A method for analyzing the effect of campus labor education by integrating moral education scenes

Abstract

The application relates to the technical field of wisdom education and Internet of Things data processing, and discloses a campus labor education effect data analysis method fusing a moral education scene, which synchronously collects acceleration and UWB positioning data by using an intelligent terminal, and extracts micro-dimension action efficiency values and macro-dimension displacement flux values in parallel. By constructing a motion topological coupling coefficient and introducing a statistical dispersion suppression mechanism, physical interlocking verification is carried out on limb work and space coverage, and pseudo-labor behaviors such as empty waving and wandering are accurately removed in a multi-dimensional feature space. The system generates multi-dimensional evaluation indexes including labor saturation, concentration and responsibility based on information entropy. By introducing a task-adaptive dual-domain feature extraction mechanism, frequency domain energy integration is used for periodic labor such as sweeping, and time domain stiffness energy accumulation is used for discrete labor such as picking, so that the technical limitation of low recognition rate of traditional methods under non-steady-state actions is overcome.

Description

Technical Field

[0001] This invention relates to the field of smart education and Internet of Things data processing technology, specifically a data analysis method for the effectiveness of campus labor education that integrates moral education scenarios. Background Technology

[0002] With labor education being formally incorporated into the comprehensive education system of primary and secondary schools, how to objectively and scientifically monitor and evaluate students' labor processes has become a key issue in the digital transformation of education. In the current construction of smart campuses, wearable devices based on Internet of Things (IoT) technology have begun to be applied to student physical health monitoring and behavior analysis, but existing data processing technologies still have significant limitations in evaluation applications targeting specific labor scenarios.

[0003] Current labor evaluation technologies primarily rely on single-dimensional sensor data, either calculating steps and acceleration energy consumption solely based on inertial sensors, or statistically analyzing movement distance and trajectory heatmaps based solely on positioning technology. This single-modal or simply superimposed data processing approach ignores the fact that labor behavior is, in its physical essence, a spatiotemporal unity of microscopic physical movements and macroscopic spatial displacement. Specifically, existing step counting or energy integration algorithms cannot distinguish between effective and pseudo-labor behavior. When the monitored subject is in a state of vigorous arm-waving in place, or wandering rapidly within a work area without performing hand actions, traditional algorithms often misjudge high-intensity labor input due to the detection of high-amplitude acceleration signals or long movement paths. This lack of a physical characteristic logical verification mechanism makes the system highly susceptible to subjective deception or uncontrolled actions, resulting in evaluation results that fail to accurately reflect the student's actual labor effectiveness.

[0004] Furthermore, in mixed labor scenarios involving periodic actions such as sweeping and mopping as well as discrete actions such as picking up and wiping, existing technologies often employ fixed threshold judgment models, lacking the ability to adaptively extract features from different action modalities, resulting in low accuracy in recognizing non-steady-state actions. Simultaneously, limited by hardware costs, low-power sensors often exhibit random drift and mechanical vibration noise; without multi-dimensional data interlocking verification, this noise is frequently misintegrated as effective workload. Regarding evaluation metrics, existing coverage calculations are mostly based on the geometric statistics of original trajectory points, failing to eliminate trajectory data from invalid wandering periods. This leads to inflated quantitative evaluations of responsibility or work uniformity, making it difficult to meet the needs of accurate and objective profiling of labor process quality in educational settings. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a data analysis method for the effectiveness of campus labor education that integrates moral education scenarios. This method solves the problem that existing labor evaluation methods, which rely on a single dimension (such as simply counting steps or energy points), cannot effectively distinguish between genuine labor and pseudo-labor behaviors such as standing still or loitering, thus leading to a lack of objectivity and accuracy in the evaluation results.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a data analysis method for the effectiveness of campus labor education integrated with moral education scenarios, comprising the following steps: Acquire inertial sensor data sequences and spatial positioning data sequences of the target object within a preset work period. Determine the feature extraction mode based on the work task attributes performed by the target object; the feature extraction mode includes periodic and discrete modes. Based on the feature extraction mode, perform adaptive feature extraction on the inertial sensor data sequences to generate action performance values. Calculate the displacement flux value of the target object within the corresponding time window based on the spatial positioning data sequence. Construct a motion topology coupling coefficient, which characterizes the correlation between the action performance value and the displacement flux value. Compare the motion topology coupling coefficient with preset valid work verification conditions to filter out valid work frames. Based on the filtered valid work frames, generate a moral education effect evaluation result including focus index or responsibility index.

[0007] In one specific embodiment, the step of generating motion performance values ​​includes: when the labor task attribute corresponds to a continuous operation such as sweeping or mopping, it is determined to be a periodic mode, and motion performance values ​​are extracted using a frequency domain analysis method; when the labor task attribute corresponds to a sudden operation such as picking up or wiping, it is determined to be a discrete mode, and motion performance values ​​are extracted using a time domain analysis method. By adopting heterogeneous feature extraction logic for different labor natures, the accuracy of physical representation of different job types is ensured.

[0008] Preferably, the step of extracting the action performance value using the frequency domain analysis method specifically includes: firstly, calculating the resultant acceleration sequence after removing the gravity component to eliminate DC component interference caused by the sensor wearing angle; performing a fast Fourier transform on the resultant acceleration sequence to obtain the spectral distribution; selecting a preset action frequency range, calculating the spectral energy density integral within the preset action frequency range, and using the integral result as the action performance value to quantify the action intensity of periodic labor.

[0009] In one specific embodiment, the step of extracting the motion performance value using a time-domain analysis method specifically includes: calculating the resultant acceleration sequence after removing the gravity component; obtaining the first derivative of the resultant acceleration sequence to obtain the stiffness sequence, where the stiffness sequence represents the rate of change of acceleration; calculating the cumulative energy of the stiffness sequence within a time window, and using the cumulative result as the motion performance value. This step achieves the quantification of non-steady-state, discrete motion characteristics by capturing instantaneous changes in acceleration.

[0010] Preferably, the step of calculating the displacement flux value of the target object within the corresponding time window specifically includes: when it is determined to be a discrete mode, considering the asynchrony between discrete actions and physical displacement on the time axis, the calculation range of the current time window is extended backward, and the spatial positioning data sequence is extracted using the extended time window; the cumulative Euclidean distance of the spatial positioning coordinates within the extended time window is calculated as the displacement flux value.

