A lighting control system based on speech recognition and magnet control

Abstract

The application relates to the technical field of intelligent lighting control, in particular to a lighting control system based on voice recognition and magnetic control, which comprises a voice triggering module, a magnetic disturbance path construction module, a behavior orientation recognition module, a voice-magnetic action joint recognition module and a lighting state control module.In the application, through frame segment division and sound wave feature extraction on continuous audio streams, effective voice instructions can be accurately recognized in a complex voice environment, magnetic induction signals are mapped through time sequence magnetic flux direction, a disturbance path atlas is formed, detailed tracking of spatial magnetic field changes is realized, non-contact spatial operations have high direction sensitivity, angle change analysis of the magnetic flux disturbance path effectively depicts the movement trend of a user in the space, the accuracy of behavior recognition is improved, the accuracy and safety of equipment operation are guaranteed on the premise of meeting the user's spatial behavior intention, and the man-machine interaction efficiency and use experience in multiple scenes are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent lighting control technology, and in particular to a lighting control system based on voice recognition and magnetic control. Background Technology

[0002] The field of intelligent lighting control technology involves adjusting and controlling lighting components through integrated circuits and control technology. Its core aspects include lighting equipment state switching, brightness adjustment, sensor control, voice recognition, and magnetic signal response, enabling adaptive control of lighting in different environments or application scenarios. This technology systematically combines voice input devices, magnetic induction devices, and microcontroller units to achieve contactless interactive control between users and lighting devices. This is particularly suitable for space-constrained environments or situations requiring enhanced user experience, such as portable products, smart packaging, and interactive displays. The control methods often employ integrated voice recognition chips, magnetic sensors, low-power LEDs, and other hardware tools combined with embedded control logic for signal response and output control.

[0003] Among them, the voice recognition and magnetic control lighting control system refers to the method of turning LED lights on or off through physical switches or mechanical buttons. Its main technical aspects include the on / off control of the light source inside the gift box, the setting of the lighting duration, and the switching management of the control power supply. The operation control is completed by manual buttons, box lid opening and closing contacts, or simple photosensitive triggering devices. In terms of voice recognition, it mostly relies on a single volume threshold or keyword matching. The magnetic control part mostly uses a magnet to trigger a Hall element to achieve single lighting switch control. This type of method has a simple structure but lacks intelligent judgment and multi-dimensional interaction capabilities, making it difficult to meet the composite control requirements based on dual voice and magnetic field recognition.

[0004] Existing intelligent lighting controls generally rely on simple voice recognition or magnetic induction triggering, lacking the ability to deeply integrate spatial actions and semantic information. This results in poor robustness of control in diverse usage scenarios. Voice-based command recognition is easily affected by environmental noise, echoes, or conversations among multiple people, leading to misrecognition or missed commands, especially noticeable in public spaces and complex acoustic environments. Magnetic triggering uses simple proximity-based magnetic field change detection, lacking in-depth analysis of magnetic field disturbance paths and spatial directions. It cannot accurately distinguish user intentions, easily leading to misoperations by non-target users. For example, a user may inadvertently trigger a magnetically controlled area while moving, causing unexpected switching operations on the lighting equipment. Existing lighting controls lack awareness of user spatial behavior trends and cannot adjust control strategies based on the user's dynamic movement within the space. This results in slow response and insufficient command accuracy in large spaces or multi-device interconnected scenarios, severely impacting user experience and the level of intelligence in equipment management. Summary of the Invention

[0005] To address the shortcomings of existing technologies, such as the lack of deep integration of spatial actions and semantic information, which leads to poor robustness in diverse usage scenarios, and the vulnerability of voice-based command recognition to interference from environmental noise, echoes, or multi-person conversations, resulting in misrecognition or command omissions, especially in public spaces and complex acoustic environments, this invention provides a lighting control system based on voice recognition and magnetic control. The technical solution is as follows: Voice-based command recognition is easily affected by environmental noise, echoes, or conversations, leading to misoperation by non-target users. For example, a user may inadvertently trigger the magnetic control area while moving, causing unexpected switching on / off operations of the lighting equipment. Furthermore, existing lighting controls lack awareness of user spatial behavior trends and cannot adjust control strategies based on dynamic user movement. This results in slow response and insufficient command accuracy in large spaces or multi-device interconnected scenarios, severely impacting user experience and the level of intelligent equipment management.

[0006] On the one hand, a lighting control system based on voice recognition and magnetic control is provided, the system comprising:

[0007] The voice triggering module divides the continuous audio input stream within the lighting environment into frames in chronological order, detects the presence of dense frequency distributions in adjacent frames, extracts audio with amplitude variations and stable durations from consecutive frames, and generates voice command recognition segments.

[0008] The magnetic disturbance path construction module identifies the magnetic flux data recorded by the magnetic sensor in the corresponding area of ​​the segment through the voice command, extracts the induction value by sorting by time, performs direction mapping on the magnetic flux in adjacent time periods, marks the direction sequence and start and end point index of the disturbance segment, and generates a magnetic flux disturbance path map.

[0009] The behavior orientation recognition module calls the direction sequence of the disturbance segment in the magnetic flux disturbance path map, performs segment angle judgment, extracts the direction change angle of each path segment, compares the angle with the set direction of the lighting equipment, and retrieves the direction continuity sequence in the angle sequence to obtain the user's spatial movement trend record.

[0010] The speech-magnetic action recognition module performs intersection judgment on the time axis based on the path segment time sequence in the user space movement trend record, identifies whether there are overlapping intervals, assigns matching tags between semantics and magnetic control behavior, and generates a speech-magnetic linkage control list.

[0011] As a further aspect of the present invention, the voice command recognition segment includes a semantic segment start and end boundary index, an audio energy concentration frame group number, and a transferable control channel identifier; the magnetic flux disturbance path map includes a disturbance direction sequence identifier, a path segment time sequence label, and a magnetic flux continuous change mapping unit; the user space movement trend record includes an orientation stable path set number, an orientation change angle distribution label, and a lighting area orientation overlap description; and the voice-magnetic linkage control list includes a path semantic matching label, an overlap time period positioning number, and a linkage command source label.

