient reverberation. When a valid human voice frequency band is detected, the device activates the beamforming unit to enhance the voice signal strength in the target direction and simultaneously performs voiceprint feature filtering to isolate interfering sound sources from non-target individuals.

[0036] The processed audio signal is digitized at a 32kHz sampling rate and fed into the speech processing unit in a standardized compressed format for conversion. The speech processing unit converts the sound wave sequence into a text sequence based on a pre-compiled vocabulary rule library. The core conversion process includes spectral feature comparison, language rule matching, and contextual correction. The resulting text interaction data is structured and contains text content, time stamps, and confidence values. The device automatically records a summary of the acoustic features during the recognition process for subsequent quality optimization.

[0037] A cross-data domain fusion processing framework is established to process lighting parameters and text instruction information. The light distribution map in the ambient lighting data is mapped to the scene's physical space via a position coordinate system conversion. The key elements of the text interaction data are converted into a set of operation elements through grammatical structure analysis, extracting the three elements of core operation target, action type, and intensity description. The lighting spatial information and text instruction elements are precisely aligned on the execution time axis, with a maximum allowable offset threshold of 50 milliseconds. Logical relationship mapping is implemented in the semantic fusion stage: the spatial object name of the text instruction is associated with the corresponding coordinate position in the lighting data grid; the action intensity descriptor is converted into a numerical adjustment ratio; and the light parameter attribute words are translated into specific executable lighting control variables.

[0038] The encoded output forms a semantic description data structure, which includes coordinate location identifiers, lighting adjustment type, proportional change, effective area range, and execution priority parameters. During information generation, the system automatically detects conflicts between the environmental state and the rationality of the instruction logic, and outputs a revised executable semantic description based on a preset rule base.

[0039] This embodiment fully captures the scene light distribution characteristics through the standardized layout and high-frequency sampling mechanism of spectral sensors, and combines frequency domain band processing technology and correction methods to significantly improve the spatial resolution and measurement stability of ambient light data. After gridding, the light environment information forms a high-precision two-dimensional distribution map, providing a reliable physical environment basis for intelligent control. The voiceprint feature filtering mechanism effectively blocks non-command interference, and the sound field calibration program automatically compensates for the reverberation effect. The speech-to-text conversion process integrates spectral feature comparison and context correction processing to ensure that text interaction data accurately reflects the user's operational demands. Through the fusion of lighting parameters and text instructions, the natural language description is accurately mapped to the physical space coordinates, realizing the structured coupling of environmental status and user needs. The joint application of action intensity quantification and conflict detection rule library ensures the operational feasibility of semantic description information and forms consistent control of physical space and behavioral intentions.

[0040] In one embodiment, a pre-trained light driving model is used to analyze the lighting requirements of the semantic description information to obtain lighting requirement parameters, including: The lighting-driven model consists of a semantic understanding layer, an intent inference layer, a parameter optimization layer, and an output fusion layer. The semantic understanding layer utilizes a graph attention network (GAT) architecture, where nodes are represented as the spatial coordinates of illuminated areas within the building, and edge weights define the light propagation coefficients between areas. The spatial coordinate parsing unit calculates node feature vectors using a graph attention mechanism, aggregating the optical characteristic parameters of adjacent areas. The temporal modeling module integrates a gated recurrent unit to process historical user action time series data and output a feature tensor with a time decay factor. The language processing unit implements word embedding projection operations, mapping action vocabulary into intensity gradient scalars in a 300-dimensional semantic space. The feature verification unit compares coordinate topological relationships using a cosine similarity algorithm, generating a four-dimensional semantic feature tensor with confidence weights.

[0041] The intent inference layer is constructed using a two-channel fully connected network. Channel one is a 128-node dense network that converts brightness requirements, while channel two is a color temperature mapping network with radial basis functions. The basic requirement conversion unit generates illumination gradient values ​​through matrix operations in the fully connected layer. The user adaptation unit performs a non-negative matrix factorization algorithm to decompose the user profile library into a 32-dimensional feature matrix, which is loaded into the preference compensation module. The physiological rhythm unit is equipped with a Gaussian kernel function generator and outputs circadian rhythm correction coefficients. The device constraint unit has a built-in driver parameter boundary table, and the out-of-bounds detection unit applies the ReLU function to truncate illegal values. The conflict arbitration unit runs the constraint propagation protocol to resolve parameter conflicts.

[0042] The parameter optimization layer deploys an adaptive compensation algorithm. The color temperature compensation unit calculates tristimulus value offsets based on standards and outputs primary color drive scaling coefficients. The brightness analysis unit implements a logarithmic function to process illuminance differences, and the compensation remapping unit applies the inverse of the light decay curve for nonlinear calibration. The dynamic adjustment unit uses an exponential moving average model to generate weighting factors, and continuous operation triggers the enhancement unit to increase the intensity of the effect. The device protection unit implements a gradual ramp constraint algorithm to limit the rate of parameter change within the device safety threshold.

[0043] The output fusion layer is constructed using a multi-domain fusion controller. The interpolation selection unit calls four curve generators: Bessel, linear, exponential, and sinusoidal. In the command conversion unit, the color temperature processing subunit outputs the current ratio using a PWM duty cycle conversion table, and the tilt control subunit generates the servo angle control word. The resource scheduling unit runs the banker's algorithm, and the priority manager assigns preemption levels to functional areas. The data compression unit implements inter-frame differential encoding, and the repeated field marking unit uses flag bits to reduce data capacity, ultimately outputting a standard DALI command stream.

[0044] More specifically, standardized, encapsulated semantic description information is imported into the pre-compiled lighting driver model, activating the contextual analysis capabilities of the semantic understanding layer. The spatial coordinate parsing unit deployed within the semantic understanding layer utilizes a graph attention network (GAT) architecture to locate the target area based on the location identifiers in the description information. Using the graph attention mechanism, it aggregates the optical properties of adjacent areas and simultaneously indexes the baseline brightness value and color temperature distribution of the corresponding location in the ambient lighting data.