[0011] In one specific embodiment, the step of constructing the motion topology coupling coefficient specifically includes: calculating the statistical variance of the inertial sensor data sequence and the statistical variance of the spatial positioning data sequence within the current time window. The product of the motion performance value and the displacement flux value is used as the principal term, and the statistical variances of the inertial sensor data sequence and the spatial positioning data sequence are used as penalty terms. The motion topology coupling coefficient is constructed such that its magnitude is positively correlated with the principal term and negatively correlated with the penalty term. The motion topology coupling coefficient is used to suppress noise interference caused by sensor hardware drift or non-laborious random jitter, and to enforce the physical interlocking logic between microscopic motion and macroscopic displacement.

[0012] Preferably, the step of selecting valid work frames specifically includes: calculating the ratio of the action efficiency value to the displacement flux value using a manifold constraint algorithm; determining whether the motion topology coupling coefficient is greater than a preset minimum threshold, and simultaneously determining whether the ratio falls within a preset valid range; marking the current time window as a valid work frame only when both of the above conditions are met. This eliminates pseudo-work behaviors such as idling in place (ratio too large) or wandering while holding an object (ratio too small).

[0013] In one specific embodiment, the step of generating a moral education effectiveness evaluation result that includes a focus index or a sense of responsibility index includes: counting the total duration of the time window marked as an effective labor frame; calculating the proportion of the total duration of the time window of the effective labor frame to the total duration of the preset labor period, and mapping the proportion to a focus index.

[0014] Preferably, the step of generating a moral education effectiveness evaluation result including a focus index or a responsibility index further includes: obtaining a set of spatial positioning coordinate points corresponding to all effective labor frames; dividing the preset labor area into several grid units, and statistically analyzing the probability distribution of the set of spatial positioning coordinate points falling into each grid unit; calculating an information entropy value based on the probability distribution, and mapping the information entropy value to a responsibility index. The higher the information entropy value, the more uniformly the coverage of the responsibility area is achieved by the target object's work trajectory while maintaining an effective labor state.

[0015] In one specific embodiment, after acquiring the inertial sensor data sequence and spatial positioning data sequence of the target object within a preset working period, the method further performs a preprocessing step: using the sampling timestamp of the inertial sensor data sequence as a reference, performing linear interpolation on the spatial positioning data sequence with a lower sampling frequency; generating a time-aligned synchronization state data vector, and performing sliding slicing processing on the synchronization state data vector according to a preset step size, thereby providing an aligned data source for subsequent coupling calculations.

[0016] This invention, through the aforementioned technical solution, utilizes a motion topology coupling mechanism to perform nonlinear verification on multi-source sensor data. This objectively eliminates pseudo-motion data lacking physical displacement support, as well as invalid displacement data lacking motion characteristics. Compared to single-dimensional monitoring methods, this invention achieves quantitative analysis of the effective investment level and spatial responsibility coverage in campus labor education without increasing additional hardware costs.

[0017] This invention provides a data analysis method for the effectiveness of campus labor education that integrates moral education scenarios. It has the following beneficial effects: 1. This invention introduces a task-adaptive dual-domain feature extraction mechanism, which uses frequency domain energy integration for periodic labor such as sweeping and time domain force energy accumulation for discrete labor such as picking up. This overcomes the technical limitation of low recognition rate of traditional methods under non-steady-state actions and ensures the quantitative accuracy of labor behaviors with various physical characteristics.

[0018] 2. This invention constructs a motion topology coupling coefficient to physically and logically interlock the micro-motion efficiency and macro-displacement flux, and introduces statistical variance as a penalty term. This can effectively eliminate false labor data such as idling tools or aimless wandering, suppress noise interference caused by sensor hardware drift, and ensure the objectivity and authenticity of the labor validity verification results.

[0019] 3. This invention calculates the spatial coverage entropy value based on the filtered effective labor frame sequence, maps the distribution characteristics of physical work trajectories into a responsibility index, eliminates the interference of invalid wandering trajectories on the evaluation results, and enables the evaluation results to objectively reflect the uniformity of the target object's work coverage of the responsible area under effective labor conditions, providing verifiable quantitative data support for moral education evaluation. Attached Figure Description

[0020] Figure 1 This is the overall main flowchart of the labor education evaluation method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the task mode determination in an embodiment of the present invention. Figure 3 This is a flowchart of frequency domain energy integration extraction under periodic modes according to an embodiment of the present invention; Figure 4 This is a flowchart of the stiffening energy accumulation extraction process under discrete modes according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the time-domain combined acceleration waveform according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the frequency domain energy spectral density according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the microscopic spatial displacement trajectory according to an embodiment of the present invention.

[0021] Among them, 131 is the data synchronization and alignment module; 132 is the task mode determination module; 133 is the feature extraction and calculation module; 134 is the coupling verification and screening module; and 135 is the index mapping and evaluation module. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0024] See attached document Figure 1 The present invention provides a data analysis system for the effectiveness of campus labor education that integrates moral education scenarios, which may include: an inertial acquisition terminal, a spatial positioning base station, and a back-end processing server.

[0025] An inertial data acquisition terminal is worn on the end of the target object's limbs to acquire raw data on the target object's microscopic movements in real time. Spatial positioning base stations are deployed around the perimeter of the labor education area to acquire the target object's position data in two-dimensional physical space in real time. The inertial data acquisition terminal and the spatial positioning base stations transmit the acquired data sequences to the backend processing server in real time or in batches via a wireless communication protocol.

[0026] The background processing server is used to process and quantify the received multi-source heterogeneous data. The background processing server specifically includes: a data synchronization and alignment module 131, a task mode determination module 132, a feature extraction and calculation module 133, a coupling verification and screening module 134, and an index mapping and evaluation module 135.

[0027] The data synchronization and alignment module 131 performs data preprocessing operations, acquiring inertial sensor data sequences and spatial positioning data sequences, and using the timestamps of the inertial data as a reference, performing linear interpolation on the spatial positioning data to generate a synchronization state data vector. Subsequently, the data synchronization and alignment module 131 performs sliding slices on the synchronization state data vector according to a preset time length and step size to form a continuous time window sequence.

[0028] The task mode determination module 132 determines the modal branch of the task currently being performed by the target object based on the preset labor task scheduling information or the functional area attributes to which the positioning coordinates belong. The modal branches include periodic modes and discrete modes, which correspond to repetitive rhythmic tasks and sudden, non-steady-state tasks, respectively.