[0012] As a further aspect of the present invention, the voice triggering module includes:

[0013] The frame segment construction submodule divides the audio stream into equal-length frames based on the continuous audio input stream in the lighting environment in chronological order, extracts the frequency information and sound wave amplitude values ​​in each frame, marks the time index and data sampling position of the constructed frame segments, retains the complete frame segment sequence and the time scale range of each segment, and generates a continuous frame segment basic information set.

[0014] The frequency vibration extraction submodule extracts the frequency change sequence of the continuous frames and records the amplitude fluctuation trend based on the basic information set of the continuous frame segments, detects the continuous and dense frequency distribution between the continuous frame segments, and determines whether the amplitude value in the continuous frame segments shows a unidirectional change trend, and generates a frequency vibration matching identifier sequence.

[0015] The segment recognition submodule calls the identifier frame segment in the frequency vibration matching identifier sequence, performs boundary determination on the start and end frame segments in the time series, extracts the audio segments covered by the identifier frame segment and adds sequence number and recognition label to generate voice command recognition segment.

[0016] As a further aspect of the present invention, the magnetic disturbance path construction module includes:

[0017] The sensing data recording submodule extracts the magnetic flux data of the magnetic sensor within the continuous time period of the area based on the range of the lighting control area marked in the voice command recognition section, records the sensing values ​​in chronological order, and adds timestamps and spatial location indexes to the data points to generate a magnetic flux sequence structure set.

[0018] The disturbance direction labeling submodule calls the magnetic flux at adjacent time points in the magnetic flux sequence structure set, analyzes the change amplitude and extracts the difference direction, calculates the disturbance trend feature value, classifies the set of data points with the same trend in continuous difference direction as disturbance segments, records the start index, end index and direction label of each segment, and generates disturbance direction segment set information;

[0019] The path map generation submodule, based on the time label, direction type, and sequence index of the disturbance segment in the disturbance direction segment set information, connects the disturbance segments in a timeline to form a complete magnetic disturbance path sequence, and performs structural classification on the direction turning points and duration segments of the path segments to generate a magnetic flux disturbance path map.

[0020] As a further aspect of the present invention, the disturbance trend characteristic value is expressed by the formula:

[0021] ;

[0022] in, Represents the characteristic value of the disturbance trend. Represents time points in the magnetic flux sequence The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The average value of magnetic flux within a given time point.

[0023] As a further aspect of the present invention, the behavior orientation recognition module includes:

[0024] The orientation angle extraction submodule calls the orientation sequence data of the disturbance segment in the magnetic flux disturbance path map, analyzes the orientation changes between adjacent path segments, extracts the turning angle value between each path direction and the previous direction, records the frame position and change type of the turning angle, and generates a path orientation turning angle set.

[0025] The angle comparison submodule, based on the direction change data in the path direction angle set, calls the lighting direction value set by the lighting equipment, calculates the angle between multiple path directions and lighting directions, filters data segments whose angle values ​​tend to be consistent within a continuous time period, records the time series number of the angle change trend, and generates a lighting direction angle sequence.

[0026] The path classification submodule distinguishes the degree of convergence of angles by span based on the continuous angle trend change segments in the lighting direction angle sequence, and performs distribution statistics based on the number and frequency of path turning angles. It then selects path segments with stable direction changes and limited fluctuation range of angle difference as feature segments to generate user spatial movement trend records.

[0027] As a further aspect of the present invention, the speech-magnetic action recognition module includes:

[0028] The time segment comparison submodule calls the path segment time series in the user space movement trend record and compares it with the start and end time index marked in the voice command recognition segment. It extracts the intersection interval of the two time series, locates the boundary of the time range of the intersection part, calculates the degree of intersection overlap index, records the overlapping segment number and duration, and generates the linkage segment intersection information.

[0029] The directional relationship mapping submodule extracts the path direction and the lamp position command direction value included in the semantics within the intersection segment according to the time range indicated by the intersection information of the linkage segment, determines whether the mapping relationship between the two directions in spatial angle tends to be consistent, and identifies the degree of direction consistency and mapping coverage, and generates a direction correspondence index set.

[0030] The matching tag submodule calls the time overlap tag in the intersection information of the direction consistency parameter and the linkage segment in the direction correspondence index set, combines the two results to construct the event identifier structure, assigns identifiable matching tags, and generates the speech-magnetic linkage control list.

[0031] As a further aspect of the present invention, the intersection overlap index is expressed by the formula:

[0032] ;

[0033] in, Indicators representing the degree of overlap, The time series path segment value represents the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the time index value of the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the number of sampling points within the o-th overlapping segment. This represents the absolute difference between the k-th sampling points within the 0-th overlapping segment.

[0034] As a further aspect of the present invention, the system also includes a lighting status control module:

[0035] The lighting status control module calls the working status parameters of the lighting equipment through the magnetic linkage control list, compares the relationship between the path direction in the matching tag and the set direction of the lamp, and if the real-time direction relationship meets the control permission area and the lamp status is within the activation condition, then pushes the control signal into the lighting control interface to generate the lighting action execution result.

[0036] The execution results of the lighting action include the control signal number, the activation response status of the lighting unit, and the area lighting control permission verification flag.

[0037] As a further aspect of the present invention, the lighting status control module includes:

[0038] The status parameter reading submodule calls the lamp number corresponding to the matching tag in the magnetic linkage control list, retrieves the real-time working status parameters of the lighting lamp, extracts the lamp activation status mark, operating mode type and control area identifier, and generates a lighting equipment status information set.

[0039] The direction permission verification submodule compares the path direction with the set illumination direction of the lamp based on the control area identifier in the lighting equipment status information set, and determines whether the section to which the path belongs is covered by the real-time control permission area. It also determines whether the lamp is in an activatable condition based on the lamp status and generates a lighting trigger judgment result.