[0045] The time series modeling module extracts the current operation time and historical contemporaneous records. A gated recurrent unit integrates with the time series modeling module to process the operation time series and establish a time decay correction factor. A pre-set synonym rule library is then invoked, and the language processing unit performs a 300-dimensional word embedding projection operation. This expands the feature dimension of the adjustment type description, identifying the intensity gradient relationship between action terms such as "enhance" and "elevate." The user behavior analysis unit retrieves a dataset of user operation records from the past ten days to detect the frequency characteristics and amplitude deviations implicit in repeated operations. This processing generates a four-dimensional semantic feature tensor in real time. The feature verification unit verifies the coordinate topology using a cosine similarity algorithm. Each column in the tensor records the feature type code, intensity conversion factor, time decay factor, and quality assessment weight. A completeness check is performed before output, and an exception marker is injected into any missing or conflicting coordinate parameters. When the feature tensor is transmitted to the lower-level processing unit, a spatial resolution identifier and color gamut coverage description are automatically appended.

[0046] The semantic feature tensor is input to the intent inference layer, initializing the demand calculation process. The basic demand conversion module constructs a dual-channel fully connected network (a 128-node dense network for the brightness branch and a radial basis function network for the color temperature branch) based on a standard parameter comparison table. This module converts the adjustment types in the semantic elements into physical reference values. Brightness adjustment maps to the illuminance gradient, and color temperature adjustment maps to the correlated color temperature offset.

[0047] The dynamic adaptation engine activates the user profile database. The user adaptation unit performs non-negative matrix decomposition to generate a 32-dimensional feature matrix and performs multi-dimensional demand correction: the baseline demand intensity is compensated by the adjustment amplitude deviation value based on the user's historical operation statistics. The circadian rhythm unit configures a Gaussian kernel function to generate circadian rhythm coefficients, matches the color temperature preference parameters for different time periods in the circadian rhythm parameter table, and loads the preset lighting mode associated with the activity type code. The device constraint unit has a built-in driver parameter boundary table and applies the ReLU truncation function. It calibrates the feasible domain of demand values ​​based on the lamp performance parameter table and automatically triggers threshold truncation for out-of-bounds parameters.

[0048] The calculation process generates an initial lighting requirement information structure, which contains the target illuminance value, target color temperature value, gradient time parameters, and transition curve type code. This structure includes a built-in device capability detection flag and injects a degraded execution flag into parameter items with driver constraints. The conflict arbitration unit runs the constraint propagation protocol to resolve parameter conflicts, performs demand conflict detection before output, and activates the emergency recalculation protocol when the time parameters and change range exceed the luminaire's response capabilities.

[0049] The initial lighting requirement information is imported into the parameter optimization layer to perform environmental adaptation processing. The parameter optimization layer deploys an adaptive compensation engine. The environmental state analysis unit loads the chromaticity distribution map in the ambient lighting data in real time, extracts the discrete data of the current color temperature value and the target color temperature parameter of the target area. The color temperature compensation module performs coordinate mapping based on the standard chromaticity diagram, runs the three stimulus value offset calculation, generates the primary color drive proportional coefficient and loads it into the adjustment parameter queue. The brightness analysis engine synchronously processes the illuminance distribution grid in the ambient lighting data, implements the logarithmic function to process the illuminance difference gradient, compares the target illuminance value of the initial requirement to obtain the difference gradient matrix, and the compensation remapping unit loads the inverse of the lamp light decay curve to implement nonlinear calibration.

[0050] The dynamic adjustment unit analyzes special adjustment markers from recent user operations and generates weighting factors using an exponential sliding average model. For continuous interventions, a parameter weighting mechanism is activated, superimposing the dynamic adjustment factors on the baseline compensation value. During the compensation process, the device protection unit implements a gradual ramp constraint algorithm to limit the parameter change rate to a safe threshold and automatically triggers a smooth transition algorithm for parameters that exceed the physical limits of the luminaire.

[0051] The optimization results are output as a structured data package containing the required color temperature and brightness parameters, as well as the dynamic change parameters. This package includes the compensated target color temperature, illuminance, gradient time range, and transition curve configuration code. A temperature drift warning flag is injected before data packet transmission, activating protective parameter corrections when the ambient temperature and humidity sensor values ​​exceed preset thresholds.

[0052] The multi-dimensional lighting requirement parameter set enters the output fusion layer for integration. The output fusion layer constructs a multi-domain fusion controller. The spatiotemporal alignment unit analyzes the effective time stamp of the color temperature requirement parameter and the spatial scope of the brightness requirement parameter, initiating a resynchronization protocol for parameters with spatiotemporal conflicts. The priority weighting engine invokes the scene weight table in the preset rule library, assigning a high priority coefficient to the requirements in the working area and applying an attenuation factor to the requirements in the transition area.

[0053] The parameter integration core component configures the code based on the transition curves of the dynamically changing parameters. The interpolation selection unit calls the Bessel / linear / exponential / sinusoidal curve generators according to the code, using an interpolation algorithm to generate parameter sequence frames with 0.1-second intervals. Each data unit within the sequence frame contains four-dimensional control instructions: the instruction conversion unit outputs the color temperature drive current ratio through the PWM duty cycle conversion table, and the illumination PWM duty cycle and tilt angle control subunit generates the servo angle control word and the auxiliary light channel enable flag. The resource scheduling unit runs the banker algorithm and configures the priority manager to detect device resource contention within the instruction set and initiate instruction sequence reconstruction or dynamic resource allocation strategies based on the priority coefficients.

[0054] The final output lighting requirement parameters are encapsulated as an instruction stream data structure. The data compression unit implements inter-frame differential encoding technology. The data structure is divided into a basic control segment and an extended information segment. The basic segment contains hexadecimal-encoded instructions that can directly drive the device, while the extended segment stores timing check codes, device status monitoring indexes, and abnormal rollback protocol identifiers. When generating the instruction stream, the repeated field marking unit uses flag bits to reduce the data volume, automatically compressing the data and removing redundant and repeated information between consecutive frames.