[0029] The feature extraction and calculation module 133 receives the modal branch determination instruction. Within each time window, the feature extraction and calculation module 133 performs feature calculation on the synchronization state data vector to generate a motion efficiency value representing the intensity of limb movements and a displacement flux value representing the physical coverage area.

[0030] In calculating the motion performance value, the system first needs to calculate the resultant acceleration modulus after removing the gravity component, and the calculation method is as follows: ; in, For a moment The magnitude of the combined acceleration; , , For a moment The original sampled values ​​of the triaxial acceleration; It is the gravitational acceleration constant; It should be noted that in all the mathematical models and calculation formulas of this invention, the time variable... Both refer to the target sampling time under the unified reference time base after being processed by the data synchronization and alignment module 131.

[0031] The coupling verification and screening module 134 constructs a motion topology coupling coefficient based on the motion efficiency value and displacement flux value output by the feature extraction and calculation module 133, combined with the statistical variance of the original data within the time window. Using a manifold constraint algorithm, this coefficient is compared with a preset verification threshold and ratio range to identify and mark valid work frames.

[0032] The indicator mapping and evaluation module 135 performs statistics on all marked valid work frames within a preset work period. This module generates a focus index by calculating the duration percentage of valid work frames and calculates information entropy using the probability distribution of valid work frames within each grid cell, thereby generating a responsibility index.

[0033] like Figure 2 As shown, the overall operation flow of the analysis method provided by this invention is as follows: Step S100: The inertial motion sequence and spatial trajectory sequence of the target object are acquired in real time by physical sensors, and the timing calibration and data slicing are completed by the data synchronization and alignment module 131.

[0034] Step S200: The task mode determination module 132 identifies the physical characteristics of the current labor and directs the calculation path to the corresponding feature extraction branch based on the identification results.

[0035] Step S300: The feature extraction and calculation module 133 calculates the motion and displacement features within each time window under the selected modal branch. For discrete tasks, this step also includes backtracking and expanding the displacement calculation window.

[0036] Step S400: The coupling verification and filtering module 134 constructs motion topology coupling coefficients, performs nonlinear logic verification, removes invalid motion data or displacement data that do not conform to physical correlation logic, and retains valid work frames.

[0037] Step S500: Based on the filtered data stream, the indicator mapping evaluation module 135 outputs the evaluation results of the labor education effect for the target object according to the preset moral education quantitative model.

[0038] The present invention provides a specific implementation of timestamp reference unification and data alignment. This process is performed by data synchronization and alignment module 131 to eliminate the asymmetry between inertial sensor data sequences and spatial positioning data sequences in terms of sampling frequency and time step.

[0039] In one specific embodiment, the inertial sensor data sequence output by the inertial acquisition terminal has a high sampling frequency, typically set between 50Hz and 100Hz, while the spatial positioning data sequence output by the spatial positioning base station has a lower sampling frequency, typically between 1Hz and 10Hz. To achieve coupled analysis of the two heterogeneous data in the same physical time slice, the data synchronization and alignment module 131 first extracts the high-frequency timestamp set from the inertial sensor data sequence as a global reference time base.

[0040] Preferably, for each discrete sampling moment in the reference time base, the data synchronization and alignment module 131 retrieves the adjacent original sampling points in the spatial positioning data sequence. The data synchronization and alignment module 131 employs a linear interpolation algorithm to calculate the estimated spatial coordinates corresponding to the target moment based on the time ratio between two original spatial positioning sampling points. The calculation formula is as follows: ; in, Time under reference time base The spatial coordinate estimation vector; For spatial positioning data sequences, the next adjacent The previous original sampling time; For spatial positioning data sequences, the next adjacent The subsequent original sampling time; To be at the original sampling time The obtained original spatial coordinate vector; To be at the original sampling time The original spatial coordinate vector obtained.

[0041] Through the above calculations, the spatial positioning data sequence is mapped onto a high-frequency reference time base. The data synchronization alignment module 131 further concatenates the triaxial acceleration values ​​at the same moment with the interpolated two-dimensional spatial coordinate values ​​to construct a synchronization state data vector, calculated as follows: ; in, For a moment The synchronization state data vector; For a moment The triaxial acceleration values; For a moment Two-dimensional spatial coordinate values; This is a vector transpose operation.

[0042] In one specific embodiment, after constructing the synchronization state data vector sequence, the data synchronization alignment module 131 uses a sliding time window technique for slicing. The system presets the length of the sliding time window and the sliding step size. For the first... Within a given time window, the data synchronization and alignment module 131 extracts all synchronization state data vectors within that time span, forming the input sample set for the subsequent feature extraction and calculation module 133. The sliding step size ensures a preset proportion of data overlap between adjacent time windows, thereby guaranteeing that dynamic features crossing window boundaries can be continuously captured during the labor activity.

[0043] Preferably, the data synchronization alignment module 131 also includes data quality verification logic. If, during the interpolation process, the time difference between two retrieved original sampling points, i.e., the difference between the subsequent sampling time and the preceding sampling time, exceeds a preset packet loss threshold, the system determines that the positioning data loss is severe during that period. At this time, the data synchronization alignment module 131 will stop the interpolation calculation for that period and mark or remove the corresponding synchronization state data vector as null to prevent invalid interpolation points caused by missing observations from interfering with subsequent coupling verification results.

[0044] This invention provides a specific implementation of signal normalization and degravity processing. This process is executed by the data synchronization alignment module 131 after timing alignment is completed, in order to eliminate the interference of static gravity components caused by the randomness of the wearing angle of the inertial acquisition terminal, and to unify the benchmark for feature extraction.

[0045] In the data processing flow, due to the random orientation of the target object wearing the inertial acquisition terminal, the output values ​​of each axis of the triaxial accelerometer are continuously affected by the gravitational acceleration component. The data synchronization and alignment module 131 extracts the triaxial acceleration components from the synchronization state data vector, and realizes the extraction of dynamic net acceleration by calculating the vector summation value of the triaxial acceleration and subtracting the standard gravitational acceleration constant.