[0040] The control signal output submodule pushes the corresponding activation signal to the lighting control interface based on the list of lamps that meet the control conditions in the lighting trigger judgment result, and labels each signal with lamp number, trigger type and execution time label, and generates lighting action execution result.

[0041] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0042] By segmenting continuous audio streams into frames and extracting acoustic features, effective voice commands can be accurately identified in complex speech environments. Dual judgment based on frequency density and amplitude changes improves the accuracy of voice command interception and significantly reduces the probability of misjudgment caused by environmental noise interference. Magnetic induction signals, through time-series magnetic flux direction mapping, form a disturbance path map, enabling detailed tracking of spatial magnetic field changes and giving non-contact spatial operations high directional sensitivity. Analysis of angular changes in magnetic flux disturbance paths effectively characterizes the user's movement trends in space, improving the accuracy of behavior recognition. This allows lighting control to not only rely on single voice triggers but also make auxiliary judgments based on spatial behavior changes. Time-series intersection analysis based on dual-channel voice and magnetic field data can quickly identify the correlation between user intent and actual spatial actions, thereby achieving more accurate command matching in dynamic scenes and avoiding false triggers or missed triggers that may occur with single-channel recognition. The control strategy achieves efficient scheduling of lighting equipment by comparing the working status, spatial orientation, and authorized areas of the lighting equipment in real time. While meeting the user's spatial behavior intentions, it ensures the accuracy and safety of equipment operation and greatly improves the efficiency of human-computer interaction and user experience in multiple scenarios. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of a lighting control system based on voice recognition and magnetic control provided in an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0046] Figure 3 This is a flowchart of the voice triggering module in this invention;

[0047] Figure 4 This is a flowchart of the magnetic disturbance path construction module in this invention;

[0048] Figure 5 This is a flowchart of the behavior orientation recognition module in this invention;

[0049] Figure 6 This is a flowchart of the language magnetic action association module of the present invention;

[0050] Figure 7 This is a flowchart of the lighting status control module in this invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0053] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0054] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0055] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0056] This invention provides a lighting control system based on voice recognition and magnetic control, such as... Figure 1-2 The diagram shown illustrates a lighting control system based on voice recognition and magnetic control. The system includes:

[0057] The voice triggering module is based on the continuous audio input stream in the lighting environment. It divides the frame segments according to the time sequence, extracts frequency data and sound wave amplitude information, detects whether there is a dense frequency distribution in adjacent frame segments, extracts audio with amplitude changes and stable duration in continuous frame segments, determines the end point, and extracts the audio between the start and end frames to form independent segments, generating voice command recognition segments.

[0058] The magnetic disturbance path construction module identifies the magnetic flux data recorded by the magnetic sensor in the corresponding area of ​​the segment through voice commands, extracts the induction value by sorting by time, performs direction mapping on the magnetic flux in adjacent time periods, marks the direction sequence and start and end point index of the disturbance segment, and generates a magnetic flux disturbance path map.

[0059] The behavior orientation recognition module calls the direction sequence of the disturbance segment in the magnetic flux disturbance path map, performs segment angle judgment, extracts the angle of change of each path direction, compares the angle with the set direction of the lighting equipment, retrieves the direction continuity sequence in the angle sequence, and records the frequency and span distribution of the path broken angle. The continuous direction convergence segment is marked as the orientation stable path segment, thus obtaining the user's spatial movement trend record.

[0060] The speech-magnetic action recognition module judges the intersection of the path segments in the user's spatial movement trend record on the time axis to identify whether there is an overlapping interval. If the time periods of the two overlap and the path direction corresponds to the lamp position command in the semantics, it is determined to be a related event, and a matching label between the semantics and the magnetic control behavior is assigned to generate a speech-magnetic linkage control list.

[0061] The lighting status control module calls the working status parameters of the lighting equipment through the magnetic linkage control list, compares the relationship between the path direction in the matching tag and the set direction of the lamp, and if the real-time direction relationship meets the control permission area and the lamp status is within the activation condition, then pushes the control signal into the lighting control interface to generate the lighting action execution result.

[0062] The voice command recognition section includes semantic segment start and end boundary indexes, audio energy concentration frame group numbers, and transmittable control channel identifiers. The magnetic flux disturbance path map includes disturbance direction sequence identifiers, path segment time sequence labels, and magnetic flux continuous change mapping units. The user space movement trend record includes orientation stable path set numbers, direction change angle distribution labels, and lighting area direction overlap descriptions. The voice-magnetic linkage control list includes path semantic matching labels, overlap time period location numbers, and linkage command source labels. The lighting action execution results include control signal numbers, lighting unit activation response status, and area lighting control permission verification flags.

[0063] Specifically, such as Figure 2 , 3 As shown, the voice triggering module includes:

[0064] The frame segment construction submodule divides the audio stream into equal-length frames based on the continuous audio input stream in the lighting environment in chronological order, extracts the frequency information and sound wave amplitude values ​​in each frame, marks the time index and data sampling position of the constructed frame segments, retains the complete frame segment sequence and the time scale range of each segment, and generates a continuous frame segment basic information set.

[0065] The system continuously collects ambient audio data from audio acquisition devices deployed around smart lighting equipment. Audio data is acquired in real-time at a fixed sampling rate, and the acquired signal stream is processed into frames of equal duration (20 milliseconds per frame). Frequency features and amplitude information are extracted from each frame. During this process, the system constructs multiple buffer areas to store the original audio data and converted spectrum data of the current frame. A time index is set according to the start time of each frame acquisition, and a sampling position index is formed by recording the sample position of each frame in the continuous audio stream. Each frame is structurally encapsulated to form a multi-dimensional data frame containing frequency features, amplitude intensity, time markers, and sampling point annotations. By constructing a sequence according to the frame acquisition order, the system maintains the integrity of the temporal connection between frame segments. In a smart lighting scenario, if a user controls the lighting equipment at home using voice, the system receives this continuous voice stream, performs preliminary data structuring, and ensures that subsequent frequency and vibration extraction and voice segment recognition are based on accurate and orderly frame data, generating a continuous frame segment basic information set.

