Agent intelligent light string accurate tuner control algorithm based on voice

Abstract

The invention discloses a voice-based Agent intelligent light string accurate tuner control algorithm, and particularly relates to the technical field of intelligent light string control, and the algorithm comprises the steps: obtaining a user voice instruction through a voice collection unit, and collecting the multi-dimensional perception data of a current environment through an environment perception unit, the multi-dimensional sensing data at least comprises environment illumination intensity, an environment noise decibel value and a human body infrared existence signal. According to the invention, the Agent intelligent decision-making module fuses a voice instruction and multi-dimensional environment perception data, achieves the deep scene adaptation of light control, extracts user demands based on context semantic analysis, associates historical instructions to construct a user preference model, and dynamically optimizes the voice recognition sensitivity and the light color rendering effect according to the environment change. Light control is upgraded to active adaptation, a full-process precise control closed loop of dynamic grouping, preview confirmation and fault self-repairing is constructed, and the problems that traditional light string control is fixed in grouping, blind in execution and prone to fault diffusion are solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent light string control technology, and more specifically, to a voice-based agent-based intelligent light string precise tuner control algorithm. Background Technology

[0002] Voice-controlled string lighting technology has been widely used in various fields such as commercial space decoration, outdoor landscape lighting, industrial workshop auxiliary lighting, and exhibition atmosphere creation. With the development of smart string lighting, the demand for voice-controlled string lighting continues to grow.

[0003] Existing voice-controlled light strings rely solely on voice recognition for linear command-to-execution responses. They cannot simultaneously utilize ambient light, noise, and infrared sensor data to optimize control. Voice recognition is susceptible to environmental interference and cannot parse contextual semantics or correlate historical commands, resulting in a disconnect between control effectiveness and requirements. Furthermore, they lack a binding mechanism between the physical location of LED beads and their hardware addresses, hindering flexible control range allocation. Colors are fixedly mapped to RGB / GRB values ​​without considering ambient color temperature calibration. There is no preview-to-confirmation process; commands are executed directly, leading to high adjustment costs when errors occur. Lamp bead current and voltage data are not collected in real-time, making it difficult to detect individual lamp bead anomalies promptly, allowing faults to easily spread. There is no automatic repair strategy after a fault, requiring manual troubleshooting, increasing maintenance costs and potentially causing service interruptions. Colors and modes are mostly preset, lacking user customization, multi-user recognition, and permission adaptation. The simplistic architecture fails to meet the needs of complex scenarios and real-time requirements. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, this invention provides a voice-based Agent-based intelligent light string precise tuner control algorithm to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a voice-based agent-based intelligent light string precise tuner control algorithm, comprising the following steps: S1: Acquire user voice commands through the voice acquisition unit, and simultaneously acquire multi-dimensional perception data of the current environment through the environmental perception unit. The multi-dimensional perception data includes at least ambient light intensity, ambient noise decibel value, and human infrared presence signal. S2: The voice command is converted into text data through the voice recognition module, and the text data and the multi-dimensional perception data are sent to the Agent intelligent decision module simultaneously. The Agent intelligent decision module accesses the preset historical command database, performs contextual semantic analysis on the text data, extracts the color sequence, lighting mode, LED control range and brightness level in the user's needs, and generates scene adaptation parameters in combination with the multi-dimensional perception data. S3: The Agent intelligent decision-making module accesses the preset color mapping table, converts the color descriptors in the color sequence into 32-bit initial color values ​​in GRB or RGB format, and performs color calibration on the initial color values ​​according to the ambient light intensity to generate a target color array to compensate for the influence of ambient light and make the target color consistent with the user's expectations in human visual perception. S4: The Agent intelligent decision-making module dynamically divides the LED beads into N non-overlapping LED bead groups according to the LED bead control range and the preset spatial position coordinates of the LED string, and determines the color arrangement of each LED bead group according to the lighting mode and scene adaptation parameters. The lighting mode includes sequential lighting, alternating lighting, random lighting, cyclic lighting and scene-related custom combination modes. S5: The Agent intelligent decision-making module generates a structured control instruction that includes color parameters, mode parameters, LED group division parameters, brightness level parameters and feedback instructions, and sends the structured control instruction to the lighting control terminal through the MCP protocol; S6: After receiving the structured control command, the lighting control terminal first drives the corresponding LED group to execute the color arrangement at 50% brightness and performs a preview for 3 to 5 seconds. At the same time, it generates and plays the execution preview voice through the voice synthesis module. S7: If no cancellation or modification instruction is received from the user during the preview, the lighting control terminal will drive the corresponding LED group to execute the complete lighting effect. If a cancellation instruction is received from the user, the current control process will be terminated. If a modification instruction is received, the process will return to S2 to re-perform semantic parsing. S8: During the execution process, the lighting control terminal collects the current and voltage data of each LED in real time and feeds the collected data back to the Agent intelligent decision-making module. The Agent intelligent decision-making module updates the historical instruction database based on the collected data to optimize the accuracy of subsequent instruction parsing.