[0046] Specifically, the data synchronization and alignment module 131 calculates the resultant acceleration magnitude using the formula mentioned earlier (i.e., the formula referenced in step S200) for the sampled data at each discrete moment. Through this calculation process, the acceleration vector in three-dimensional space is transformed into a one-dimensional resultant acceleration scalar. This resultant acceleration scalar only reflects the dynamic acceleration changes generated by the movement of the target object's limbs and is independent of the tilt angle of the inertial acquisition terminal in three-dimensional space, thereby eliminating the influence of the device wearing posture on the motion intensity quantification results.

[0047] After obtaining the resultant acceleration sequence, the data synchronization and alignment module 131 further performs signal normalization processing. This signal normalization process aims to map the numerical fluctuations caused by differences in the amplitude of motion of different target objects to a unified numerical range. In a specific embodiment, the system uses a max-min normalization algorithm to perform a linear transformation on the resultant acceleration sequence within the current time window, calculated as follows: ; in, For a moment Acceleration values ​​after normalization; For a moment The magnitude of the combined acceleration; This represents the maximum value of the resultant acceleration modulus within the current time window. This represents the minimum value of the resultant acceleration modulus within the current time window. It is a preset, extremely small positive number to prevent the denominator from being zero.

[0048] Preferably, to further suppress signal glitches caused by high-frequency vibration or hardware thermal noise, the data synchronization alignment module 131 applies a median filtering algorithm to the combined acceleration sequence after completing the gravity removal process. The system sets a sliding filter window, and replaces the center point value with the median of the values ​​within the sliding filter window. This process can filter out non-characteristic pulse interference in the acceleration sequence, ensuring that the input signal acquired by the subsequent feature extraction calculation module 133 has smooth envelope characteristics.

[0049] In one specific embodiment, the data sequence after signal normalization and degravation processing is rewritten back into the synchronization state data vector and used as the reference input for the feature extraction calculation module 133. This ensures that subsequent frequency domain analysis in periodic modes and time domain stiffness calculation in discrete modes are performed on signals with uniform range and gravity background noise eliminated, thereby improving the comparability of labor efficiency evaluation indicators among different target objects.

[0050] This invention provides a specific implementation of a sliding time window slicing strategy. This process is executed by the data synchronization and alignment module 131 after the construction of the synchronization state data vector is completed. The aim is to divide the continuous heterogeneous time series into data samples with temporal locality, so as to provide a calculation benchmark for the subsequent feature extraction calculation module 133.

[0051] In the data synchronization and alignment module 131, the sequence of synchronized state data vectors after time-series calibration and normalization is treated as a continuous matrix space. To capture the dynamic evolution characteristics of the target object during the labor process, the system uses fixed-length, overlapping sliding windows to traverse and extract the sequence. The set of data points contained in each sliding window constitutes an independent analysis frame, ensuring that subsequent motion efficiency calculations and displacement flux calculations are performed within the same time domain boundary.

[0052] In one specific embodiment, the slicing process of the sliding time window is jointly controlled by a window length parameter and a sliding step size parameter. The window length determines the duration of historical observations upon which each feature extraction relies, while the sliding step size determines the frequency of data updates. The difference between the window length and the step size forms an overlapping region, which is used to ensure that labor actions that cross the window boundaries (such as a complete broom sweeping motion) can be captured by two or more consecutive windows, thereby avoiding feature loss due to hard slicing.

[0053] Specifically, for the first in the synchronization state data vector sequence The formulas for calculating the start and end indices of each time window in the original sequence are as follows: ; ; in, For the first The index of the starting sampling point of each time window in the sequence; For the first The index of the end sampling point of each time window in the sequence; The sequence number of the time window is a positive integer; This represents the total number of sampling points included within the preset time window; This is the number of sampling points corresponding to the preset window sliding step size.

[0054] Preferably, the system dynamically adjusts the task attributes output by the task mode determination module 132. The value of . For periodic modal tasks, the system sets . The physical duration of the time window is made greater than twice the preset action cycle to ensure that the spectral resolution meets the feature extraction requirements during frequency domain analysis; for discrete modal tasks, the system shortens the time window. The value of is determined and combined with the subsequent displacement backtracking mechanism to improve the sensitivity of sudden action recognition.

[0055] In one specific embodiment, after completing the slicing, the data synchronization alignment module 131 packages the synchronization state data vector within each window into a multidimensional tensor. This multidimensional tensor simultaneously contains both the acceleration component and the spatial position component within that time period, providing a physical alignment input source for the coupling verification and filtering module 134 to perform motion topology correlation analysis. By setting... The system achieves near real-time monitoring of labor status, ensuring that the generated effective labor frame sequence has logical continuity on the time axis.

[0056] This invention provides a specific implementation of the mapping between task tags and regional attributes. This process is executed by the task modality determination module 132, aiming to solve the technical problem that a single feature extraction algorithm cannot simultaneously adapt to steady-state rhythmic actions and non-steady-state sudden actions, and to realize dynamic routing of feature extraction logic.

[0057] The task modality determination module 132, serving as the pre-control unit of the feature extraction calculation module 133, has the core function of determining the mathematical model to be followed for data processing within the current time window. Internally, this module maintains a labor task attribute mapping table, which defines the correspondence between different labor types and feature extraction modalities. Specifically, the system defines continuous, repetitive tasks such as sweeping and mopping as periodic modes, which physically manifest as sinusoidal waveforms with a dominant frequency. Conversely, the system defines sudden, random tasks such as picking up trash and wiping stains as discrete modes, which physically manifest as sparsely distributed pulse waveforms on the time axis.

[0058] In one specific embodiment, the task mode determination module 132 determines the current labor task attributes through two parallel paths: the scheduling information index path and the spatial region trigger path.

[0059] In the course scheduling information index path, the task modality determination module 132 communicates with the academic affairs management database to obtain the preset labor task ID of the target object within the current time period. The task modality determination module 132 queries the attribute mapping table. If the task ID corresponds to a large-area cleaning task, a periodic modality control signal is generated; if the task ID corresponds to a fixed-point cleaning or roving retrieval task, a discrete modality control signal is generated.

[0060] Preferably, when the preset scheduling information is missing or unclear, the task modality determination module 132 initiates a spatial region trigger path. This path utilizes the spatial positioning coordinates output by the data synchronization alignment module 131 to execute a polygon fence determination algorithm. The system pre-divides the campus labor area into several functional blocks and binds a default labor attribute to each block. For example, the long corridor area is bound to the mopping attribute (periodic modality), and the flower bed or lawn area is bound to the picking attribute (discrete modality). When the spatial coordinates of the target object continuously fall into a specific functional block, the task modality determination module 132 automatically activates the feature extraction modality corresponding to that block.