[0066] The frequency vibration extraction submodule extracts the frequency change sequence of continuous frames and records the amplitude fluctuation trend based on the basic information set of continuous frame segments. It detects the continuous and dense frequency distribution between continuous frame segments, determines whether the amplitude value in the continuous frame segment shows a unidirectional change trend, and generates a frequency vibration matching identifier sequence.

[0067] The system performs frame-by-frame analysis on the frequency information in consecutive frames. During this process, it is necessary to identify the main frequency components and their amplitude fluctuation characteristics in each frame, and track whether the frequency between frames remains stable or shows a concentrated trend in the continuous interval. The system performs frequency continuity detection with a certain frame window. It is set in a sliding window of four frames. If the difference in the main frequency value of each frame is small and all are within the specified frequency band, it is judged as a combination of frequency-dense frames. At the same time, the main frequency amplitude value of each frame is tested for unidirectionality, that is, to observe whether the amplitude value in consecutive frames always shows an increasing or decreasing state. In a smart lighting environment, the user issues the voice command "turn on the lights". The system detects that the main frequency is concentrated at about 1200Hz from the 10th to the 15th frame, and the amplitude value increases from 65dB to 80dB. Then, the frame sequence can be marked as having a significant frequency vibration trend. The frames that meet the conditions are marked with "1" in the sequence to support the subsequent extraction of key segments and generate a frequency vibration matching identification sequence.

[0068] The segment recognition submodule calls the identifier frame segment in the frequency vibration matching identifier sequence, performs boundary determination on the start and end frame segments in the time series, extracts the audio segments covered by the identifier frame segment and adds sequence number and recognition label to generate voice command recognition segment;

[0069] The system performs continuous segment detection on frames marked "1", identifies the start and end frame numbers of the matching identifier, and determines the time boundary of each potential voice command. During this process, a minimum threshold for the number of consecutive frames to be identified is set to eliminate occasional frequency vibration frames caused by environmental noise. A minimum of 3 consecutive matching frames is required to constitute a valid segment. After identification, the system reconstructs the original audio signal segment based on the frame segment index, extracts the segment content and assigns a sequence number, and adds an identification tag for subsequent command recognition or control module calls. In smart lighting applications, the system sets the user's voice segment "turn off the bedroom light" to last from frame 30 to frame 38. The system identifies this segment as a valid voice command segment and marks it as Segment03, with the corresponding command tag "TURNOFFBEDROOM". This ensures that the segment has clear boundaries and identification attributes before further semantic parsing, enhancing the reliability and execution efficiency of subsequent calls and generating a voice command recognition segment.

[0070] Specifically, such as Figure 2 , 4 As shown, the magnetic disturbance path construction module includes:

[0071] The sensing data recording submodule identifies the marked lighting control area in the segment based on voice commands, extracts the magnetic flux data of the magnetic sensor in the area for a continuous time period, records the sensing values ​​in chronological order, and adds timestamps and spatial location indexes to the data points to generate a magnetic flux sequence structure set.

[0072] When a voice command is associated with a spatial location, and the recognition tag contains the keywords "bedroom" or "kitchen," the system will activate the magnetic sensors in the corresponding area. It will then extract the magnetic flux change data collected by the sensors in real time. This extraction operation will sample at fixed time intervals to ensure data integrity within continuous time periods. Simultaneously, the system needs to record the current timestamp for each set of sampled data to mark the precise location of the data on the time axis. The position of each magnetic sensor will also be marked in the spatial coordinate system. For example, using the two-dimensional coordinate index set in the building layout diagram, the magnetic sensor at the bedroom door will be set to (2, 5). The collected data points will be appended... With this index, a data point Di = {timestamp ti, position (xi, yi), magnetic flux value φi} is formed. The system arranges the data points in chronological order to form a magnetic flux sequence structure set. This structure set can be used to track the magnetic field response process of a user's entry, exit, or movement trajectory. For example, if a user enters the bedroom and says "turn on the light," the system extracts a data stream with a significant change in φ value from the magnetic sensor on the bedroom door frame before and after receiving the instruction. For example, from t = 3.2 seconds to t = 4.1 seconds, the φ value increases from 20 units to 70 units. Combined with the spatial index, a complete magnetic flux response sequence within this time period can be formed, generating a magnetic flux sequence structure set.

[0073] The disturbance direction labeling submodule calls the magnetic flux at adjacent time points in the magnetic flux sequence structure set, analyzes the change amplitude and extracts the difference direction, calculates the disturbance trend feature value, classifies the set of data points with the same trend in continuous difference direction as disturbance segments, records the start index, end index and direction label of each segment, and generates disturbance direction segment set information;

[0074] The disturbance trend characteristic value is calculated using the following formula:

[0075] ;

[0076] in, Represents the characteristic value of the disturbance trend. Represents time points in the magnetic flux sequence The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The average value of magnetic flux within a given time point;

[0077] The calculation logic of the formula is mainly based on the change pattern of adjacent time points in the magnetic flux sequence, taking into account both abrupt changes in amplitude and reversals in direction, and calculating the time points. , , The differences among the three magnetic flux values ​​are combined to construct a weighted difference sequence, reflecting the turning point characteristics of the instantaneous disturbance direction. The average value of the magnetic flux before and after a certain time point is introduced to normalize the current value, forming a relative offset of the trend change. The above difference term is then combined with the normalization standard as the numerator of the characteristic value, and the denominator is composed of the squared difference between the magnetic flux value and its mean, which plays the role of normalization and enhancing the weight of change. The calculated disturbance trend characteristic value Δφᵢ can sensitively reflect the instantaneous dynamics of trend intensification or reversal in the magnetic flux disturbance sequence, and is effectively used for subsequent inflection point identification and time series segmentation analysis.