[0055] This embodiment uses the semantic understanding layer of the lighting driver model, combined with spatial coordinate parsing and temporal analysis, to establish a dynamic mapping between environmental states and user commands. The intent inference layer integrates international lighting standards and user behavior characteristics to generate precise physical parameter requirements. A hierarchical processing mechanism eliminates semantic ambiguity, ensuring that demand calculations conform to actual optical environment conditions. The chromaticity diagram-driven compensation scheme in the parameter optimization layer provides real-time correction for deviations between the target color temperature and physical constraints. The brightness compensation module utilizes lamp attenuation characteristics for calibration, and the user interaction data processing unit captures operational characteristics to update weight parameters. This mechanism dynamically adapts output parameters to device aging, environmental interference, and changes in user preferences. The spatiotemporal alignment protocol in the output fusion layer eliminates boundary conflicts in control commands, and four-dimensional command sequence generation technology enables coordinated management of color temperature driving, illumination adjustment, device angle, and auxiliary light sources. The conflict arbitration unit dynamically allocates resources based on scene priorities to ensure stable execution of lighting requirements in core areas. A three-level verification mechanism comprises a quality marker for the semantic feature tensor, a temperature drift warning for optimized parameters, and an abnormal rollback protocol for the command stream. Integrity verification and device compatibility verification are integrated throughout the processing flow to prevent control failures.

[0056] In one embodiment, the parameter optimization layer combines ambient light data and user interaction input to perform color temperature compensation and brightness adjustment on the initial lighting requirement information to obtain color temperature requirement parameters, brightness requirement parameters, and dynamic change parameters, including: Initial lighting requirement information is imported into the parameter optimization layer's optical parameter decomposition unit, initiating the processing flow. This unit loads the specified spectral analysis specifications and separates the physical quantity data packets of the lighting parameters from the initial requirement information. This processing activates the spatial feature separation mechanism, demarcating the target action area based on the spatial coordinate identifiers in the requirement information and establishing a chromaticity coordinate mapping index for that area. The spectral matching unit uses the standard light source feature library to perform type recognition, determine the color temperature type attributes of the required light source, and match the corresponding blackbody radiation trajectory coordinate points. The brightness analysis component intercepts the target illuminance value and its action range radius parameter, converting the luminous flux requirement value according to the measurement standard.

[0057] The separation operation is implemented in three stages: chromaticity coordinates are calculated to output the base value of correlated color temperature; illuminance gradient analysis generates a base brightness reference; and time-domain analysis extracts gradient time and transition curve characteristics. The resulting output is a dual-channel output of base color temperature parameters and base brightness parameters. The base color temperature parameter is packaged as the color temperature value, the color tolerance range, and the dominant wavelength offset vector. The base brightness parameter stores the target illumination value, minimum / maximum boundary thresholds, and an attenuation compensation factor. The parameter set is automatically transmitted with a spatiotemporal synchronization index tag for the ambient light data, allowing subsequent processing units to access the calibration data in real time.

[0058] The base color temperature parameters and the chromaticity distribution grid in the ambient light data are synchronously input into the color temperature demand calculation engine. The coordinate index markers in the base color temperature parameters are used to locate the corresponding grid cells in the ambient light data. The color difference analysis module extracts the coordinate parameters of the uniform color space within the grid cells and compares them with the target chromaticity points in the base color temperature parameters to calculate the original color difference value.

[0059] The chromatic adaptation conversion unit performs tristimulus value transformation, calculating and eliminating the deviation in the spectral power distribution between the ambient light and the target color temperature. This calculation process consists of four core operations: measuring the correlated color temperature of the ambient light source, solving the coefficients of the chromatic adaptation transformation matrix, applying the matrix to transform the target chromaticity point, and remapping the corrected color coordinates onto the chromaticity diagram. The optical path difference compensation module is also activated, applying the inverse square law to compensate for the radiant brightness attenuation of the color coordinates based on the optical distance parameter between the measurement plane and the luminaire as specified in the ambient light data.

[0060] Before output, light source consistency verification is performed to check whether the deviation of the corrected color coordinates from the blackbody locus exceeds the range of the third-order MacAdam ellipse. The final color temperature requirement parameters generated include the actual target color temperature value, the maximum color tolerance allowed value, the chromaticity coordinate correction amount, and the environmental adaptability assessment coefficient. The data is encapsulated in a standard format for color temperature requirement parameters.

[0061] The user operation record database stores historical control data and extracts the target user's effective operation sequences over the past six months. The analysis unit screens brightness and color temperature control events based on human-computer interaction specifications and establishes a time-space bidimensional analysis framework. It extracts operation frequency characteristics by daily time period and identifies differences in operation intensity during peak hours in the morning and evening. It correlates the operation location coordinates with environmental activity types and clusters high-frequency usage areas to form a heat map. Parameter deviation statistics calculate the deviation rate between the user-set value and the system's recommended value, converting this into a feature vector input weight allocation matrix. This outputs a user preference parameter table, including the time period activity coefficient, regional weight index, color temperature bias offset, and brightness adjustment amplitude ratio. Boundary constraints are applied before output to ensure that the preference values ​​are within the physical limits of the device.

[0062] Basic brightness parameters and user preference information are analyzed for demand. The preference mapping unit performs a composite operation based on the brightness adjustment amplitude ratio: a linear superposition strategy is used within the normal fluctuation range, and a logarithmic function progressive compensation mechanism is activated within the abnormal deviation range.

[0063] The time period correction module matches the current time parameters with the time period activity coefficient table in the preference information and dynamically scales the basic brightness value. The spatial adaptation unit parses the area identifier for the basic parameter and performs a convolution operation with the user area weight index to generate a gradient compensation value. The device state adaptation component loads the luminous flux characteristic curve from the device usage data record and calculates the attenuation compensation factor based on the operating hours for reverse correction. The result is a brightness requirement parameter structure containing the core illumination target value, the allowable fluctuation threshold, the effective time range, and the regional difference parameters. The structure is encapsulated with embedded data integrity indicators and abnormal status warning codes.

[0064] The color temperature and brightness requirement parameters are input into a dynamic response matching system for collaborative optimization. A time alignment unit analyzes the temporal attribute differences between the two parameters and achieves time domain synchronization through a time remapping protocol. The brightness regional difference coefficient and color temperature spatial distribution characteristics are extracted to perform coupling compatibility testing.

[0065] An optimization calculation method is applied to solve the optimal dynamic response curve and generate three control logic paths: a synchronous control path, in which color temperature and brightness gradually change on the same time scale; a sequential control path, in which color temperature changes are triggered after 60% of the brightness adjustment is completed; and a regional decoupling path, in which high-weight regional parameters are independently controlled.

[0066] The control logic path results are converted into dynamically changing parameters, including the master mode identifier, the gradual change time parameter, the response function index, and the synchronization flag bit. The conflict resolution component detects resource contention and reconfigures the control logic according to the preset priority.