[0006] Preferably, the historical instruction database stores the user's voice instructions and corresponding execution parameters within the past 72 hours, and the context semantic parsing includes: identifying scene keywords in the text data, associating the color sequence, brightness level and lighting mode corresponding to the same scene keyword in the historical instruction database, and generating a parameter recommendation list adapted to the current scene; The parameter recommendation list contains multiple sets of candidate parameters, preferably 1 to 3 sets.

[0007] Preferably, the Agent intelligent decision-making module also dynamically adjusts the sensitivity of the speech recognition module according to the ambient noise decibel value: when the ambient noise decibel value reaches or exceeds 60 decibels, the wake-up threshold of speech recognition is increased by 20% to 30%, and the beamforming technology of the microphone array is enabled to enhance user voice acquisition. When the ambient noise level is below 60 decibels, the default wake-up threshold is restored.

[0008] Preferably, the preset spatial coordinates of the light string are obtained in the following way: when the light string is installed, the user marks the physical position of each light bead or light bead segment by voice command, and the Agent intelligent decision module binds the voice-marked physical position with the hardware address of the light string to generate a spatial coordinate library.

[0009] Preferably, the custom combination mode associated with the scene is implemented in the following way: After parsing the scene keywords in the text data, the Agent intelligent decision-making module configures differentiated working parameters for the divided multiple LED groups based on the preset lighting strategy corresponding to the scene keywords. The working parameters include at least color, brightness and dynamic mode. The preset lighting strategy includes configuring at least one LED group with different brightness levels and dynamic effects than other LED groups to achieve a collaborative lighting scene that matches the scene keywords.

[0010] Preferably, the Agent intelligent decision-making module also constructs a user preference model based on the user's selection record of the parameter recommendation list: if the user selects a certain type of color sequence 3 times in a row, the priority of that type of color sequence in the parameter recommendation list will be increased in subsequent analysis of the same scenario.

[0011] Preferably, the feedback instruction further includes a fault warning sub-instruction: when the current or voltage data of the lamp beads collected by the lighting control terminal exceeds the normal parameter range, the Agent intelligent decision module generates a fault warning voice and automatically adjusts the lamp bead group division parameters, skips the faulty lamp beads, and reissues the control instruction; The normal parameter range is ±10% of the rated current and ±5% of the rated voltage.

[0012] Preferably, the method further includes a multi-user voiceprint recognition step: the Agent intelligent decision-making module pre-stores the voiceprint features of multiple authorized users. In S2, the voiceprint features of the current user are first extracted by the speech recognition module and matched with the pre-stored voiceprint features. If the match is successful, the personalized parameter library corresponding to the user is called. If the match fails, only basic control permissions are granted.

[0013] Preferably, the Agent intelligent decision-making module adopts an edge computing and cloud collaborative architecture: basic semantic parsing, historical command retrieval, environmental data processing, and LED group division are completed at the edge terminal. Only when multiple LED strings are parsed for collaborative control or complex scene combinations are the commands sent to the cloud-based large language model for deep parsing. The cloud parsing results are then returned to the edge terminal for execution.

[0014] Preferably, the color map supports user-defined extensions: users input custom color names and corresponding RGB or GRB values ​​via voice commands, and the Agent intelligent decision-making module associates and stores the custom color names with their corresponding values ​​in the color map, which can then be directly accessed using the custom color name.