[0078] Meaning of parameters and derivation of formulas:

[0079] Magnetic flux data were acquired using an RM3100 triaxial magnetometer with a sampling rate of 10Hz and an acquisition interval of 0.1 seconds. The acquired magnetic flux data are as follows:

[0080] Time point :magnetic flux Wb;

[0081] Time point :magnetic flux Wb;

[0082] Time point :magnetic flux Wb;

[0083] Calculation time point Disturbance trend eigenvalues The steps are as follows:

[0084] Calculate the average value of magnetic flux :

[0085] ;

[0086] Calculate the numerator:

[0087] ;

[0088] ;

[0089] ;

[0090] Sum of numerators: ;

[0091] Calculate the denominator:

[0092] ;

[0093] ;

[0094] ;

[0095] Summation: ;

[0096] Denominator: ;

[0097] Calculate the characteristic value of the disturbance trend:

[0098] ;

[0099] This result indicates that the time point The perturbation trend characteristic value is approximately 0.0008 Wb. This perturbation trend characteristic value is used to characterize the severity and directional turning characteristics of magnetic flux changes over time. By weighted normalization processing based on the differences in magnetic flux data at three adjacent time points and the deviation from the average value, the ability to identify local perturbation abrupt changes is enhanced. This index can be used to extract trend inflection points in the magnetic flux perturbation sequence, and to help distinguish between stable and abrupt change stages. It indicates that the change trend of magnetic flux is relatively stable at this time point, without significant perturbation. This value can be used to further analyze the perturbation segment in the magnetic flux sequence and help identify abnormal regions of magnetic flux change.

[0100] The path map generation submodule, based on the time label, direction type, and sequence index of the disturbance segment in the disturbance direction segment set information, connects the disturbance segments in a timeline to form a complete magnetic disturbance path sequence, and performs structural classification on the direction turning points and duration segments of the path segments to generate a magnetic flux disturbance path map.

[0101] The system integrates the time labels, direction types, and segment order of disturbance segments, connects them in chronological order to form a continuous magnetic disturbance path, representing the user's behavioral trajectory and magnetic field interference process within the lighting control area. This process requires extracting the start and end times of each disturbance segment and identifying the direction change points of adjacent disturbance segments. A change from "ascending" to "descending" is defined as a direction change point. The system records the time and spatial index of each change point and calculates the duration of each disturbance segment, constructing a time-direction dual-structure information set. In a smart lighting scenario, if a user moves from the doorway to the window and then back in the kitchen, the magnetic sensor collects multiple disturbance direction change data segments. The system sequentially splices these segments into path segments, forming a disturbance path map containing multiple direction change nodes and duration segments. For example, if the path sequentially presents an "ascending-ascending-descending-descending-ascending" structure, the system can sequentially number the segments P1 to P5, generate a complete map, and store it for subsequent analysis, thus generating a magnetic flux disturbance path map.

[0102] Specifically, such as Figure 2 , 5As shown, the behavior orientation recognition module includes:

[0103] The orientation angle extraction submodule calls the orientation sequence data of the disturbance segment in the magnetic flux disturbance path map, analyzes the orientation changes between adjacent path segments, extracts the turning angle value between each path direction and the previous direction, records the frame position and change type of the turning angle, and generates a path orientation turning angle set.

[0104] The system extracts the directional attribute values ​​of each disturbance segment and arranges them in chronological order according to the path segments to form a directional sequence. Based on this, the system performs angle difference analysis on the directional values ​​between two adjacent path segments, that is, identifies the turning angle formed between the direction of the previous path segment and the direction of the current path segment. The system needs to combine the spatial index information of the path segments to identify the direction vector of the path segments and calculate the angle between the direction vectors on the two-dimensional plane to form the path direction turning angle value. This operation is performed once at each direction change point. For example, if a user walks from the living room northward and then turns eastward, the system identifies that the direction vectors of the preceding and following path segments point to the north and east respectively, forming a 90-degree directional angle. The system marks the turning angle position as the turning angle occurrence frame segment and records the change type such as "right turn", "left turn" or "return". In this way, the system constructs a set of data structures consisting of the path segment sequence index, turning angle value and change type to reflect the directional change relationship between each segment and the previous segment in the complete path and its specific location index, generating a path direction turning angle set.

[0105] The angle comparison submodule uses the direction change data in the path direction angle set to call the lighting direction value set by the lighting equipment, calculates the angle between multiple path directions and lighting directions, filters the data segments whose angle values ​​tend to be consistent in a continuous time period, records the time series number of the angle change trend, and generates the lighting direction angle sequence.

[0106] The system calls the preset lighting direction data of the lighting equipment, sets the main living room light direction to 30 degrees east of north, and uses this direction data as a comparison benchmark. The system calculates the angle between each segment of the path direction and the lighting direction in the path direction angle set. After constructing a direction vector based on the user's actual path direction, it calculates the angle value with the lighting direction vector and arranges the obtained angle values ​​in the order of the path segments into an angle time series. During this process, the system sets a continuous time window to identify whether the angle value remains relatively consistent over a period of time, that is, whether the angle change amplitude is less than the preset fluctuation range (set not to exceed ±10 degrees). If the condition is met, the angle time series segment is marked as a consistent segment and its time series number is recorded. If the angles in path segments P3 to P6 are 42 degrees, 44 degrees, 45 degrees, and 43 degrees respectively, and the change amplitude is within 5 degrees, it can be determined as a segment with a stable angle trend, and a lighting direction angle sequence is generated.

[0107] The path classification submodule distinguishes the degree of convergence of angles by span based on the continuous angle trend change segments in the lighting direction angle sequence, and performs distribution statistics based on the number and frequency of path turning angles. It then selects path segments with stable direction changes and limited fluctuation range of angle difference as feature segments to generate user spatial movement trend records.