[0061] Once the mode is determined, the task mode determination module 132 sends a mode switching command to the feature extraction and calculation module 133. This mode switching command logically controls the branch selection of subsequent calculation processes: if the command is for a periodic mode, the energy integration calculation unit based on frequency domain analysis is activated; if the command is for a discrete mode, the stiffening energy accumulation calculation unit based on time domain analysis is activated. This adaptive mapping mechanism based on task attributes ensures that the mathematical tools used in subsequent steps match the physical nature of the current labor behavior, avoiding feature overload or distortion caused by using steady-state analysis methods to process transient signals.

[0062] Reference Figure 3 The present invention provides a specific implementation of frequency domain energy integration technology under periodic modes. This technology is executed by feature extraction and calculation module 133 after receiving periodic mode instructions, and is mainly aimed at continuous labor tasks with rhythmic characteristics such as sweeping and mopping.

[0063] In the periodic mode, the feature extraction and calculation module 133 treats the normalized combined acceleration sequence within the current time window as a discrete time series signal. Since this type of labor action physically manifests as reciprocating limb movements, its energy is mainly concentrated in a specific low-frequency range. To quantify the effective work intensity of this type of action, the system uses Fast Fourier Transform (FFT) to map the time-domain signal to the frequency domain and generates action efficiency values ​​through spectral energy analysis.

[0064] Specifically, the feature extraction calculation module 133 first performs a calculation on the current window with a length of [missing information]. Resultant acceleration sequence Apply a window function (such as the Hanning or Hamming window) to reduce spectral leakage. Then, the system performs a Discrete Fourier Transform to obtain the signal's spectral distribution sequence, calculated as follows: ; in, Frequency Index The corresponding complex spectral coefficients; The first in the current time window The combined acceleration sampling point value; These are the preset window function coefficients; This represents the total number of sampling points within the time window. It is the imaginary unit.

[0065] After acquiring the spectral distribution, the feature extraction and calculation module 133 calculates the power spectral density at each frequency point. The system has a preset effective operating frequency range. This range is set based on the biomechanical characteristics of the human body when performing cleaning operations, and typically covers the frequency band from 0.5Hz to 3Hz to eliminate interference from high-frequency noise (such as equipment vibration) and extremely low-frequency drift (such as slow posture adjustment).

[0066] Preferably, the feature extraction calculation module 133 calculates the spectral energy density integral within the preset action frequency range and uses the integral result as the action effectiveness value for the current time window. This action effectiveness value intuitively represents the intensity of the target object's limb activity under a specific rhythm, and its calculation formula is as follows: ; in, This represents the motion performance value under periodic modes. Frequency Index The corresponding spectral magnitude; Lower limit of preset action frequency range The corresponding discrete frequency index; Upper limit of preset action frequency range The corresponding discrete frequency index.

[0067] Through the aforementioned frequency domain integration processing, the feature extraction calculation module 133 can transform the seemingly chaotic acceleration waveform into a single quantitative index. When the target object is in a normal sweeping or mopping state, its motion frequency falls within a preset range and has a stable amplitude, thereby generating a high... The calculated motion performance value is lower if the target object is stationary or in a state of random, irregular swaying. Conversely, if the target object is stationary or swaying randomly, its spectral energy will be dispersed or fall below the effective threshold, resulting in a decrease in the calculated motion performance value. This extraction method based on frequency domain energy focusing provides high signal-to-noise ratio microscopic motion feature input for subsequent coupling verification with displacement data.

[0068] Reference Figure 4 This invention provides a specific implementation of a stiffening energy accumulation technology under discrete modes. This technology is executed by the feature extraction and calculation module 133 after receiving the discrete mode instruction, and is mainly aimed at sudden and unsteady labor tasks such as picking up garbage and fixed-point wiping.

[0069] In discrete modes, the motion characteristics of the target object manifest as sparsely distributed, pulse-like abrupt changes along the time axis, lacking periodicity. In this case, simple acceleration amplitude analysis or frequency domain analysis is insufficient to effectively capture the motion characteristics. To quantify the explosive force of such transient movements, the feature extraction calculation module 133 introduces the physical quantity of jerk force. Jerk force is defined as the first derivative of acceleration with respect to time, i.e., the rate of change of acceleration. This physical quantity can sensitively reflect the physiological functional state of the human body at the instant of initiation, cessation, or change of force direction.

[0070] Specifically, the feature extraction calculation module 133 first obtains the normalized combined acceleration sequence within the current time window. Then, the system uses the finite difference method to calculate the first derivative of this acceleration sequence, generating an acceleration sequence. To eliminate the influence of high-frequency quantization noise on the differential calculation, the calculation process is usually combined with a low-pass differential filter, and its discretization calculation formula is as follows: ; in, For a moment The stiffening value; For a moment The normalized chemical acceleration value; The previous sampling time The normalized chemical acceleration value; This represents the sampling time interval.

[0071] In obtaining the stiffening sequence Subsequently, the feature extraction calculation module 133 calculates the cumulative energy of the sequence within the current time window. Since discrete labor typically consists of a series of brief exertion and retraction processes, its force intensity value fluctuates rapidly between positive and negative. The system characterizes the total amount of abrupt changes in limb movement within this time period by calculating the sum of squares or the integral of the absolute value of the force intensity. The calculation formula is as follows: ; in, This represents the motion performance value under discrete modes; For the first time window The stiffness value of each sampling point; This represents the total number of sampling points within the current time window. This represents the sampling time interval.

[0072] Preferably, the feature extraction calculation module 133 outputs the final... Before applying the value, a baseline threshold filtering operation will also be applied. Only when... The value is only retained and passed to subsequent modules when it exceeds a preset static noise baseline; otherwise, the system sets the motion efficiency value for that time window to zero. This processing method effectively suppresses the low-amplitude acceleration change rate generated by the target object in a non-labor state (such as normal walking or standing micro-movements), ensuring that the extracted motion efficiency value is only associated with effective work actions with a certain burst intensity. Through the above time-domain analysis technique, this invention can accurately quantify discrete labor behaviors that have dispersed energy in the frequency domain but abrupt changes in the time domain.