[0015] The technical effects and advantages of this invention are as follows: By integrating voice commands with multi-dimensional environmental perception data through the Agent intelligent decision-making module, deep scenario-based adaptation of lighting control is achieved. It can not only extract user needs based on contextual semantic parsing, but also build user preference models by associating historical commands. It can also dynamically optimize voice recognition sensitivity and lighting color rendering effect according to environmental changes, upgrading lighting control from passive response to active adaptation. It can automatically match user habits and scenario needs, significantly improving the interaction stability and control fit in complex environments, and avoiding the control failure problem caused by environmental interference or misunderstanding of needs in traditional systems. By constructing a precise closed-loop control system with dynamic grouping, preview confirmation, and fault self-repair, the system solves the problems of fixed grouping, blind execution, and easy fault propagation in traditional light string control. In terms of control precision, voice annotation binds the physical location and hardware address of the LED beads, allowing for flexible division of non-overlapping LED bead groups to achieve differentiated lighting effects. In terms of interactive reliability, the preview mechanism allows users to confirm the effect before full execution, reducing misoperation. In terms of fault handling, the system can monitor the current and voltage of the LED beads in real time, automatically generate warnings when anomalies are detected, and skip the faulty LED bead to readjust the control strategy. This closed-loop design not only ensures the accurate presentation of lighting effects but also avoids the impact of single-point failures on overall control, significantly improving the system's control precision and operational reliability. By adopting an edge computing and cloud-based collaborative architecture, the edge can quickly process basic control needs of a single light string to ensure low-latency response, while the cloud supports in-depth analysis of multi-light string collaboration or complex scenarios, balancing response speed and complex control capabilities. On the other hand, it supports custom expansion of color mapping tables and multi-user voiceprint recognition. Users can create their own colors via voice and call them directly. The system can also distinguish user identities through voiceprints and match different users' personalized parameter libraries and permissions. This design not only meets users' personalized customization needs for lighting effects, but also adapts to the differentiation of permissions and preferences in multi-user shared scenarios, giving the system stronger scalability and humanized service capabilities, breaking through the limitations of traditional light string control functions that are single and difficult to adapt to diverse needs. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the algorithm flow structure of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: This embodiment provides a voice-based Agent-based intelligent light string precision tuner control algorithm. Its core lies in driving an Agent with intelligent decision-making capabilities through voice interaction and environmental perception to achieve refined, scenario-based, and adaptive control of the light string.

[0019] The system hardware mainly includes: Voice acquisition unit: An array-type voice acquisition unit containing multiple microphones is adopted, preferably a 6-microphone array arranged in a ring, to achieve sound source localization and noise reduction functions.

[0020] Environmental sensing unit: includes a digital ambient light sensor for collecting ambient light intensity, with a measurement range covering 0 to 65535 Lux; a digital microphone for collecting ambient noise decibel values; and an infrared pyroelectric sensor for detecting the presence of human signals.

[0021] Agent intelligent decision-making module: Implemented by an embedded processor, which has the computing power to support running lightweight artificial intelligence models.

[0022] Lighting control terminal: Includes main control MCU and LED driver circuit, used to drive independently addressable smart LED beads, such as WS2812B series or compatible types of light strings.

[0023] The execution steps of the control algorithm are as follows: S1: Multimodal data acquisition As the user speaks a voice command, the environmental perception unit simultaneously collects multi-dimensional perception data of the current environment, including ambient light intensity, ambient noise levels (decibels), and the presence of human infrared signals. This data is sampled at specific intervals, such as per second, and transient interference can be removed using a smoothing filtering algorithm.