[0108] Further processing is required for path segments marked with consistent angle change trends. This involves determining the convergence degree of each angle trend, i.e., analyzing whether the angle value fluctuates around a fixed directional value. The system sets an angle convergence interval; when the angle value fluctuates within ±5 degrees of a central value, it is considered a high-convergence segment. The system statistically analyzes the number and type distribution of turning angles concentrated on the corresponding path segment. If only one slight directional change occurs in a path segment with consistent angle trends, the system identifies that path segment as a stable directional change segment. Then, the fluctuation range is determined based on the angle fluctuation amplitude threshold. For example, if the angle is between 40 and 44 degrees with a fluctuation amplitude of 4 degrees, the system records that segment as a stable angle segment. Path filtering is performed based on turning frequency, classifying path segments with few directional changes and limited angle fluctuations as characteristic path segments. For example, when a user walks from the kitchen to the living room, if the overall path direction is stable, with few turning angle changes and the direction of travel is consistent with the lighting direction, then this path segment will be identified as a typical spatial movement trend and recorded in the user movement trend database, generating a user spatial movement trend record.

[0109] Specifically, such as Figure 2 , 6 As shown, the speech-magnetic action association module includes:

[0110] The time segment comparison submodule calls the path segment time series in the user space movement trend record and compares it with the start and end time index marked in the voice command recognition segment. It extracts the intersection interval of the two time series, locates the boundary of the time range of the intersection part, calculates the degree of intersection overlap index, records the overlapping segment number and duration, and generates the linkage segment intersection information.

[0111] The degree of overlap is measured by the following formula:

[0112] ;

[0113] in, Indicators representing the degree of overlap, The time series path segment value represents the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the time index value of the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the number of sampling points within the o-th overlapping segment. This represents the absolute difference between the k-th sampling points within the 0-th overlapping segment;

[0114] The formula's calculation logic is used to quantify the overlap strength of two time-series path sequences in terms of spatial movement trends. This process extracts path segment sample values ​​from the corresponding overlapping sections of the two time series. With time index value The algorithm calculates the average difference and absolute deviation of each sampling point along the path value. Then, it normalizes the path sequence average and time index values ​​separately to handle the impact of different sample sizes and scales of change, enhancing the consistency of the calculations. The formula consists of three parts: first, the mean difference of path values, reflecting trend shift; second, the time series center deviation, reflecting temporal misalignment; and third, the absolute difference of sampling points, reflecting single-point error. The integrated result is... The smaller the index, the closer the spatial-temporal characteristics of the two trajectories are and the more significant the overlap; the larger the index, the more significant the differences in trend and temporal sequence between the two path segments. The index can be used in tasks such as behavioral pattern recognition, path prediction and time segment aggregation.

[0115] Meaning of parameters and derivation of formulas:

[0116] In the intersection interval, the first Within each overlapping segment, the number of sampling points is The corresponding number of sampling points within 1 second, with a sampling frequency of 16kHz and a sampling interval of 10ms, is calculated as follows: Each frame contains 16kHz × 10ms = 160 sampling points;

[0117] Time series path segment values ​​of user space movement trend records Data is acquired via an inertial measurement unit (IMU), using accelerometer and gyroscope data. Velocity and displacement information are obtained through filtering and integration. The time index value is marked in the voice command recognition segment. The audio signal is processed by speech recognition to extract features and identify the start and end times of the voice command.

[0118] Set the following parameter values: The mean value is 0.5, and the standard deviation is 0.1. The mean value is 0.6, and the standard deviation is 0.1. ;

[0119] The calculation process is as follows:

[0120] Calculate the Euclidean distance term:

[0121] ;

[0122] Calculate the average value: ;

[0123] Calculate the absolute value difference term: ;

[0124] Substitute the above results into the formula:

[0125] ;

[0126] The results show that the intersection overlap index is 1.9798. The intersection overlap index is used to measure the similarity of two time period path sequences in terms of spatial movement trends. It is calculated by normalizing the path value difference, time index offset and sampling point absolute error to reflect the consistency of the trajectory in the overlapping area. This index can be used to determine the matching degree and overlap characteristics of user behavior trajectory. The larger the value, the greater the difference in time series path segment values ​​between the user's spatial movement trend record and the voice command recognition segment, and the lower the degree of overlap. This index can be used to evaluate the synchronization between user behavior and voice commands, and further analyze the correlation between user behavior and voice commands.

[0127] The directional relationship mapping submodule extracts the path direction and the lamp position command direction value included in the semantics within the intersection segment based on the time range indicated by the intersection information of the linkage segment. It determines whether the mapping relationship between the two directions in spatial angle tends to be consistent, and identifies the degree of direction consistency and mapping coverage, and generates a directional correspondence index set.

[0128] The system retrieves the intersection time period marked in the intersection information of the linkage segments and extracts the path direction data within that time window. Based on the direction label in each frame of the path segment, the system obtains the user's movement direction within this interval. Simultaneously, it extracts the target direction of the lighting control contained in the voice command from the semantic recognition module. If the statement contains "turn on the left light," the system presets the lighting position direction as "west." The system performs spatial angle mapping analysis between this direction value and the path direction to determine whether the user's movement direction is consistent with the target lighting direction. If the user moves in a west-northwest direction during the intersection time period, while the lighting direction is due west, the system calculates the angle deviation between the two. If the deviation is within the set tolerance range (e.g., ±15 degrees), it is marked as "consistent direction." If the deviation exceeds the set range, it is marked as "inconsistent direction." The system also needs to count the proportion of data frames with consistent direction in the intersection time period as a mapping coverage indicator. For example, if 16 out of 20 frames have consistent direction, the coverage rate is 80%. The system summarizes the direction consistency degree and coverage data corresponding to each intersection segment to generate a direction correspondence index set.

[0129] The matching tag submodule calls the direction correspondence index set for direction consistency parameters and the time overlap tags in the intersection information of linkage segments, combines the two results to construct an event identifier structure, assigns identifiable matching tags, and generates a speech-magnetic linkage control list.

[0130] Based on the directional consistency parameters provided by the directional correspondence index set and the time overlap labels in the intersection information of the linkage segments, the two results are combined to construct an event identifier structure. That is, for each intersection segment, a judgment result is constructed on whether its spatial behavior direction matches the directional intent of the voice command. For the intersection number Overlap01, its directional consistency is set to "high consistency" and the time overlap rate is 66%. The system combines its event features into a structure containing multiple parameters such as number, time, directional consistency, and mapping coverage. The system performs matching label assignment on this structure. If the directional consistency is "high consistency" and the coverage rate exceeds 75%, the label is assigned as "action matching - command valid". If the directional deviation is large or the coverage rate is insufficient, the label is "action deviation - command not triggered". Each record includes the intersection segment number, directional consistency label, time overlap ratio, and event executable identifier, which is used to guide the intelligent lighting system to respond according to the voice command or to trigger the linkage control logic flow, and to generate a voice-magnetic linkage control list.