[0073] The present invention provides a specific implementation of displacement flux accumulation calculation, which is performed in parallel by feature extraction calculation module 133 when processing each sliding time window, with the aim of quantifying the length of the target object's movement trajectory in macroscopic physical space.

[0074] In the feature extraction and calculation module 133, the displacement flux value is defined as the sum of the physical paths traversed by the target object within a specific time span. This differs from simple start-end displacement vectors and reflects the traversal characteristics of the labor process. The system acquires the synchronization state data vector within the current time window and extracts a two-dimensional spatial positioning coordinate sequence after linear interpolation and alignment. This sequence consists of several discrete time and space point pairs, recording the real-time position of the target object within the labor area.

[0075] The feature extraction and calculation module 133 uses a step-accumulation method to calculate the displacement flux. Specifically, the feature extraction and calculation module 133 iterates through each sampling time within the time window, calculates the Euclidean distance between two adjacent sampling times, and sums these infinitesimal distances as a scalar. This calculation process simulates the mathematical logic of the path integral, and the calculation formula is as follows: ; in, This represents the displacement flux value within the current time window. For the first time window Two-dimensional spatial coordinates of each sampling point; For the first time window Two-dimensional spatial coordinates of each sampling point.

[0076] Preferably, to prevent signal drift or measurement errors from the positioning base station from causing artificially high displacement flux in a stationary state, the feature extraction calculation module 133 introduces micro-motion suppression logic during the cumulative calculation process. The system presets a micro-displacement threshold, which is set based on the average measurement accuracy of the positioning technology (e.g., 10-20 cm for UWB technology). When calculating the Euclidean distance between two adjacent points, if the calculated result is less than the threshold, the system determines that the position change is sensor noise, forcibly sets the corresponding micro-element distance to zero, and excludes it from the total flux.

[0077] Through the above calculation logic, the displacement flux value obtained by the system can objectively characterize the running amount or patrol coverage of the target object within a unit time window. When the target object performs large-scale mopping or patrol and picking tasks, its trajectory points show a continuous extension distribution in space, thus generating a high displacement flux value; while when the target object is resting in place or only moving within a very small area, even if there is motion efficiency value (such as waving in place), its displacement flux value will remain at a very low level. This physical quantity provides a macroscopic verification basis for the subsequent construction of motion topology coupling coefficients.

[0078] This invention provides a specific implementation of a forward backtracking extension mechanism for discrete tasks. This mechanism is triggered by the feature extraction calculation module 133 when it determines that the current task mode is a discrete mode, aiming to solve the asynchronous problem of physical displacement and limb movements on the time axis in discontinuous labor.

[0079] When performing discrete labor tasks such as picking up and wiping, the behavior of the target object typically exhibits a sequential characteristic of movement, pause, and operation. That is, the target object first reaches the work point through physical displacement (resulting in high displacement throughput and low motion efficiency), then pauses to perform manual operations (resulting in high motion efficiency and low displacement throughput). If the system calculates both simultaneously within a fixed, short sliding time window, the motion and displacement characteristics may become misaligned in time, leading to reduced coupling within the single window and resulting in misjudgment as ineffective labor.

[0080] To address the aforementioned technical issues, the feature extraction calculation module 133 introduces a forward backtracking expansion strategy when calculating displacement flux. Specifically, when the task modality determination module 132 issues a discrete modality command, the feature extraction calculation module 133 does not directly use the time boundary of the current action analysis window. Instead, the feature extraction and calculation module 133 extends the spatial data extraction window towards the historical timeline based on a preset backtracking factor.

[0081] In one specific embodiment, the starting point of the expanded spatial data calculation window is shifted forward by a preset time offset from the starting point of the current window. Feature extraction calculation module 133 in the extended time range The system obtains a spatial positioning coordinate sequence and calculates the cumulative Euclidean distance of the extended sequence.

[0082] Preferably, to prevent irrelevant trajectory interference from excessively long backtracking times, the system introduces a time-decay weighted mechanism. When calculating the displacement flux within the extended window, the system assigns different weights to displacement elements within different time periods. Displacements closer to the current action occurrence time have higher weights; displacements farther away have their weights decay exponentially. The formula for calculating the weighted displacement flux is as follows: ; in, This is the displacement flux value after forward backtracking expansion; To extend the time window The total number of sampling points contained therein; For the first in the extended window Two-dimensional spatial coordinates of each sampling point; For the first The time weighting coefficient corresponding to each sampling point, this coefficient approaches the time of sampling. And it increases monotonically.

[0083] Through the above mechanism, while keeping the motion efficiency calculation window (strictly corresponding to the current limb burst moment) unchanged, the system logically incorporates the displacement corresponding to the approach process required to generate the motion into the same evaluation unit. This ensures that even in a stop-and-go work mode, effective labor behavior can form a strong coupling match between high motion efficiency and high displacement flux in the feature space, avoiding feature fragmentation caused by time slicing.

[0084] This invention provides a specific implementation method for constructing coupling coefficients. This process is executed by the coupling verification and screening module 134, which aims to physically and logically interlock the micro-dimensional limb movement features with the macro-dimensional spatial displacement features, thereby quantifying the true effectiveness of labor behavior.

[0085] The coupling verification and screening module 134 receives the action performance value (recorded according to the current modality) output by the feature extraction and calculation module 133. or Collectively referred to as ) and displacement flux value (denoted as or Collectively referred to as Since motion efficiency and displacement flux belong to different physical dimensions and have significantly different numerical ranges, the system first standardizes these two input variables, mapping them to a dimensionless unit interval. .

[0086] After obtaining the standardized feature variables, the coupling verification and screening module 134 constructs the motion topology coupling coefficient. This motion topology coupling coefficient is designed according to the dual-response principle, meaning that a high value is only observed when the target object simultaneously exhibits both limb movement intensity and corresponding spatial coverage. To suppress abnormally high values ​​in a single dimension (such as extremely high movement values ​​generated by vigorous waving in place, or spurious displacement values ​​generated by positioning drift), the system introduces a nonlinear product term. Furthermore, to further eliminate random noise interference from the sensor itself, the coupling verification and screening module 134 introduces statistical variance as a penalty factor in the denominator.