[0024] S2: Speech Recognition and Intelligent Semantic Parsing Before the speech recognition module converts the text data, the system first performs a multi-user voiceprint recognition step: the Agent intelligent decision-making module pre-stores voiceprint feature models of multiple authorized users. The system extracts voiceprint features from the current voice command and matches them with the pre-stored voiceprint feature models. If the match is successful, the system identifies the current user and calls the user's corresponding personalized parameter library, which includes their commonly used color preferences, brightness levels, mode preferences, etc., for subsequent analysis. If the match fails or the user is not identified, the system only grants basic control permissions, such as switching on / off and fixed color temperature adjustment, and restricts the use of custom color sequences and dynamic modes. The speech recognition module converts speech commands into text data, which is then sent synchronously to the Agent intelligent decision-making module along with the environmental perception data. The Agent intelligent decision-making module accesses a preset historical command database, such as an embedded database, which stores the user's voice commands and corresponding execution parameters within the recent period, such as 72 hours. The module performs contextual semantic analysis on the text data, identifies scene keywords, and associates the color sequences, brightness levels, and lighting modes corresponding to the same scene keywords in the historical command database to generate a parameter recommendation list adapted to the current scene. The parameter recommendation list contains multiple sets of candidate parameters, preferably 1 to 3 sets. The Agent intelligent decision-making module also constructs a user preference model based on the user's selection record of the parameter recommendation list: if the user selects a certain type of color sequence, such as "cool color tone", three times in a row, the priority of that type of color sequence in the parameter recommendation list will be increased in subsequent analysis of the same scene; The Agent intelligent decision-making module can generate a parameter recommendation list using a weighted scoring algorithm. The algorithm comprehensively considers multiple factors such as historical usage frequency and environmental data matching degree. The weight of each factor can be adjusted according to actual application. In a preferred embodiment, 1 to 3 sets of candidate parameters can be generated for final decision-making. The weighted scoring algorithm calculates a comprehensive score for each candidate parameter scheme. The score can be calculated as a weighted sum of the scores of each factor and their corresponding weights. Finally, the scheme with the highest score is selected as the recommendation result.

[0025] S3: Color Mapping and Ambient Adaptive Color Calibration The Agent intelligent decision-making module accesses a preset color map table and converts color descriptors into 32-bit initial color values ​​in GRB or RGB format. The color map table supports user-defined extensions: users can input custom color names and corresponding RGB or GRB values ​​via voice commands, such as "custom color A, RGB: 255, 100, 50". The Agent intelligent decision-making module will associate the custom color name with the corresponding value and store it in the color map table, which can then be directly accessed using the custom color name. To further improve color rendering accuracy, the system performs color rendering calibration on the initial color value based on the color temperature value corresponding to the ambient light intensity. The calibration process can be implemented through a mapping function, the purpose of which is to compensate for the influence of ambient light and control the color rendering deviation of the target color under the current ambient light within a range that is imperceptible to human vision. In a preferred embodiment, the calibration can control the color rendering deviation to within 5%.

[0026] S4: Dynamic LED grouping and mode decision The Agent intelligent decision-making module dynamically divides the LED string into N non-overlapping LED groups based on the LED control range and the preset spatial coordinates of the LED string. N is a positive integer. The preset spatial coordinates of the LED string are obtained in the following way: When the LED string is installed, the user marks the physical position of each LED or LED segment with voice commands, such as "LEDs 1-10 on the left side of the living room". The Agent intelligent decision-making module binds the voice-marked physical position with the hardware address of the LED string to generate a spatial coordinate library. Subsequently, the Agent determines the color arrangement for each LED group based on the parsed lighting mode and scene adaptation parameters. The custom combination mode associated with the scene is implemented in the following way: After parsing the scene keywords in the text data, the Agent intelligent decision module configures differentiated working parameters for the divided LED groups based on the preset lighting strategy corresponding to the scene keywords, including at least color, brightness and dynamic mode. The preset lighting strategy includes configuring at least one LED group with different brightness levels and dynamic effects than other LED groups to achieve a collaborative lighting scene that matches the scene keywords. The lighting modes include sequential lighting, alternating lighting, random lighting, cyclic lighting, and custom combination modes associated with the scene.

[0027] S5: Structured Instruction Generation and Issuance The Agent intelligent decision-making module generates structured control instructions that include color parameters, mode parameters, LED group division parameters, brightness level parameters, and feedback instructions. These instructions are then sent to the lighting control terminal via a communication protocol, such as a proprietary MCP protocol based on serial communication. The system also has environmental adaptive capabilities: the Agent intelligent decision-making module dynamically adjusts the sensitivity of the speech recognition module according to the ambient noise decibel value. For example, when the ambient noise decibel value reaches or exceeds 60 decibels, the wake-up threshold of speech recognition is increased by 20% to 30%, and the beamforming technology of the microphone array is enabled to enhance user voice acquisition; when the ambient noise decibel value is below 60 decibels, the default wake-up threshold is restored. The Agent intelligent decision-making module adopts an edge computing and cloud collaborative architecture: basic semantic parsing, historical command retrieval, environmental data processing, and LED group division are completed on the edge terminal and the lighting control terminal is completed locally. Only when multi-LED string collaborative control is parsed, such as "synchronous gradual change of LED strings throughout the house" or complex scene combinations, such as "movie viewing and curtain linkage", the command is sent to the cloud large language model for deep parsing. The cloud parsing result is returned to the edge terminal for execution.