[0067] This embodiment automatically identifies the differences in lighting preference intensity at different time periods. A spatial clustering algorithm generates a regional heat distribution map to accurately capture the user's core activity area. Brightness offset statistics quantify the user's personalized adjustment characteristics to solve the problem of deviation between standardized lighting parameters and actual needs. The linear and logarithmic dual-mode operation mechanism of the preference adjustment amplitude ratio adapts to different scenarios of routine operation and abnormal deviation. The dynamic scaling function of the time period activity coefficient realizes precise adaptation to circadian rhythms. The reverse calibration of the device light attenuation curve eliminates the impact of hardware performance degradation and ensures the long-term stability and effectiveness of the output parameters. Response surface modeling technology realizes the parameter fusion control of color temperature and brightness, and provides three control paths: synchronization / sequential / decoupling. The regional differentiation processing mechanism supports spatial weight zoning control. The conflict arbitration strategy ensures the stability of the core area through priority reassignment, breaking through the problem of parameter conflict in traditional lighting systems.

[0068] In one embodiment, color temperature requirement calculation is performed on the basic color temperature parameters based on the ambient light data to obtain the color temperature requirement parameters, including: The ambient light data is fed into the spectral analysis unit for component decomposition. This unit activates a multi-channel signal separation mechanism, breaking down the radiant brightness data into discrete spectral components within the 380nm-780nm range according to the visible spectrum band division rules defined by the standard.

[0069] The analysis process includes three operations: first, activating the spectral matching filter group to eliminate the stray light interference components in the equipment acquisition; second, performing characteristic wavelength extraction to identify the mixing ratio coefficient of natural light source and artificial light source; finally, performing band energy integral calculation to generate the radiation flux distribution table of each 5nm interval band.

[0070] The processing process applies a linear interpolation algorithm to fill in sensor blind spots, using the interpolation benchmark as a guideline for the energy gradient trends of adjacent channels. The output spectral distribution data uses a standardized matrix structure, with row vectors corresponding to spatial grid coordinates and column vectors storing wavelength-radiant flux pairs, along with additional light source type identifiers and ambient light mixing coefficients. Spectral fidelity verification is performed before data packaging, with the color difference threshold range calculated by inversely calculating the color coordinates.

[0071] The spectral distribution data and the basic color temperature parameters are synchronously loaded into the offset calculation engine. The spatial mapping unit locates the corresponding row vector in the spectral matrix based on the coordinate index mark in the basic color temperature parameters to establish the spectral scope of the target area.

[0072] The color difference analysis core performs a two-stage calculation: the first stage solves the theoretical spectral difference between the current spectral distribution and the target color temperature, and uses a weighted algorithm to calculate the deviation weight of each wavelength band; the second stage evaluates the impact of ambient light interference and activates the adaptive compensation mechanism when it is identified that the proportion of natural light components exceeds 30%.

[0073] The compensation logic performs two operations: adjusting the dominant wavelength offset vector to match the blackbody radiation trajectory; and analyzing the fluctuation amplitude of the correlated color temperature to generate a tolerance compensation coefficient. The calculation engine outputs a color temperature compensation value data structure consisting of a base compensation Kelvin value, a secondary compensation gradient value, and a wavelength domain compensation vector. Device driver capability verification is performed before output, triggering hierarchical dimensionality reduction for required values ​​outside the programmable color temperature range.

[0074] The base color temperature parameters and color temperature compensation values ​​are simultaneously imported into the overlay calculation core for parameter fusion. This calculation process activates a two-stage fusion logic: the primary overlay unit performs an algebraic addition operation on the reference color temperature value in the base color temperature parameters and the principal component of the compensation value to generate an intermediate compensation value.

[0075] The secondary gradient processing unit loads the gradient components of the compensation value and implements spatial gradient overlay based on the optical path distribution in the ambient lighting data. For example, linear compensation is used within 0.5 meters of the luminaire, while exponential decay compensation is initiated 0.5 meters away. The overlay operation is performed using the correlated color temperature conversion formula to perform parameter conversion, and tristimulus value verification is performed simultaneously to ensure that the calculated results meet the constraints of the chromaticity diagram.

[0076] The output of the corrected color temperature parameter structure contains four core data points: the fused color temperature value, the effective radius, the gradient control flag, and the color tolerance range. The structure is encapsulated with the overlay path flag, and the adaptive conversion protocol is triggered for calculation results that exceed the standard light source spectrum.

[0077] The corrected color temperature parameters are input into the range constraint system for feasibility verification. The color temperature programmable range table in the device driver feature library is retrieved and the corrected color temperature value is compared with the physical limit boundary value. The constraint process implements a dual-track verification system: the main channel executes the boundary value truncation algorithm and directly sets the out-of-limit parameters to the boundary extreme value; the auxiliary channel activates the smooth transition mechanism and calculates the gradient transition path from the current correction value to the allowable extreme value. The spatial adaptability unit implements the effective radius constraint processing based on the effective radius data in the parameters and the lamp layout distribution map: the mean constraint is activated for the overlapping coverage area, and the original radius is maintained for the independent control area. The final generated color temperature requirement parameters include the required color temperature value, the tolerance value, the spatial range identifier and the device compatibility mark.

[0078] This embodiment uses multi-channel signal separation technology and a characteristic wavelength extraction mechanism to accurately separate natural light and artificial light components in mixed light source scenarios. Spectral fidelity control ensures that reconstructed color coordinates are consistent with human perception, eliminates the interference of ambient stray light on basic color temperature parameters, and overcomes the measurement accuracy bottleneck in complex lighting environments. A dual compensation operation integrates the reference color temperature and spectral offset, automatically switching between linear and exponential attenuation compensation modes based on spatial distance. Gradient superposition technology driven by the optical path distribution state achieves a smooth color temperature transition from the center to the edge of the lamp, solving the color shift problem caused by distance attenuation in traditional lighting systems.

[0079] In one embodiment, light field demand analysis is performed on the lighting demand parameters according to the ambient lighting data and the user interaction input to generate light field control information, including: Ambient lighting data is processed for feature extraction. The visible light band radiance data stream is fed into the illuminance analysis unit, which calculates the luminous flux per unit area according to photometric measurement standards. The analysis unit creates a two-dimensional illuminance distribution map with a 0.1-meter grid resolution. Each cell in the map records the lux value and the standard deviation of fluctuation within one second. Adjacent grid cells are differentiated using spatial gradients to generate the X / Y axial illuminance gradient vector.