[0087] Specifically, the formula for calculating the motion topology coupling coefficient is as follows: ; in, The motion topology coupling coefficients within the current time window; This represents the standardized motion performance value. The displacement flux value is the standardized value. This represents the statistical variance of the combined acceleration sequence within the current time window, used to characterize the instability noise of micro-motions. The statistical variance of the positioning coordinate sequence within the current time window is used to characterize the jitter noise of macroscopic positioning. This is a preset penalty weighting factor used to adjust the degree to which noise suppresses the coupling coefficient.

[0088] Preferably, the standardization process uses the Sigmoid function or linear extremum mapping to ensure that the input variables... and Non-negative. The coupling coefficient constructed using the above formula. It has physical filtering characteristics: when the target object is in a stationary, idling state, although... High, but Approaching zero, the numerator becomes extremely small, and the overall... The value decreases; when the target object is in an invalid wandering state, although High, but Lower, also leading to The value is limited. Furthermore, if the sensor data exhibits high-frequency oscillations or drift, the variance term in the denominator will be affected. It will increase, and further compress. The value is used to automatically suppress low-quality or fake labor data.

[0089] This invention provides a specific implementation of noise suppression using statistical dispersion. This process is executed synchronously by the coupling verification and screening module 134 during the calculation of motion topology coupling coefficients. The aim is to quantify and suppress signal noise caused by hardware drift, environmental multipath effects, or uncontrolled mechanical vibration, and prevent it from interfering with the determination of valid working frames.

[0090] In IoT sensing environments, inertial acquisition terminals may experience high-frequency uncontrolled vibrations due to collisions, and spatial positioning base stations may experience coordinate drift due to signal obstruction. These two types of noise often manifest as high-amplitude, transient changes, which can easily be misinterpreted as high-intensity labor if calculated solely based on amplitude or flux. The coupling verification and screening module 134 introduces statistical dispersion analysis, constructing a penalty term in the denominator of the coupling coefficient by calculating the signal's fluctuation stability within a sliding time window.

[0091] Specifically, the coupling verification and screening module 134 first calculates the statistical dispersion of the resultant acceleration sequence within the current time window, i.e., the acceleration noise variance. The system treats the resultant acceleration sequence within the window as a random variable sample set and calculates its deviation from the local mean. To more accurately separate effective motion (low-frequency component) from vibration noise (high-frequency component), the system preferentially uses the residual variance method, calculated as follows: ; in, The variance of acceleration noise within the current time window; For a moment The magnitude of the combined acceleration; For a moment The baseline value of the combined acceleration after being filtered by the moving average.

[0092] Simultaneously, the coupling verification and filtering module 134 calculates the statistical dispersion of the spatial positioning sequence within the current time window, i.e., the positioning drift variance. Considering that positioning drift typically manifests as disordered jumps around a certain point, the system calculates the dispersion of the original coordinate point relative to its fitted trajectory line, using the following formula: ; in, This represents the location drift variance within the current time window. For a moment The original two-dimensional spatial positioning coordinates; For a moment The corresponding trajectory coordinates after linear fitting or spline interpolation using the least squares method.

[0093] Preferably, the coupling verification and screening module 134 will calculate the... and Substitute this into the denominator of the aforementioned formula for the motion topology coupling coefficient. When sensor drift exists in the monitoring data source (leading to...) (A dramatic increase) or mechanical vibrations outside the normal physiological range of the human body (leading to) When the number of quotients increases dramatically, the denominator increases, thus forcibly lowering the final coupling coefficient. .

[0094] This suppression mechanism leverages the physical characteristics of smooth motion and continuous trajectory in effective labor activities, while noise exhibits random and discrete characteristics. Through the penalty effect of statistical variance, the system can effectively filter out pseudo-high-energy data frames generated by non-human factors such as equipment loosening and signal interference, while retaining genuine large-amplitude labor movements (such as the regular arm swinging during sweeping), thus improving the robustness of feature verification.

[0095] This invention provides a specific implementation of the logic for determining the effective labor manifold space. This process is executed by the coupled verification and screening module 134, which aims to delineate the logical boundary of effective labor in the multidimensional feature space based on the aforementioned calculated physical indicators, thereby realizing the binary classification of labor behavior.

[0096] The coupling verification and filtering module 134 constructs a two-dimensional verification plane, where the horizontal axis represents the motion efficiency value and the vertical axis represents the displacement flux value. On this plane, the system defines an effective region constrained by the motion topology coupling coefficient and the energy-displacement ratio, called the labor manifold space. Only when the data point corresponding to the time window falls within this closed region is the system considered the time window as a valid labor frame.

[0097] Specifically, the decision logic includes constraints in two dimensions: The first dimension is the intensity coupling constraint. The system compares the calculated motion topology coupling coefficient with a preset coupling threshold. This constraint is used to eliminate low-intensity invalid movements, such as unconscious limb swinging or slow walking.

[0098] The second dimension is the form ratio constraint. The system calculates the ratio of displacement flux value to motion efficiency value within the current time window, denoted as the motion displacement ratio. This ratio reflects the proportional relationship between the work done by the limbs and spatial movement. For a specific labor task (such as sweeping), there exists a fixed biomechanically consistent range between the frequency of limb movement and the speed of body movement. A value that is too large indicates that the target object is moving quickly but with weak hand movements (likely simply walking); if... A value that is too small indicates that the target object's hand movements are violent but its position remains essentially unchanged (likely a stationary hand gesture). The coupled verification and filtering module 134 checks whether this ratio is within the preset valid range. The internal verification logic is as follows: ; in, This represents the displacement flux value for the current time window; This represents the action performance value for the current time window. This is the preset lower limit of the motion displacement ratio; This is the preset upper limit of the motion displacement ratio.

[0099] Preferred ratio range The system adaptively adjusts the range based on the task type output by the task modality determination module 132. For periodic cleaning tasks, the range is set to a medium range to match the gait characteristics of sweeping while walking; for discrete picking tasks, the system allows a wider range to accommodate the alternating behavior patterns of moving to find and stopping to pick up.

[0100] The coupling verification and filtering module 134 integrates the verification results from the two dimensions mentioned above to generate a binary judgment identifier. Only when a data frame simultaneously satisfies both the strength coupling constraint and the morphological ratio constraint is the system marked as 1 (valid labor frame); otherwise, it is marked as 0 (invalid frame). The system then outputs a Boolean sequence of 0s and 1s, serving as the basis for the subsequent indicator mapping and evaluation module 135 to perform duration statistics and entropy calculations. Through this dual constraint mechanism, the present invention accurately locks down data clusters that conform to the physical characteristics of real labor within the feature space, eliminating pseudo-labor data that, although high in energy, has an abnormal morphology.

