Direct-current fan and illumination integrated intelligent driving system based on multi-mode perception

Abstract

The invention provides a DC fan and illumination integrated intelligent driving system based on multi-mode perception. The system comprises a data acquisition module, an analysis decision module, a driving execution module and an updating optimization module. The data acquisition module is used for sensing multi-modal information of an environment and a user; the analysis decision module is used for carrying out fusion analysis on the multi-modal information to realize a personalized and scenario-based wind-solar linkage control strategy; the driving execution module is used for controlling and driving the fan and the lighting device; the updating optimization module is used for continuously optimizing a system control strategy; according to the invention, through combination of multi-mode perception and self-learning control, intelligent linkage and self-adaptive adjustment of the fan and the lighting device are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control and home automation technology, and particularly relates to a DC fan and lighting integrated intelligent driving system based on multi-modal perception. BACKGROUND

[0002] With the development of smart home and Internet of Things technology, automatic control and human-computer interaction experience of home environment have gradually become the focus of research and application. Existing fan and lighting devices are mostly controlled independently, and users need to operate separately, which cannot realize coordinated adjustment according to environmental changes and user behaviors, resulting in low energy utilization, response lag and poor user experience.

[0003] At present, some intelligent control systems realize automatic adjustment through temperature and humidity sensors or light sensors, but most of them still rely on single modal perception information, lack comprehensive judgment of user activity state, environmental context and subjective comfort, and are difficult to realize personalized dynamic control. In addition, the traditional control strategy is mostly static threshold setting, which cannot be self-optimized according to user habits and real-time feedback, and the intelligent degree of the system is limited.

[0004] In practical application, multi-source information such as light conditions, user position, activity intensity and voice instructions has a significant impact on environmental comfort and control requirements. If visual, acoustic and environmental multi-modal perception information can be fully utilized, combined with adaptive control and learning optimization strategy, linkage and self-evolution between fan and lighting devices can be realized, which will help to improve the comfort, energy saving and interactive intelligence level of indoor environment.

[0005] After checking the related disclosed technical solutions, the technical solution with publication number CN207634359U proposes an intelligent fan control system, which includes an Arduino control mainboard, a steering device, a sensor expansion board, a DC motor and a second stepping motor. The steering device includes a first stepping motor and a tray, and the first stepping motor is connected with the Arduino control mainboard and the tray respectively. The sensor expansion board is inserted on the Arduino control mainboard, and the sensor expansion board is connected with an HC-SR501 infrared sensor, an L298N motor driving module, three E18 infrared sensors and two LED light emitting diodes respectively. The L298N motor driving module is connected with the DC motor. This scheme judges whether a person exists and the position of the person through the infrared sensor, controls the steering, start and stop of the fan, overcomes the inconvenience of mechanical knob control of the traditional fan, increases the user experience, has the characteristics of intelligence and power saving, improves the life quality, and is energy-saving and environment-friendly. However, this scheme mainly relies on a single infrared sensor to judge the existence and position of the person, and the perception dimension, decision-making ability and intelligent degree are limited. SUMMARY

[0006] The purpose of this invention is to address the shortcomings of current technologies by proposing an integrated intelligent drive system for DC fans and lighting based on multimodal sensing.

[0007] The present invention adopts the following technical solution: A multimodal sensing-based integrated intelligent drive system for DC fans and lighting is disclosed. The system includes a data acquisition module, an analysis and decision-making module, a drive execution module, and an update and optimization module. The data acquisition module is used to sense multimodal information of the environment and users. The analysis and decision-making module is used to perform fusion analysis on the multimodal information to realize personalized and contextualized wind and solar linkage control strategies. The drive execution module is used to control and drive the fan and lighting devices. The update and optimization module is used to continuously optimize the system control strategy.

[0008] The data acquisition module includes an environmental sensing unit, a visual sensing unit, and an acoustic sensing unit; the environmental sensing unit is used to acquire environmental sensing information, including environmental temperature, humidity, and light intensity information; the visual sensing unit is used to acquire environmental image information; and the acoustic sensing unit is used to acquire voice command interaction information issued by the user in the environment.

[0009] The analysis and decision-making module includes a data fusion unit, a user behavior recognition unit, and an adaptive decision-making unit. The data fusion unit is used to align multimodal information with timestamps. The user behavior recognition unit is used to analyze user activity status by combining environmental perception information and environmental image information. The adaptive decision-making unit is used to issue control commands for fans and lighting devices by combining user activity status and user voice command interaction information.