[0131] Specifically, such as Figure 2 , 7 As shown, the lighting status control module includes:

[0132] The status parameter reading submodule calls the lamp number corresponding to the matching tag in the magnetic linkage control list, retrieves the real-time working status parameters of the lighting lamp, extracts the lamp activation status mark, operating mode type and control area identifier, and generates a lighting equipment status information set.

[0133] The system retrieves the real-time status parameters of the current lighting fixtures one by one according to the corresponding luminaire number. During this process, communication with the intelligent lighting control platform is required. The system queries the activation status flag of the luminaire, setting the status field to "ON" to indicate activation and "OFF" to indicate deactivation. At the same time, it extracts the operating mode type of the luminaire, such as "automatic dimming," "constant brightness," or "manual control," and obtains the control area identifier to which the luminaire belongs. This identifier is generally represented by a zone number or spatial label, such as "ZONE1" representing the living room area. Through parameters, the system can generate a status record item for each luminaire corresponding to a matching tag. The luminaire number L001 is set to "OFF" status, "automatic dimming" mode, and "ZONE3" control area. The system integrates the status items of the luminaires into a lighting equipment status information set for subsequent control permission judgment and direction consistency verification. When the user issues the command "turn on the kitchen light" and the matching tag points to the luminaire with the number L002, the system will search for the status item of L002 in the status information set as a prerequisite for determining whether the command can be executed, and generate the lighting equipment status information set.

[0134] The direction permission verification sub-module compares the angle between the path direction and the set irradiation direction of the lamp based on the control area identifier in the lighting device status information set, determines whether the section to which the path belongs is covered by the real-time control permission area, combines the lamp status to determine whether it is in an activatable condition, and generates a lighting trigger judgment result;

[0135] Extract the set irradiation direction of the area where the lamp is located. Set the set direction of a certain lamp to "20 degrees west of south". The system needs to compare this direction value with the user's traveling direction in the path direction data. By comparing whether the path segment is consistent with the lamp direction in the spatial angle, it is judged whether the path has a physical association with the current lamp, and whether the spatial area to which the path belongs is within the control permission range of the lamp. When the user path segment mark is set to "ZONE3" and the set control area of the lamp is also "ZONE3", it is regarded as effective path coverage, otherwise it is judged as permission mismatch.Combined with the activation status and operation mode of the lamp, it is judged whether the lamp is in an activatable condition. If the status is "OFF" and the operation mode is "automatic dimming", it can be judged that the activation condition is met. Otherwise, even if the direction and area match, no action is taken. The system records the verified results in a structured manner. Each record includes the path segment index, lamp number, direction comparison status, permission verification status, and activation condition judgment conclusion, and generates a lighting trigger judgment result.

[0136] The control signal output sub-module pushes the corresponding activation signal to the lighting control interface according to the list of lamps that meet the control conditions in the lighting trigger judgment result, and marks each signal with the lamp number, trigger type, and execution time label, and generates a lighting action execution result;

[0137] Process each lamp that meets the control conditions item by item. The system generates a control signal structure according to the lamp numbers recorded in the list, sets the trigger type field, such as "automatic turn on", "mode switch", etc., and marks the lamp number and the current timestamp as the execution time label in the signal. Set that lamp L005 needs to be automatically turned on. The system will generate a control signal containing the fields: the lamp number is L005, the trigger type is "automatic turn on", and the execution time is "2025-05-28 14:36:12". This signal is pushed to the lighting control execution through the control interface to ensure that the command reaches the target lamp as planned. The system also needs to record the execution feedback status of each control signal, and summarize the successfully executed signals into a lighting action execution result for subsequent user behavior tracking or system response evaluation. For example, the result list generated after a round of complete linkage control contains five lighting action records, corresponding to the automatic activation actions of five lamps respectively. Each record identifies the specific lamp, execution time, and control type, and generates a lighting action execution result.

[0138] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A lighting control system based on voice recognition and magnetic control, characterized in that, The system includes: The voice triggering module divides the continuous audio input stream within the lighting environment into frames in chronological order, detects the presence of dense frequency distributions in adjacent frames, extracts audio with amplitude variations and stable durations from consecutive frames, and generates voice command recognition segments. The magnetic disturbance path construction module identifies the magnetic flux data recorded by the magnetic sensor in the corresponding area of ​​the segment through the voice command, extracts the induction value by sorting by time, performs direction mapping on the magnetic flux in adjacent time periods, marks the direction sequence and start and end point index of the disturbance segment, and generates a magnetic flux disturbance path map. The magnetic disturbance path construction module includes: The sensing data recording submodule extracts the magnetic flux data of the magnetic sensor within the continuous time period of the area based on the range of the lighting control area marked in the voice command recognition section, records the sensing values ​​in chronological order, and adds timestamps and spatial location indexes to the data points to generate a magnetic flux sequence structure set. The disturbance direction labeling submodule calls the magnetic flux at adjacent time points in the magnetic flux sequence structure set, analyzes the change amplitude and extracts the difference direction, calculates the disturbance trend feature value, classifies the set of data points with the same trend in continuous difference direction as disturbance segments, records the start index, end index and direction label of each segment, and generates disturbance direction segment set information; The disturbance trend characteristic value is expressed by the formula: ； in, Represents the characteristic value of the disturbance trend. Represents time points in the magnetic flux sequence The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The magnetic flux, Representing a point in time The average magnetic flux within a given time point; The path map generation submodule, based on the time label, direction type, and sequence index of the disturbance segment in the disturbance direction segment set information, connects the disturbance segments in a timeline to form a complete magnetic disturbance path sequence, and performs structural classification on the direction turning points and duration segments of the path segments to generate a magnetic flux disturbance path map. The behavior orientation recognition module calls the direction sequence of the disturbance segment in the magnetic flux disturbance path map, performs segment angle judgment, extracts the direction change angle of each path segment, compares the angle with the set direction of the lighting equipment, and retrieves the direction continuity sequence in the angle sequence to obtain the user's spatial movement trend record. The speech-magnetic action recognition module performs intersection judgment on the timeline based on the path segment time sequence in the user space movement trend record, identifies whether there are overlapping intervals, assigns matching tags between semantics and magnetic control behavior, and generates a speech-magnetic linkage control list. The speech-magnetic action association module includes: The time segment comparison submodule calls the path segment time series in the user space movement trend record and compares it with the start and end time index marked in the voice command recognition segment. It extracts the intersection interval of the two time series, locates the boundary of the time range of the intersection part, calculates the degree of intersection overlap index, records the overlapping segment number and duration, and generates the linkage segment intersection information. The directional relationship mapping submodule extracts the path direction and the lamp position command direction value included in the semantics within the intersection segment according to the time range indicated by the intersection information of the linkage segment, determines whether the mapping relationship between the two directions in spatial angle tends to be consistent, and identifies the degree of direction consistency and mapping coverage, and generates a direction correspondence index set. The matching tag submodule calls the time overlap tag in the intersection information of the direction consistency parameter and the linkage segment in the direction correspondence index set, combines the two results to construct the event identifier structure, assigns identifiable matching tags, and generates the speech-magnetic linkage control list.