[0080] The color temperature data processing channel is simultaneously activated, analyzing the chromaticity coordinate values ​​in the raw data and constructing a color temperature distribution heatmap using a quadratic surface fitting method. After Gaussian filtering to reduce noise on the heatmap, the maximum color temperature difference within the nine-square grid is calculated to generate the color temperature gradient amplitude. The output data is encapsulated in a composite data structure: the light intensity information unit contains the base illuminance value, fluctuation amplitude value, and spatial gradient field; the color temperature gradient data unit stores the main color temperature value, gradient vector, and gradient tolerance threshold. All output parameters are automatically flagged for ambient temperature drift correction.

[0081] User interaction data is processed through a multimodal parsing framework. The explicit demand parsing channel receives the text data stream output by speech recognition and uses syntactic dependency tree analysis to extract the three elements of the action entity, action type, and intensity modifier. This is then normalized and mapped using a pre-set rule base to generate the explicit demand information structure.

[0082] The implicit behavior analysis channel processes raw point cloud data. After spatial filtering and noise reduction, the point cloud is used to model joint motion trajectories and identify the sequence of upper limb posture angle changes. Infrared thermal imaging data is simultaneously input, and the temperature distribution map is used to reconstruct the gesture morphology using an edge detection algorithm. The behavior pattern analysis unit integrates posture angle and gesture trajectory features to output implicit behavior information data structures including the action direction vector, movement amplitude index, and limb steady-state characteristic values. The dual-channel results are matched by the timestamp alignment engine and then packaged for output, maintaining a time tolerance of 20 milliseconds. The output data packet is annotated with the action confidence score and device interference flag.

[0083] The compensation calculation unit combines the ambient lighting data with the explicit user input for illumination intensity. The compensation calculation unit then calculates the spatial difference between the two-dimensional illumination distribution map provided by the illumination intensity information and the target illumination value of the explicit user input. The compensation rule engine then processes this difference matrix to generate compensation parameters.

[0084] The main compensation channel applies linear compensation to the difference range, while the gradient compensation channel initiates exponential compensation for differences exceeding the threshold. The device attenuation adapter component simultaneously loads the luminaire's light attenuation characteristic curve and performs reverse calibration on the compensation value. Spatial processing identifies partitions based on the scope of explicit requirements: independent compensation for core areas and average compensation for peripheral areas. The output basic light compensation parameter structure includes the compensated illuminance value, partition identifier, maximum gradient value, and device compatibility flag. This structure is encrypted and verified for anti-interference packaging.

[0085] Implicit behavioral information and color temperature gradient data are fed into the color temperature calculation engine. The motion parameter mapping unit analyzes the action direction vectors in the implicit behavior and locates the corresponding scope in the color temperature gradient space. The adjustment coefficient generator performs a two-stage process: when the motion amplitude index is ≤0.3, the baseline color temperature is maintained; when it is >0.3, an offset is calculated based on a 300K per unit amplitude. When the limb steady-state characteristic value is >0.8, a smoothing coefficient of 0.9 is applied; when it is ≤0.8, a step coefficient of 0.5 is used.

[0086] The spatial mapping module convolves the calculated results with the color temperature gradient data, outputting a color temperature adjustment coefficient set containing an offset, an effective time window, and a gradient indicator. The coefficient set is transmitted with a dynamic priority tag, activating a degradation strategy when system load exceeds the limit.

[0087] The light field synthesis system inputs illumination requirements, basic illumination compensation parameters, and color temperature adjustment coefficients. The spatiotemporal alignment unit performs three-dimensional matching: in the spatial dimension, a regional overlay model is established using a coordinate mapping engine. In the temporal dimension, the transition curve for the required parameters and the adjustment coefficient time window are calibrated. The synthesis calculation is performed in four steps: the basic compensation parameter values ​​are superimposed on the required illumination baseline value; the color temperature adjustment coefficient offset is injected into the required color temperature value; the adjustment coefficient gradient flag resets the required gradient logic; and the device compatibility flag drives the power allocation strategy.

[0088] The command stream for generating light field control information includes a regional lighting control block, which is divided into a four-channel data structure: an illumination PWM duty cycle channel, a color temperature and current ratio channel, a gradient time parameter channel, and a fault rollback code channel. The command stream is encapsulated as a timestamp sequence frame, using a transmission verification mechanism. A regional topology check code is embedded in each frame to prevent spatial out-of-bounds.

[0089] This embodiment achieves intelligent and precise light field control through the coordinated processing of ambient lighting data and user interaction input. Multi-dimensional feature extraction of ambient lighting data, combined with high-precision grid analysis, dynamically generates a two-dimensional illuminance distribution map and color temperature gradient field, providing refined data support for lighting adjustment. Through dual-channel analysis of explicit and implicit demands, it accurately captures explicit demands from user voice commands while intelligently identifying latent demands through gesture trajectory modeling, enhancing the naturalness and accuracy of human-computer interaction. The compensation calculation unit utilizes a partitioned, differentiated processing strategy, combined with device attenuation characteristic calibration, to ensure the reliability and adaptability of lighting compensation. Color temperature adjustment achieves dynamic and smooth transitions through the convolution of motion parameters and gradient data, avoiding the discomfort caused by sudden color temperature changes. The resulting light field control command stream integrates multi-channel parameters, and through spatiotemporal alignment and verification mechanisms, ensures the precise transmission and execution of control signals, significantly improving the comfort and energy efficiency of the lighting environment.

[0090] In one embodiment, a basic compensation calculation is performed based on the light intensity information and the explicit demand information to obtain basic light compensation parameters, including: The light intensity information is segmented. The preset illumination interval division criteria define three core ranges: a low-light zone corresponds to illumination values ​​below a certain lower threshold, a transition zone lies in the middle illumination range, and a high-light zone corresponds to illumination values ​​above a certain upper threshold. The segmentation operation implements a spatial grid traversal algorithm, scanning and processing the two-dimensional illumination distribution map in the light intensity information element by element.