[0010] Furthermore, the adaptive decision-making unit includes an association mapping subunit and a voice adjustment subunit; the association mapping subunit is used to combine environmental perception information and user activity status to establish a mapping relationship between environmental perception information and user activity status and fan lighting device control parameters, and generate control commands accordingly to achieve adaptive adjustment of fan speed and lighting brightness; the voice adjustment subunit adjusts the control parameters of the fan and lighting device in real time based on real-time user voice command interaction information.

[0011] Furthermore, the specific workflow of the association mapping subunit is as follows: S11: For each system-identifiable user activity state, combining historical experience and prior knowledge to pursue thermally neutral comfort and optimal visual illumination, pre-establish the environmental perception information parameter range corresponding to the target comfort environment under each user activity state. S12: Obtain real-time environmental perception information parameters and compare them with the target environmental perception information parameter range in the previous step. Calculate the deviation magnitude and deviation direction of the environmental perception information parameters in each dimension from the midpoint of the target environmental perception information parameter range, and output a unified multi-dimensional deviation vector. S13: Using a multi-dimensional deviation vector as the core input, the system generates control commands for adjusting fan speed and lighting brightness through PID control logic to achieve adaptive adjustment. Furthermore, the voice adjustment subunit includes a semantic recognizer and a mapping controller; the semantic recognizer is used to identify control command keywords and control degree words in voice command interaction information; the mapping controller contains a preset command action mapping library, used to convert control command keywords and control degree words into specific control parameter commands.

[0012] Furthermore, the drive execution module includes a main controller, a fan motor drive circuit, and a lighting drive circuit; the main controller is used to receive control commands output by the analysis and decision module, and to coordinate and control the fan motor drive circuit and the lighting drive circuit according to the control commands. Furthermore, the update and optimization module includes a runtime data acquisition unit and a self-learning unit; the runtime data acquisition unit is used to collect multi-source sensing information during system operation and to timestamp and synchronize the multi-source sensing information; the multi-source sensing information includes environmental sensing information, user behavior information, and device control information; the user behavior information includes user activity status and voice command interaction information records; the device control information includes the output and execution delay of control commands; the self-learning unit is used to adaptively update and optimize the system's control strategy by combining the multi-source sensing information.

[0013] Furthermore, the self-learning unit optimizes the target environment perception information parameter range to reduce the frequency of user intervention in system adjustments and minimize deviations in environmental comfort. The specific workflow of the self-learning unit is as follows: S21: Continuously extract interaction records containing explicit user feedback from multi-sensory information as training samples. Each training sample consists of: a status label of the user's activity state, a sequence of environmental perception information in the activity state, a record of user feedback voice command interaction information, and device control information. S22: For a certain user activity state, when the accumulated training samples in that activity state reach a specified threshold number, an update of the target environment perception information parameter range for that user activity state is triggered; the initial center point of the target environment perception information parameter range is set as... The interval width is The target environment perception information parameter range for this user's activity state can be optimized in the following ways: ; in, The center point of the optimized target environment perception information parameter range. The center point of the target environment perception information parameter range before optimization; The learning rate has a range of values. ; For the current optimization phase The validity weights of each training sample. This represents the total number of training samples in the current optimization process. For the current optimization phase Sensitivity weights of training samples For the first The environmental perception parameters, adjusted by the output control commands, are recorded in the training samples after user voice command interaction. This is to prevent extremely small positive numbers with a denominator of 0; ; in, The optimized interval width, and satisfying It always stays within the preset interval width; The interval width before optimization. This is the width sensitivity adjustment coefficient, used to control the range of width variation within the interval. Its value is set through pre-experimentation and has a specific range. ; The number of adjustments recorded based on user voice command interaction information within the previous specified statistical period prior to the current optimization time; The number of adjustments recorded based on user voice command interaction information within the second specified statistical period prior to the current optimization time.