2. The lighting control system based on voice recognition and magnetic control according to claim 1, characterized in that, The voice command recognition segment includes a semantic segment start and end boundary index, an audio energy concentration frame group number, and a transferable control channel identifier. The magnetic flux disturbance path map includes a disturbance direction sequence identifier, a path segment time sequence label, and a magnetic flux continuous change mapping unit. The user space movement trend record includes an orientation stable path set number, an orientation change angle distribution label, and a description of the overlap of lighting area orientations. The voice-magnetic linkage control list includes a path semantic matching label, an overlap time period positioning number, and a linkage command source label.

3. The lighting control system based on voice recognition and magnetic control according to claim 1, characterized in that, The voice triggering module includes: The frame segment construction submodule divides the audio stream into equal-length frames based on the continuous audio input stream in the lighting environment in chronological order, extracts the frequency information and sound wave amplitude values ​​in each frame, marks the time index and data sampling position of the constructed frame segments, retains the complete frame segment sequence and the time scale range of each segment, and generates a continuous frame segment basic information set. The frequency vibration extraction submodule extracts the frequency change sequence of the continuous frames and records the amplitude fluctuation trend based on the basic information set of the continuous frame segments, detects the continuous and dense frequency distribution between the continuous frame segments, and determines whether the amplitude value in the continuous frame segments shows a unidirectional change trend, and generates a frequency vibration matching identifier sequence. The segment recognition submodule calls the identifier frame segment in the frequency vibration matching identifier sequence, performs boundary determination on the start and end frame segments in the time series, extracts the audio segments covered by the identifier frame segment and adds sequence number and recognition label to generate voice command recognition segment.

4. The lighting control system based on voice recognition and magnetic control according to claim 3, characterized in that, The behavior orientation recognition module includes: The orientation angle extraction submodule calls the orientation sequence data of the disturbance segment in the magnetic flux disturbance path map, analyzes the orientation changes between adjacent path segments, extracts the turning angle value between each path direction and the previous direction, records the frame position and change type of the turning angle, and generates a path orientation turning angle set. The angle comparison submodule, based on the direction change data in the path direction angle set, calls the lighting direction value set by the lighting equipment, calculates the angle between multiple path directions and lighting directions, filters data segments whose angle values ​​tend to be consistent within a continuous time period, records the time series number of the angle change trend, and generates a lighting direction angle sequence. The path classification submodule distinguishes the degree of convergence of angles by span based on the continuous angle trend change segments in the lighting direction angle sequence, and performs distribution statistics based on the number and frequency of path turning angles. It then selects path segments with stable direction changes and limited fluctuation range of angle difference as feature segments to generate user spatial movement trend records.

5. The lighting control system based on voice recognition and magnetic control according to claim 1, characterized in that, The degree of overlap index is expressed by the formula: ； in, Indicators representing the degree of overlap, The time series path segment value represents the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the time index value of the k-th sampling point within the 0-th overlapping segment of the intersection interval. This represents the number of sampling points within the o-th overlapping segment. This represents the absolute difference between the k-th sampling points within the 0-th overlapping segment.

6. The lighting control system based on voice recognition and magnetic control according to claim 1, characterized in that, The system also includes a lighting status control module: The lighting status control module calls the working status parameters of the lighting equipment through the magnetic linkage control list, compares the relationship between the path direction in the matching tag and the set direction of the lamp, and if the real-time direction relationship meets the control permission area and the lamp status is within the activation condition, then pushes the control signal into the lighting control interface to generate the lighting action execution result. The execution results of the lighting action include the control signal number, the activation response status of the lighting unit, and the area lighting control permission verification flag.

7. The lighting control system based on voice recognition and magnetic control according to claim 6, characterized in that, The lighting status control module includes: The status parameter reading submodule calls the lamp number corresponding to the matching tag in the magnetic linkage control list, retrieves the real-time working status parameters of the lighting lamp, extracts the lamp activation status mark, operating mode type and control area identifier, and generates a lighting equipment status information set. The direction permission verification submodule compares the path direction with the set illumination direction of the lamp based on the control area identifier in the lighting equipment status information set, and determines whether the section to which the path belongs is covered by the real-time control permission area. It also determines whether the lamp is in an activatable condition based on the lamp status and generates a lighting trigger judgment result. The control signal output submodule pushes the corresponding activation signal to the lighting control interface based on the list of lamps that meet the control conditions in the lighting trigger judgment result, and labels each signal with lamp number, trigger type and execution time label, and generates lighting action execution result.

Images