[0091] The scanning process activates a dynamic boundary optimization mechanism. When the illumination difference between adjacent grid cells exceeds a set gradient threshold, smoothing filtering is initiated to eliminate abrupt boundary changes caused by measurement noise. The segmentation results are generated into independent data sets: the highlight area data set records the grid coordinates, average illumination value, and fluctuation characteristics of the high-illuminance area; the low-illuminance area data set stores the coordinate sequence, minimum illumination value, and gradient change vector of the low-illuminance area. Data encapsulation uses a binary bitmap to mark spatial distribution features, and a segmentation quality report is simultaneously output, noting the proportion of unprocessed transition areas and boundary smoothness indicators.

[0092] Explicit demand information and highlight area data are input into the regional balance calculation unit. The demand parsing module extracts the target illumination value and range parameters from the explicit demand and performs a position matching operation with the spatial coordinate set of the highlight area data. The balance calculation is a two-stage process: the primary balance stage calculates the absolute difference between the current highlight area average illumination and the target illumination. If the difference is within the allowable fluctuation range, the status quo is maintained. If it exceeds the threshold, a linear attenuation adjustment coefficient is generated.

[0093] The advanced optimization phase analyzes illumination fluctuation data. When the fluctuation exceeds a set tolerance, a dynamic balancing algorithm is activated. The algorithm extracts the gradient change vector of the highlight area and generates a compensation vector field in the opposite direction of the gradient. The compensation range is defined by the required radius parameter. The balancing parameters are output as a multidimensional data structure, including the regional balancing adjustment coefficient, the maximum allowable attenuation value, the coordinates of the spatial scope boundary, and a device compatibility flag. The data structure is compressed and encoded, and then embedded with a timestamp and regional topology verification information.

[0094] Explicit demand information and low-light area data are input into the compensation calculation unit to perform illumination enhancement processing. The demand analysis module extracts the target illumination value and range parameters from the explicit demand and performs spatial position matching with the coordinate sequence of the low-light area data. The compensation calculation implements a two-layer processing mechanism: the basic compensation stage calculates the absolute difference between the target illumination and the minimum illumination value in the low-light area. If the difference is within the set fluctuation range, a linear compensation base is generated. The gradient enhancement stage analyzes the gradient change vector in the low-light area and generates a compensation enhancement field along the positive gradient direction.

[0095] The spatial compensation strategy operates based on the scope parameters: independent enhancement compensation is applied to the core area, while the mean compensation algorithm is applied to the edge areas. The device attenuation adapter component loads the luminaire's light attenuation characteristic curve and performs a reverse calibration on the compensation value to eliminate the impact of device performance attenuation. The output low-light compensation parameter structure contains the compensated illuminance value, spatial scope identifier, maximum enhancement gradient value, and device compatibility flag. The structure is encrypted and encapsulated, and then the regional topology check code is appended.

[0096] Highlight balance parameters and lowlight compensation parameters are fed into the compensation integration engine for collaborative processing. The spatial mapping unit parses the spatial scope identifiers of the two parameters and initiates a mean fusion algorithm for overlapping areas, while maintaining the original parameter values ​​in non-overlapping areas. The temporal alignment module calibrates the effective time windows of the two parameters, implementing linear interpolation synchronization for segments with temporal differences.

[0097] Integrated three-dimensional optimization is performed: in the illumination dimension, a weighted average algorithm is used to balance the need for attenuation in high-light areas and enhancement in low-light areas; in the spatial dimension, compensation gradients are redistributed based on regional weight coefficients; and in the device dimension, power allocation strategies are arbitrated using compatible tags. The output basic lighting compensation parameters are a composite data structure consisting of a global compensation illumination distribution map, a spatial gradient control vector, a temporal transition parameter set, and a device constraint identifier group.

[0098] This embodiment significantly improves the boundary distortion problem caused by traditional fixed partitioning by processing areas with sudden changes in light intensity through a dynamic boundary optimization mechanism, combined with a gradient threshold-driven smoothing filter algorithm. The three-dimensional illuminance distribution map generates independent data sets (highlight area / lowlight area / transition area) through spatial grid traversal, enabling lossless analysis of complex light field structures and providing a basis for millimeter-level spatial accuracy for subsequent compensation. The regional balance calculation unit adopts a hierarchical processing strategy: primary balance maintains the target illuminance within an allowable fluctuation range, and advanced optimization dynamically suppresses light overexposure in strong gradient areas through a compensation vector field. In the gradient enhancement stage, a compensation field is generated along the positive direction of illuminance change, and the compensation value is calibrated by the equipment light decay curve simultaneously to eliminate the negative impact of lamp performance degradation on the compensation accuracy of low-light areas.

[0099] In one embodiment, obtaining power load data, constructing instructions for light field control information, and generating lighting control instructions include: The target scenario's power supply network is connected to an intelligent monitoring terminal device. The terminal is equipped with a voltage transformer, a current transformer, and a power metering chip to form a data acquisition unit. Upon startup, the monitoring device executes a device addressing protocol to establish a topological mapping relationship between the lighting cluster and the power supply circuit. The data acquisition unit captures three-phase voltage waveforms, current RMS values, and instantaneous power parameters at a sampling rate of 60 times per second. The raw electrical signals undergo A / D conversion to generate a digital data stream.

[0100] Voltage data is converted to RMS values ​​using a true RMS algorithm. Current data is subjected to harmonic analysis to extract the fundamental component. Active and reactive power are calculated using discrete integration. The load status assessment engine activates a dynamic threshold adjustment mechanism, setting warning thresholds based on historical load curves and triggering safety flags for out-of-limit parameters. Output power load data is encapsulated as a structured message containing a loop identifier, voltage RMS value, current fundamental value, active power, power factor, and a safety status code. Millisecond-level timestamps and device topology checksums are appended during transmission.

[0101] Prioritize power load data and light field control information. The load status analysis unit extracts the active power value and safety status code from the power data and implements hierarchical marking operations for lighting circuits: circuits with normal safety status codes and power margins exceeding 15% are marked as high-priority circuits, while circuits with warning safety status codes or power margins below 5% are marked as low-priority circuits.