[0014] The beneficial effects achieved by this invention are: This invention achieves intelligent and personalized control of DC fans and lighting systems by combining multimodal perception with self-learning optimization. The system can accurately identify user activity status based on environmental perception information, environmental image information, and voice command interaction information, and realize adaptive linkage adjustment of wind speed and lighting brightness through correlation mapping and PID control logic. At the same time, it introduces an online self-learning mechanism based on explicit user feedback to dynamically optimize the range of comfortable environmental parameters, so that the system control strategy continuously approaches the user's actual preferences. This invention effectively reduces the frequency of manual adjustment by users, improves environmental comfort, control responsiveness, and system intelligence, and has high practicality and human-computer interaction experience value. Attached Figure Description

[0015] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0016] Figure 1 This is a schematic diagram of the overall modules of the present invention.

[0017] Figure 2 This is a schematic diagram of the workflow of the data acquisition module of the present invention.

[0018] Figure 3 This is a schematic diagram of the workflow of the analysis and decision-making module of the present invention.

[0019] Figure 4 This is a schematic diagram comparing the key performance indicators of the system of this invention with those of a traditional system.

[0020] Figure 5 This diagram illustrates the improvement in key performance indicators of the system of the present invention compared to traditional systems. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention. Other systems, methods, and / or features of this embodiment will become apparent to those skilled in the art after reviewing the following detailed description. It is intended that all such additional systems, methods, features, and advantages are included within this specification, are included within the scope of the present invention, and are protected by the appended claims. Further features of the disclosed embodiments are described in detail below, and these features will become apparent from the following detailed description.

[0022] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present patent. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0023] Example 1:

[0024] like Figure 1 , Figure 2 ,Figure 3 As shown in the figure, this embodiment provides an integrated intelligent drive system for DC fans and lighting based on multimodal perception. The system includes a data acquisition module, an analysis and decision-making module, a drive execution module, and an update and optimization module. The data acquisition module is used to sense multimodal information of the environment and the user. The analysis and decision-making module is used to perform fusion analysis on the multimodal information to realize personalized and contextualized wind and solar linkage control strategies. The drive execution module is used to control and drive the fan and lighting devices. The update and optimization module is used to continuously optimize the system control strategy. The data acquisition module includes an environmental sensing unit, a visual sensing unit, and an acoustic sensing unit; the environmental sensing unit is used to acquire environmental sensing information, including environmental temperature and humidity and light intensity information; the visual sensing unit is used to acquire environmental image information; and the acoustic sensing unit is used to acquire voice command interaction information issued by the user in the environment. The analysis and decision-making module includes a data fusion unit, a user behavior recognition unit, and an adaptive decision-making unit. The data fusion unit is used to align multimodal information with timestamps. The user behavior recognition unit is used to analyze user activity status by combining environmental perception information and environmental image information. The adaptive decision-making unit is used to issue control commands for fans and lighting devices by combining user activity status and user voice command interaction information. Furthermore, the adaptive decision-making unit includes an association mapping subunit and a voice adjustment subunit; the association mapping subunit is used to combine environmental perception information and user activity status to establish a mapping relationship between environmental perception information and user activity status and fan lighting device control parameters, and generate control commands accordingly to achieve adaptive adjustment of fan speed and lighting brightness; the voice adjustment subunit adjusts the control parameters of the fan and lighting device in real time based on real-time user voice command interaction information. Furthermore, the specific workflow of the association mapping subunit is as follows: S11: For each system-identifiable user activity state, combining historical experience and prior knowledge to pursue thermally neutral comfort and optimal visual illumination, pre-establish the environmental perception information parameter range corresponding to the target comfort environment under each user activity state. S12: Obtain real-time environmental perception information parameters and compare them with the target environmental perception information parameter range in the previous step. Calculate the deviation magnitude and deviation direction of the environmental perception information parameters in each dimension from the midpoint of the target environmental perception information parameter range, and output a unified multi-dimensional deviation vector. S13: Using a multi-dimensional deviation vector as the core input, the system generates control commands for adjusting fan speed and lighting brightness through PID control logic to achieve adaptive adjustment. Furthermore, the voice adjustment subunit includes a semantic recognizer and a mapping controller; the semantic recognizer is used to recognize control command keywords and control degree words in voice command interaction information; the mapping controller contains a preset command action mapping library, which is used to convert control command keywords and control degree words into specific control parameter commands; Furthermore, the drive execution module includes a main controller, a fan motor drive circuit, and a lighting drive circuit; the main controller is used to receive control commands output by the analysis and decision module, and to coordinate and control the fan motor drive circuit and the lighting drive circuit according to the control commands. Furthermore, the update and optimization module includes a runtime data acquisition unit and a self-learning unit; the runtime data acquisition unit is used to collect multi-source perception information during system operation and to timestamp and synchronize the multi-source perception information; the multi-source perception information includes environmental perception information, user behavior information, and device control information; the user behavior information includes user activity status and voice command interaction information records; the device control information includes the output and execution delay of control commands; the self-learning unit is used to adaptively update and optimize the system's control strategy by combining the multi-source perception information. Furthermore, the self-learning unit optimizes the target environment perception information parameter range to reduce the frequency of user intervention in system adjustments and minimize deviations in environmental comfort. The specific workflow of the self-learning unit is as follows: S21: Continuously extract interaction records containing explicit user feedback from multi-sensory information as training samples. Each training sample consists of: a status label of the user's activity state, a sequence of environmental perception information in the activity state, a record of user feedback voice command interaction information, and device control information. S22: For a certain user activity state, when the accumulated training samples in that activity state reach a specified threshold number, an update of the target environment perception information parameter range for that user activity state is triggered; the initial center point of the target environment perception information parameter range is set as... The interval width is The target environment perception information parameter range for this user's activity state can be optimized in the following ways: ; in, The center point of the optimized target environment perception information parameter range. The center point of the target environment perception information parameter range before optimization; The learning rate has a range of values. ; For the current optimization phase The validity weights of each training sample. This represents the total number of training samples in the current optimization process. For the current optimization phase Sensitivity weights of training samples For the first The environmental perception information parameters, adjusted by the output control commands, are recorded in the training samples after user voice command interaction. To prevent extremely small positive numbers with a denominator of 0, the following condition must be met: ; in, This is the effective threshold for filtering noise, and its value is a very small positive number. ; in, For the first The time distance between the time when a training sample appears and the current optimization time. The time normalization coefficient can be set to the time distance between the time of the furthest training sample and the current optimization time. For the first The number of adjustments recorded in each training sample based on user voice command interaction information. For use in normalization The coefficient; ; in, The optimized interval width, and satisfying It always stays within the preset interval width; The interval width before optimization. This is the width sensitivity adjustment coefficient, used to control the range of width variation within the interval. Its value is set through pre-experimentation and has a specific range. ; The number of adjustments recorded based on user voice command interaction information within the previous specified statistical period prior to the current optimization time; The number of adjustments recorded based on user voice command interaction information within the second specified statistical period prior to the current optimization time; This solution establishes a dynamic mapping relationship between the target comfort environment parameter range and multimodal perception information, and introduces an online self-learning mechanism based on explicit user feedback. Through correlation mapping and PID control logic, it automatically adjusts the environmental perception information parameters to the comfort range, significantly reducing the need for manual intervention by the user. By analyzing the user's historical adjustment behavior, it dynamically optimizes the center point and width of the comfort range, enabling the system control strategy to continuously approach the user's true preferences. This achieves a personalized, adaptive, and continuously optimized environmental control experience, effectively improving user comfort and the system's intelligent interaction level in different scenarios.

[0025] Example 2:

[0026] This embodiment should be understood to include at least all the features of any of the foregoing embodiments, and to further improve upon them; This embodiment provides an integrated intelligent drive system for DC fans and lighting based on multimodal perception. The system includes a data acquisition module, an analysis and decision-making module, a drive execution module, and an update and optimization module. The data acquisition module is used to sense multimodal information of the environment and the user. The analysis and decision-making module is used to perform fusion analysis on the multimodal information to realize personalized and contextualized wind and solar linkage control strategies. The drive execution module is used to control and drive the fan and lighting devices. The update and optimization module is used to continuously optimize the system control strategy. The data acquisition module includes an environmental sensing unit, a visual sensing unit, and an acoustic sensing unit; the environmental sensing unit is used to acquire environmental sensing information, including environmental temperature and humidity and light intensity information; the visual sensing unit is used to acquire environmental image information; and the acoustic sensing unit is used to acquire voice command interaction information issued by the user in the environment. The analysis and decision-making module includes a data fusion unit, a user behavior recognition unit, and an adaptive decision-making unit. The data fusion unit is used to align multimodal information with timestamps. The user behavior recognition unit is used to analyze user activity status by combining environmental perception information and environmental image information. The adaptive decision-making unit is used to issue control commands for fans and lighting devices by combining user activity status and user voice command interaction information. Furthermore, the user behavior recognition unit includes an image parser, an environment context parser, and a state determiner. The image parser is used to extract user posture features based on environmental image information, combined with image detection and pose estimation algorithms, and outputs an activity state recognition result based on visual features by combining the posture feature sequence with a pre-established classifier. The activity state recognition result is in the form of a confidence distribution for each activity state. The environment context parser calculates the similarity vector between the current environmental perception information and the center point of the target environmental perception information parameter interval under each activity state using Gaussian similarity calculation, and transforms the similarity vector into a confidence distribution for each activity state to output an activity state recognition result based on environmental reasoning. Furthermore, the state determiner determines the current user's activity state by fusing the activity state recognition results output by the image parser and the environment context parser; the fusion process is as follows: ; in, The fused activity state identification result is in the form of the activity state label corresponding to the highest confidence level; The results of activity state recognition based on visual features. The results of activity state identification based on environment reasoning. For visual feature weight coefficients, satisfying: ; in, Given the current light intensity, To ensure the minimum illumination intensity required for the visual perception unit to acquire stable environmental image information, experimental calibration was performed. This is the transition sensitivity coefficient, used to assess the sensitivity of light intensity to the visual feature weighting coefficients. Its value range is [value range missing]. ; This solution combines visual recognition results with environmental context reasoning results through a dynamic weighted fusion mechanism that adapts to illumination. This effectively overcomes the industry challenge of a sharp drop in reliability of single visual recognition in low-light environments, ensuring the robustness of user activity status recognition at all times. At the same time, by using environmental perception information to assist in reasoning, the overall accuracy and rationality of status determination are further improved, providing a solid and reliable decision-making basis for subsequent wind-solar linkage intelligent control.

[0027] Example 3:

[0028] This embodiment should be understood to include at least all the features of any of the foregoing embodiments, and to further improve upon them; This embodiment uses an open office area in a science and technology park as the application scenario. The area has a total area of ​​1,200 square meters, accommodates 180 employees, and has 125 workstations, 4 meeting rooms and 1 rest area. The system adopts a centralized control and distributed execution architecture. Based on Embodiment 1 and Embodiment 2, it provides specific equipment selection, deployment parameters and detailed quantitative analysis of implementation effects. The data acquisition module employs a high-precision, low-power IoT sensor array; the environmental sensing unit uses a Sensirion SCD40 CO2 sensor combined with a temperature and humidity sensor, with a temperature measurement range of -10°C to 60°C and an accuracy of ±0.5°C; and a humidity measurement range of 0% to 100%RH and an accuracy of ±3%RH; the light intensity monitoring uses a Bosch BHI260AP intelligent microelectromechanical system, integrating an ambient light sensor, with a measurement range of 0-150klux and an accuracy of ±5%; the visual sensing unit uses an Orbbec Astra Pro depth camera, supporting simultaneous acquisition of RGB and depth information, with a resolution of 1280x720 and an effective detection distance of 0.6-8 meters, used to accurately capture user posture and position; the acoustic sensing unit uses a Respeaker 6-Mic ring microphone array, supporting 360° sound source localization and beamforming, with a wake-up word recognition rate >97% and an effective pickup distance of 5 meters; The core hardware of the analysis and decision-making module uses the NVIDIA Jetson Xavier NX edge AI computing module, equipped with a 6-core ARM CPU and a 384-core Volta GPU, with a computing power of 21 TOPS and a power consumption of less than 20W. In the user behavior recognition unit, the pose estimation algorithm run by the image parser is optimized based on the OpenPose model, achieving an inference speed of 15 FPS on the Jetson platform, and an activity state recognition accuracy of 96.8% on the test set. The Gaussian similarity calculation cycle of the environment context parser is 1 second. The fusion weight of the state determiner... Medium, lowest light intensity The measured and calibrated value is 50 lux, and the transition sensitivity coefficient is... Set to 0.03; The drive execution module adopts a high-precision digital drive scheme. The fan motor drive circuit uses a TIDRV10987 three-phase sensorless BLDC motor driver, supporting PWM speed regulation, with a frequency range of 20Hz-20kHz and an efficiency of up to 92%. The driven fan is a custom-designed brushless DC motor with a rated voltage of 24V, a speed range of 200-2000 RPM, and a noise level below 30 dB. The lighting drive circuit uses a Mean Well HLG-240H-48B constant voltage / constant current LED driver power supply, paired with a PWM dimming controller, to achieve stepless adjustment of lighting brightness from 0-100% and continuous adjustment of color temperature from 2700K-6500K. The main controller uses an STM32H743 high-performance microcontroller, communicating with a Jetson module via a CAN bus, with an instruction execution latency of <10 milliseconds. The self-learning unit of the update and optimization module runs on the Jetson module; the specified threshold number of training samples accumulated is set to 50; in the center point optimization algorithm, the learning rate... Set to 0.1; taking ambient temperature as an example, the effectiveness threshold. Based on the characteristics of ambient temperature fluctuations, the time normalization coefficient is set to 0.8°C; Dynamically set the time span for the current optimized statistical period; This represents the sum of adjustment times for all samples within the current period; the width sensitivity adjustment coefficient in the interval width optimization formula. Set to 0.2; optimize the statistical period to 7 days (optimize the interval width in weekly units); Regarding the effectiveness of system implementation, such as Figure 4 , Figure 5 As shown, through a three-month comparative operation test (compared to a traditional system using a fixed temperature setting and manual switching), the system in this embodiment shows significant improvements in key performance indicators, including: Improved comfort: Users' proactive satisfaction rating for ambient temperature increased from 7.2 to 9.1 out of 10; through self-learning optimization, the center point of the target temperature range is aligned with the overall preferences of the user group; Intervention frequency reduced: The average number of times users adjust the scenery by voice or manually per day decreased from 4.7 times to 0.9 times, with an automation coverage rate of 80.9%, significantly reducing the user's operational burden; Energy efficiency optimization: by adjusting the range width The system dynamically adjusts (average width narrowed by 18%) and automatically enters a low-power mode when unattended, resulting in a 31.5% reduction in the total energy consumption of the wind and solar system in the area, with an average daily energy saving of approximately 18.2 kWh. Regarding the modification and replacement of technical means, the edge computing core can be replaced with an Intel Neural Compute Stick (NCS2) paired with an UP Squared motherboard. Although the computing power is slightly weaker (approximately 4 TOPS), the cost can be reduced by 40%, making it suitable for scenarios with less demanding real-time requirements. The self-learning optimization algorithm can be replaced with a statistical method based on a simple moving average. Although it is simple to implement and has low computational overhead, it cannot achieve personalization and has a slow convergence speed, resulting in an overall reduction of approximately 50% in the improvement of user experience. This embodiment fully verifies the effectiveness, reliability, and economy of the integrated intelligent drive system in a real, complex office environment through specific equipment selection, precise parameter configuration, and quantified operational data. It provides a solid engineering practice basis for the large-scale commercial promotion of the system in scenarios such as smart buildings and high-end homes.

[0029] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A DC fan and lighting integrated intelligent drive system based on multimodal sensing, characterized in that, The system includes a data acquisition module, an analysis and decision-making module, a drive execution module, and an update and optimization module. The data acquisition module is used to sense multimodal information of the environment and users. The analysis and decision-making module is used to perform fusion analysis on the multimodal information to realize personalized and contextualized wind and light linkage control strategies. The drive execution module is used to control and drive the fan and lighting devices. The update and optimization module is used to continuously optimize the system control strategy. The data acquisition module includes an environmental sensing unit, a visual sensing unit, and an acoustic sensing unit; the environmental sensing unit is used to acquire environmental sensing information, including environmental temperature and humidity and light intensity information; the visual sensing unit is used to acquire environmental image information; and the acoustic sensing unit is used to acquire voice command interaction information issued by the user in the environment. The analysis and decision-making module includes a data fusion unit, a user behavior recognition unit, and an adaptive decision-making unit. The data fusion unit is used to align multimodal information with timestamps. The user behavior recognition unit is used to analyze user activity status by combining environmental perception information and environmental image information. The adaptive decision-making unit is used to issue control commands for fans and lighting devices by combining user activity status and user voice command interaction information. The update and optimization module includes a runtime data acquisition unit and a self-learning unit. The runtime data acquisition unit is used to collect multi-source sensing information during system operation and to timestamp and synchronize the multi-source sensing information. The multi-source sensing information includes environmental sensing information, user behavior information, and device control information. The user behavior information includes the user's activity status and voice command interaction information records. The device control information includes the output and execution delay of control commands. The self-learning unit is used to adaptively update and optimize the system's control strategy by combining the multi-source sensing information.