[0028] S6: Preview and Confirmation Interaction After receiving the structured control command, the lighting control terminal first drives the corresponding LED group to execute the color arrangement at 50% brightness and performs a 3- to 5-second preview. At the same time, it generates and plays the execution preview voice through the speech synthesis module.

[0029] S7: Final Execution or Process Redirection If no cancellation or modification instruction is received from the user during the preview, the lighting control terminal drives the corresponding LED group to execute the complete lighting effect. If a cancellation instruction is received from the user, the current control process is terminated; if a modification instruction is received, the process returns to step S2 to re-perform semantic parsing.

[0030] S8: Real-time monitoring and feedback optimization During execution, the lighting control terminal collects the current and voltage data of each LED in real time and feeds the collected data back to the Agent intelligent decision-making module. The feedback instruction also includes a fault warning sub-instruction: when the LED current or voltage data collected by the lighting control terminal exceeds the normal parameter range, for example, ±10% of the rated current and ±5% of the rated voltage, the Agent intelligent decision-making module will generate a fault warning voice, such as "The current of LED X is abnormal" and automatically adjust the LED group division parameters, skip the faulty LED, and reissue the control instruction. Meanwhile, successful execution parameters are fed back to the historical instruction database. The Agent intelligent decision-making module updates the historical instruction database based on the collected data, which is used to optimize the parsing accuracy of subsequent instructions through machine learning algorithms. The method in this embodiment, by introducing a closed loop of multimodal perception, agent-based intelligent decision-making, preview confirmation, and feedback optimization, has the following advantages: It improves the robustness of voice interaction and control accuracy in complex environments, such as high-noise environments.

[0031] Environmental adaptive calibration ensures that the lighting display effect is consistent with user expectations.

[0032] Through preview and self-learning mechanisms, the user experience and system intelligence have been greatly improved.

[0033] The reliability of the system was enhanced through real-time monitoring and fault handling.

[0034] The present invention does not depend on any specific hardware model, and those skilled in the art can select appropriate existing devices or software modules to implement it based on the above functional description.

[0035] Finally, it should be noted that the accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A voice-based agent-based intelligent light string precise tuner control algorithm, characterized in that: Includes the following steps: S1: Acquire user voice commands through the voice acquisition unit, and simultaneously acquire multi-dimensional perception data of the current environment through the environmental perception unit. The multi-dimensional perception data includes at least ambient light intensity, ambient noise decibel value, and human infrared presence signal. S2: The voice command is converted into text data through the voice recognition module, and the text data and the multi-dimensional perception data are sent to the Agent intelligent decision module simultaneously. The Agent intelligent decision module accesses the preset historical command database, performs contextual semantic analysis on the text data, extracts the color sequence, lighting mode, LED control range and brightness level in the user's needs, and generates scene adaptation parameters in combination with the multi-dimensional perception data. S3: The Agent intelligent decision-making module accesses the preset color mapping table, converts the color descriptors in the color sequence into 32-bit initial color values ​​in GRB or RGB format, and performs color calibration on the initial color values ​​according to the ambient light intensity to generate a target color array to compensate for the influence of ambient light and make the target color consistent with the user's expectations in human visual perception. S4: The Agent intelligent decision-making module dynamically divides the LED beads into N non-overlapping LED bead groups according to the LED bead control range and the preset spatial position coordinates of the LED string, and determines the color arrangement of each LED bead group according to the lighting mode and scene adaptation parameters. The lighting mode includes sequential lighting, alternating lighting, random lighting, cyclic lighting and scene-related custom combination modes. S5: The Agent intelligent decision-making module generates a structured control instruction that includes color parameters, mode parameters, LED group division parameters, brightness level parameters and feedback instructions, and sends the structured control instruction to the lighting control terminal through the MCP protocol; S6: After receiving the structured control command, the lighting control terminal first drives the corresponding LED group to execute the color arrangement at 50% brightness and performs a preview for 3 to 5 seconds. At the same time, it generates and plays the execution preview voice through the voice synthesis module. S7: If no cancellation or modification instruction is received from the user during the preview, the lighting control terminal will drive the corresponding LED group to execute the complete lighting effect. If a cancellation instruction is received from the user, the current control process will be terminated. If a modification instruction is received, the process will return to S2 to re-perform semantic parsing. S8: During the execution process, the lighting control terminal collects the current and voltage data of each LED in real time and feeds the collected data back to the Agent intelligent decision-making module. The Agent intelligent decision-making module updates the historical instruction database based on the collected data to optimize the accuracy of subsequent instruction parsing.

2. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The historical instruction database stores the user's voice instructions and corresponding execution parameters within the past 72 hours. The contextual semantic parsing includes: identifying scene keywords in the text data, associating the color sequence, brightness level, and lighting mode corresponding to the same scene keyword in the historical instruction database, and generating a parameter recommendation list adapted to the current scene. The parameter recommendation list contains multiple sets of candidate parameters, preferably 1 to 3 sets.

3. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The Agent intelligent decision-making module also dynamically adjusts the sensitivity of the speech recognition module according to the ambient noise decibel value: when the ambient noise decibel value reaches or exceeds 60 decibels, the wake-up threshold of speech recognition is increased by 20% to 30%, and the beamforming technology of the microphone array is enabled to enhance user voice acquisition. When the ambient noise level is below 60 decibels, the default wake-up threshold is restored.

4. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The preset spatial coordinates of the light string are obtained in the following way: when the light string is installed, the user marks the physical position of each light bead or light bead segment through voice commands. The Agent intelligent decision module binds the voice-marked physical position with the hardware address of the light string to generate a spatial coordinate library.

5. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The custom combination pattern associated with the scene is implemented in the following way: After parsing the scene keywords in the text data, the Agent intelligent decision-making module configures differentiated working parameters for the divided multiple LED groups based on the preset lighting strategy corresponding to the scene keywords. The working parameters include at least color, brightness and dynamic mode. The preset lighting strategy includes configuring at least one LED group with different brightness levels and dynamic effects than other LED groups to achieve a collaborative lighting scene that matches the scene keywords.

6. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The Agent intelligent decision-making module also constructs a user preference model based on the user's selection records of the parameter recommendation list: if the user selects a certain type of color sequence three times in a row, the priority of that type of color sequence in the parameter recommendation list will be increased in subsequent analysis of the same scenario.

7. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The feedback instruction also includes a fault warning sub-instruction: when the current or voltage data of the lamp beads collected by the lighting control terminal exceeds the normal parameter range, the Agent intelligent decision module generates a fault warning voice and automatically adjusts the lamp bead group division parameters, skips the faulty lamp beads, and reissues the control instruction. The normal parameter range is ±10% of the rated current and ±5% of the rated voltage.

8. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: It also includes a multi-user voiceprint recognition step: the Agent intelligent decision module pre-stores the voiceprint features of multiple authorized users. In S2, the voiceprint features of the current user are first extracted by the speech recognition module and matched with the pre-stored voiceprint features. If the match is successful, the personalized parameter library corresponding to the user is called. If the match fails, only basic control permissions are granted.

9. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The Agent intelligent decision-making module adopts an edge computing and cloud collaborative architecture: basic semantic parsing, historical command retrieval, environmental data processing, and LED group division are completed at the edge terminal. Only when multiple LED strings are coordinated for control or complex scene combinations are parsed will the command be sent to the cloud large language model for deep parsing. The cloud parsing result is then returned to the edge terminal for execution.

10. The voice-based agent-based intelligent light string precise tuner control algorithm according to claim 1, characterized in that: The color map table supports user-defined extensions: users can input custom color names and corresponding RGB or GRB values ​​via voice commands, and the Agent intelligent decision-making module will associate and store the custom color names with the corresponding values ​​in the color map table, which can then be directly accessed using the custom color name.

Images