[0102] The light field parameter parsing module parses the regional lighting control block data structure within the light field control information, extracting the required illumination value, color temperature adjustment coefficient, and gradient time parameters. The priority calculator implements a three-dimensional weighting process: the lighting quality weight calculates the visual comfort coefficient based on the required illumination value, with a value range of 0 to 1.0; the energy conservation priority assigns an energy conservation factor based on the power margin ratio, with a coefficient range of 0.5 to 1.5; and the safety weight is converted into a safety factor using the safety status code, with a gradation of 0.7, 1.0, or 1.3.

[0103] The weight matrix is ​​decomposed using eigenvalues ​​to generate a comprehensive priority score. Lighting commands with a score exceeding 0.8 are marked as real-time execution commands, commands between 0.5 and 0.8 are marked as delayed execution commands, and commands below 0.5 enter the standby sequence. The output light field control sequence uses a time-slice rotation structure. The sequence unit stores the command type identifier, execution time window parameters, power budget value, and spatial topology association identifier. A load fluctuation tolerance parameter is injected into the sequence encapsulation, and the sequence reconstruction protocol is activated when the monitored grid frequency deviation exceeds 0.5 Hz.

[0104] Perform time-matching and power coordination for the light field control sequence. Load the historical grid load database and divide the time periods into three core periods: peak, flat, and off-peak based on the characteristics of the daily load curve. A differentiated strategy is implemented for matching calculations: during peak periods, the command power compression mechanism is activated to maintain base power for real-time commands, the power of delayed commands is limited to 70% of the rated value, and standby commands are forced to sleep. During flat periods, the power budget for commands is allocated based on the priority score, with a 10% power increase allowed for every 0.1% increase in the score. During off-peak periods, the power limit is lifted, and the comfort enhancement mode is enabled, allowing real-time commands to over-allocate power by 20%.

[0105] Real-time load fluctuations are monitored simultaneously, triggering a dynamic reallocation protocol when the load factor suddenly changes by more than 15%. The output time-segment control parameter structure includes a time segment identifier, a power allocation ratio matrix, a command execution window correction value, and a load mutation response flag. This structure is compressed and encoded before being injected with the grid frequency tracking coefficient.

[0106] Device drivers convert time-segment control parameters and light field control information. The encoding core implements four-channel conversion: the illumination channel converts the illumination demand value into a pulse-width modulation duty cycle code, scaling the duty cycle range based on the time-segment power ratio; the color temperature channel maps the color temperature adjustment coefficient into a multi-channel mixed light ratio code, activating the wide-band color temperature enhancement algorithm during off-peak periods; the time channel re-encodes the gradient time parameter into a sequence of step values, compressing the total duration during peak periods by 30%; and the fault-tolerant channel converts fault rollback codes into device status monitoring instructions.

[0107] The encoding process performs a spatial topology check and binds the spatial association identifiers in the control sequence to device addresses. The output of the initial control instruction set uses a layered encapsulation structure: the physical layer is a power line carrier communication frame, and the data layer contains the device address code, control instruction code, check sequence, and retransmission protocol identifier. When the instruction set is generated, a time period power margin warning flag is automatically added, triggering an instruction degradation flag when the real-time power margin falls below 5%.

[0108] The initial control instruction set is security-verified. The conflict detection engine performs a three-dimensional scan: The power dimension calculates the total power requirement of the instruction set and compares it to the real-time circuit capacity to detect over-limits; the device dimension verifies the compatibility of instruction parameters with the lighting fixture's driving capabilities; and detects whether the sudden gradient of adjacent device instructions exceeds the threshold. When a conflict is detected, a dynamic adjustment protocol is initiated: in power over-limit scenarios, standby instructions are shut down in descending order of priority, followed by power compression of delay instructions; in device over-limit scenarios, instructions outside the adjustable range are threshold-truncated; in gradient over-limit scenarios, transitional instruction frames are inserted to smooth the change curve. The adjusted instruction set is re-encoded to generate the final lighting control instructions, with the instruction structure adding a conflict resolution identifier and an execution receipt request bit.

[0109] This embodiment uses intelligent monitoring terminals to capture the three-phase voltage, current, and power parameters of the power grid in real time, and dynamically adjusts warning thresholds based on historical load curves. Time-of-day power matching technology implements differentiated strategies based on peak, flat, and off-peak load characteristics: non-critical command power is compressed during peak periods, and comfort enhancement mode is enabled during off-peak periods. Light field control parameters and power data are collaboratively processed through three-dimensional weight distribution: the visual comfort coefficient ensures lighting quality, the dynamic energy-saving factor optimizes power distribution, and the safety factor provides redundant protection. A priority scoring mechanism divides commands into three levels: real-time execution, delayed execution, and standby commands, enabling precise resource scheduling with millisecond-level response in key areas and flexible regulation in marginal areas.

[0110] Reference Figure 2 As shown, the present application also provides a lighting control system based on AI model analysis, which is applied to any of the above-mentioned lighting control methods based on AI model analysis, including: Recognition module: The recognition module is used to obtain the ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; The parsing module is used to analyze the lighting requirements of the semantic description information through the pre-trained lighting driving model to obtain the lighting requirement parameters; A processing module is used to perform light field demand analysis on the lighting demand parameters according to the ambient lighting data and user interaction input, and generate light field control information; The building module is used to obtain power load data, construct instructions for light field control information, and generate lighting control instructions.

[0111] The present application provides a lighting control system based on AI model analysis. By converting user interaction input into quantifiable lighting demand parameters, it solves the problem of lighting control deviation caused by insufficient semantic understanding in traditional systems and improves the satisfaction of users' personalized needs. By combining ambient lighting data and user interaction input to analyze light field demand, dynamic adaptive control of the light environment is achieved, overcoming the limitation of existing methods that rely only on static compensation and cannot adapt to changes in complex scenes, and improving the comfort and consistency of the light environment. By introducing power load data to construct instructions for light field control information, it ensures that the lighting control strategy meets user needs while taking into account the load balance of the power grid, avoiding energy waste and system overload risks, and improving the overall energy efficiency ratio. Based on the pre-trained lighting drive model, it can automatically adjust lighting parameters according to different scenarios and user needs, enhance adaptability and intelligence, and enable it to be widely used in various scenarios such as home, office, and business.