2. The self-learning unit optimizes the target environment perception information parameter range to reduce the frequency of user intervention in system adjustments and minimize deviations in environmental comfort. The specific workflow of the self-learning unit is as follows: S21: Continuously extract interaction records containing explicit user feedback from multi-sensory information as training samples, wherein each training sample comprises: The status label of the user's current activity state, as well as the environmental perception information sequence, user feedback voice command interaction information record, and device control information in that activity state; S22: For a certain user activity state, when the number of training samples in that activity state reaches a specified threshold, an update of the target environment perception information parameter range for that user activity state is triggered. The initial target environment perception information parameter range center point is set as The interval width is The target environment perception information parameter range for this user's activity state can be optimized in the following ways: ； in, The center point of the optimized target environment perception information parameter range. The center point of the target environment perception information parameter range before optimization; The learning rate has a range of values. ; For the current optimization phase The validity weights of each training sample. This represents the total number of training samples in the current optimization process. For the current optimization phase Sensitivity weights of training samples For the first The environmental perception parameters, adjusted by the output control commands, are recorded in the training samples after user voice command interaction. This is to prevent extremely small positive numbers with a denominator of 0; ； in, This is the effective threshold for filtering noise, and its value is a very small positive number. ； in, For the first The time distance between the time when a training sample appears and the current optimization time. The time normalization coefficient can be set to the time distance between the time of the furthest training sample and the current optimization time. For the first The number of adjustments recorded in each training sample based on user voice command interaction information. For normalization The coefficient; ； in, The optimized interval width, and satisfying It always stays within the preset interval width; The interval width before optimization. This is the width sensitivity adjustment coefficient, used to control the range of width variation within the interval. Its value is set through pre-experimentation and has a specific range. ; The number of adjustments recorded based on user voice command interaction information within the previous specified statistical period prior to the current optimization time; The number of adjustments recorded based on user voice command interaction information within the second specified statistical period prior to the current optimization time.

3. The integrated intelligent drive system for DC fans and lighting based on multimodal sensing according to claim 1, characterized in that, The adaptive decision-making unit includes an association mapping subunit and a voice adjustment subunit. The association mapping subunit is used to combine environmental perception information and user activity status to establish a mapping relationship between environmental perception information and user activity status and control parameters of the fan and lighting device, and generate control commands accordingly to achieve adaptive adjustment of fan speed and lighting brightness. The voice adjustment subunit adjusts the control parameters of the fan and lighting device in real time based on real-time user voice command interaction information.

4. The integrated intelligent drive system for DC fans and lighting based on multimodal sensing according to claim 2, characterized in that, The specific workflow of the association mapping subunit is as follows: S11: For each system-identifiable user activity state, combining historical experience and prior knowledge to pursue thermally neutral comfort and optimal visual illumination, pre-establish the environmental perception information parameter range corresponding to the target comfort environment under each user activity state. S12: Obtain real-time environmental perception information parameters and compare them with the target environmental perception information parameter range in the previous step. Calculate the deviation magnitude and deviation direction of the environmental perception information parameters in each dimension from the midpoint of the target environmental perception information parameter range, and output a unified multi-dimensional deviation vector. S13: Using a multi-dimensional deviation vector as the core input, the system generates control commands for adjusting fan speed and lighting brightness through PID control logic to achieve adaptive adjustment.

5. The integrated intelligent drive system for DC fans and lighting based on multimodal sensing according to claim 2, characterized in that, The voice adjustment subunit includes a semantic recognizer and a mapping controller; the semantic recognizer is used to identify control command keywords and control degree words in voice command interaction information; the mapping controller contains a preset command action mapping library, which is used to convert control command keywords and control degree words into specific control parameter commands.

6. The integrated intelligent drive system for DC fans and lighting based on multimodal sensing according to claim 1, characterized in that, The drive execution module includes a main controller, a fan motor drive circuit, and a lighting drive circuit; the main controller is used to receive control commands output by the analysis and decision module, and to coordinate and control the fan motor drive circuit and the lighting drive circuit according to the control commands.

Images