[0112] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0113] The scheme of the present application has been described in detail above with reference to the accompanying drawings. In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. Those skilled in the art should also be aware that the actions and modules involved in the description are not necessarily required for this application. In addition, it is understood that the steps in the method of the embodiment of the present application can be adjusted in sequence, merged and deleted according to actual needs, and the modules in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0114] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0115] Alternatively, the present application can also be implemented as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) on which executable code (or computer program, or computer instruction code) is stored. When the executable code (or computer program, or computer instruction code) is executed by a processor of an electronic device (or electronic device, server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0116] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the application herein may be implemented as electronic hardware, computer software, or combinations of both.

[0117] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems and methods according to multiple embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a part of a module, program segment or code, and the part of the module, program segment or code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0118] The embodiments of the present application have been described above. The above description is illustrative and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to the technology in the market, or to enable other persons skilled in the art to understand the embodiments disclosed herein.

Claims

1. A lighting control method based on AI model analysis, characterized in that: include: Obtain the ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; Performing a lighting demand analysis on the semantic description information through a pre-trained lighting driving model to obtain lighting demand parameters; Performing a light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input to generate light field control information; The power load data is obtained, and the light field control information is constructed into instructions to generate lighting control instructions.

2. The lighting control method based on AI model analysis according to claim 1, characterized in that: The step of obtaining the ambient lighting data of the target scene and the user interaction input, performing semantic requirement analysis, and obtaining semantic description information includes: Collecting spectral distribution data of the target scene through a preset light sensor to obtain an original light sampling signal; Performing frequency domain filtering and noise suppression on the original light sampling signal to generate ambient light data; Receive user voice input of the target scene through a preset voice interaction device, perform voice recognition, and obtain text interaction data; The environmental lighting data and the text interaction data are combined through semantic coding to output the semantic description information.

3. The lighting control method based on AI model analysis according to claim 1, characterized in that: The lighting requirement analysis of the semantic description information is performed using the pre-trained lighting driving model to obtain lighting requirement parameters, including: Inputting the semantic description information into the light driving model, and performing context association analysis on the semantic description information through the semantic understanding layer of the light driving model to obtain semantic features; The intention inference layer calculates user requirements based on the semantic features to obtain initial lighting requirement information; Performing color temperature compensation and brightness adjustment on the initial lighting requirement information by combining the ambient lighting data and the user interactive input through a parameter optimization layer to obtain color temperature requirement parameters, brightness requirement parameters and dynamic change parameters; The color temperature requirement parameter, brightness requirement parameter and dynamic change parameter are integrated into multiple dimensions through the output fusion layer to output the lighting requirement parameter.

4. The lighting control method based on AI model analysis according to claim 3 is characterized in that: The parameter optimization layer combines the ambient light data and the user interaction input to perform color temperature compensation and brightness adjustment on the initial light requirement information to obtain color temperature requirement parameters, brightness requirement parameters and dynamic change parameters, including: The optical parameter decomposition unit of the parameter optimization layer analyzes the initial lighting requirement information, extracts spectral features, and generates basic color temperature parameters and basic brightness parameters; Calculating the color temperature requirement of the ambient light data for the basic color temperature parameter and outputting the corresponding color temperature requirement parameter; Obtain historical interaction information, perform preference analysis based on the user interaction input, and output user preference information; Performing a brightness demand analysis on the basic brightness parameter according to the user preference information to obtain a brightness demand parameter; The color temperature requirement parameter is dynamically matched with the brightness requirement parameter to generate a dynamically changing parameter.

5. The lighting control method based on AI model analysis according to claim 4 is characterized in that: The step of calculating the color temperature requirement of the basic color temperature parameter by the ambient light data and outputting the corresponding color temperature requirement parameter includes: Analyzing spectral components from the ambient light data to extract spectral distribution data; Calculating a color temperature offset for the basic color temperature parameter according to the spectral distribution data to obtain a color temperature compensation value; Performing color temperature superposition calculation on the color temperature compensation value and the basic color temperature parameter to generate a corrected color temperature parameter; The color temperature range constraint is performed on the corrected color temperature parameter to obtain the color temperature requirement parameter.

6. The lighting control method based on AI model analysis according to claim 1, characterized in that: The performing light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input to generate light field control information includes: Extracting light intensity information and color temperature gradient data from the ambient light data; Analyze the intention of the user's interactive input and identify explicit demand information and implicit behavior information; Performing basic compensation calculation according to the light intensity information and the explicit demand information to obtain basic light compensation parameters; Adjusting the color temperature gradient data based on the implicit behavior information to generate a color temperature adjustment coefficient; A light field synthesis calculation is performed on the basic light compensation parameter and the color temperature adjustment coefficient according to the light requirement parameter to obtain the light field control information.

7. The lighting control method based on AI model analysis according to claim 6, characterized in that: The performing basic compensation calculation according to the light intensity information and the explicit demand information to obtain basic light compensation parameters includes: Segmenting the illumination intensity information into intervals based on a preset illumination interval threshold to obtain high light area data and low light area data; For the highlight area data, combining the explicit demand information, performing dynamic area brightness balancing, and outputting highlight area balance parameters; For the low-light area data, performing minimum illumination compensation according to the explicit requirement information to generate low-light area compensation parameters; The highlight area balance parameter and the light area compensation parameter are integrated for basic compensation to obtain the basic illumination compensation parameter.

8. The lighting control method based on AI model analysis according to claim 1, characterized in that: The acquiring of power load data, constructing instructions for the light field control information, and generating lighting control instructions include: Identifying the grid load of the target scenario and monitoring the load in real time to obtain the power load data; assigning priorities to the light field control information based on the power load data to form a light field control sequence; Performing time-division power matching on the light field control sequence to obtain time-division control parameters; Using the time-division control parameters to perform instruction encoding on the light field control information to generate an initial control instruction set; The load conflict of the initial control instruction set is detected and dynamically adjusted, and the lighting control instruction is output.

9. A lighting control system based on AI model analysis, characterized in that: The lighting control method based on AI model analysis applied to any one of claims 1 to 8 above comprises: A recognition module is used to obtain ambient lighting data of the target scene and user interaction input, perform semantic demand analysis, and obtain semantic description information; An analysis module is used to analyze the lighting requirements of the semantic description information using a pre-trained lighting driving model to obtain lighting requirement parameters; a processing module, configured to perform a light field demand analysis on the lighting demand parameters according to the ambient lighting data and the user interaction input, and generate light field control information; A construction module is used to obtain power load data, perform instruction construction on the light field control information, and generate lighting control instructions.

Images