[0101] This invention provides a specific implementation method for calculating labor saturation and concentration indicators. This process is executed by the indicator mapping evaluation module 135, which aims to transform the underlying data frame sequence into a quantitative score that intuitively reflects the student's labor status.

[0102] The indicator mapping evaluation module 135 first receives a binary sequence of valid labor frames output by the coupled verification and filtering module 134. This sequence is a time-ordered Boolean set that records whether each time slice within the entire task cycle is determined to be valid labor. In order to evaluate the quality of labor from a macro perspective, the system first calculates two basic dimensional indicators based on this sequence: saturation, which represents the total amount of labor, and focus, which represents the continuity of labor.

[0103] Labor saturation calculation: Labor saturation reflects the proportion of effective labor actually invested by the target object within the specified task time. The indicator mapping evaluation module 135 performs statistical integration on the effective markers in the binary sequence to obtain the cumulative effective time, and compares it with the preset task scheduling time.

[0104] Specifically, the system iterates through the binary sequence and accumulates the durations of all time windows marked as 1. Considering the overlapping nature of the sliding time windows, the system employs a step-size integration method to avoid redundant calculations. Subsequently, the system introduces a duration scaling function, calculated as follows: ; in, Rate the level of labor saturation; For the first The binary determination result for each time window (1 for valid, 0 for invalid); The time interval corresponding to the sliding step size of the sliding time window; This represents the total number of windows within the task cycle. The total scheduled class time for this labor task is preset for the academic affairs system; A preset compliance factor (e.g., 0.8) is used to tolerate reasonable rest or preparation time.

[0105] Labor focus calculation reflects the stability of the target's state during the labor process. If the target frequently switches between working and idle states during the labor process (i.e., exhibiting dawdling or inattentiveness), its binary sequence will show high-frequency jump characteristics. The index mapping evaluation module 135 quantifies focus by analyzing the state reversal frequency of the binary sequence.

[0106] Specifically, the system calculates the number of transitions between adjacent states in the sequence (i.e., the number of transition edges from 0 to 1 or 1 to 0). A higher number of transitions indicates a more fragmented work process and a lower focus score. The system constructs a focus model based on inverse decay, with the calculation formula as follows: ; in, Rate work focus level; For the first Binary determination results for each time window; For the first Binary determination results for each time window; This represents the total number of windows within the task cycle. The preset fragmentation tolerance factor is used to adjust the tolerance for normal intervals (such as changing water or wiping sweat).

[0107] Preferably, to prevent short-term misjudgment fluctuations from affecting attention calculation, the indicator mapping evaluation module 135 performs a morphological closing operation on the binary sequence before executing the above formula. That is, it first dilates and then erodes, filling in extremely short invalid gaps in the sequence with a length less than a preset threshold (such as 5 seconds), treating them as part of continuous work, thereby ensuring that the attention indicator focuses more on the evaluation of long-term behavioral patterns.

[0108] Through the above calculations, the system can distinguish between two completely different work states: efficient continuous work and intermittent perfunctory work. Even if the total effective time of the two may be similar, their focus scores will show differences, thus achieving a refined profile of the quality of the work process.

[0109] This invention provides a specific implementation method for quantifying labor responsibility using information entropy. This process is executed by the indicator mapping evaluation module 135, aiming to solve the technical defect of traditional evaluation methods that only focus on whether something has been done (duration) while ignoring whether it has been done completely (coverage).

[0110] In the labor education evaluation system, responsibility is defined at the physical level as the comprehensiveness and uniformity of the target object's coverage of a designated labor area. If the target object only repeats work in a local area (e.g., only sweeping a tile at their feet), although their action efficiency value and cumulative time may meet the standards, it is considered perfunctory behavior from the perspective of labor effectiveness. In order to identify such spatial laziness or opportunistic behavior, the indicator mapping evaluation module 135 introduces the Shannon Entropy model to transform the spatial distribution dispersion of the labor trajectory into a quantifiable indicator of responsibility.

[0111] Specifically, the indicator mapping evaluation module 135 first performs grid processing on the preset labor area, dividing it into... The system then traverses all valid labor frames output by the coupling verification and filtering module 134, extracts the spatial positioning coordinates corresponding to each frame, and maps them to the corresponding grid cells. The system counts the number of valid labor frames falling into each grid cell and calculates the proportion of labor density in that grid to the total valid labor, thereby constructing a spatial distribution probability model.

[0112] After obtaining the probability distribution of each grid cell, the system calculates the spatial disorder (i.e., uniformity) of labor behavior using the information entropy formula. According to information theory principles, the system's entropy is maximized when labor trajectories are uniformly distributed across all grids; and minimized when labor trajectories are highly concentrated in a few grids. The calculation formula is as follows: ; in, The spatial distribution entropy value of the current labor task; The total number of effective grid cells used to divide the labor area; For the first The proportion of effective labor frames accumulated within a single grid cell to the total effective labor frames. It is a very small positive number, used to prevent Logarithmic calculations are meaningless when the logarithm is zero.

[0113] Preferably, in order to convert the abstract entropy value into an intuitive percentage score, the indicator mapping evaluation module 135 will calculate the... The value is compared to the theoretical maximum entropy after normalization. Corresponding to the target object in all The case where exactly the same amount of labor is invested in each grid (at this time) The final formula for calculating the responsibility evaluation index is as follows: ; in, Rate your sense of responsibility for your work; This represents the actual calculated spatial distribution entropy value; This represents the theoretical maximum entropy value under the preset region grid division.

[0114] Based on the above logic, this evaluation index possesses a keen ability to identify spatial distribution. For example, considering two students who both worked for 20 minutes, student A traversed the entire classroom to clean, and their trajectory points were scattered across various grids. The calculated [indicator / indicator]... The distribution is uniform, therefore near Student A received a very high score for responsibility; while Student B merely swept in place in a corner, with their trajectory points highly concentrated on a single grid, resulting in that grid being... The value approaches 1 while others are 0, and the calculated value is... The score approaches zero, resulting in an extremely low responsibility rating. This evaluation mechanism, based on the principle of physical entropy reduction, fundamentally eliminates data manipulation through half-hearted efforts, ensuring that the evaluation results truly reflect the extent to which workers fulfill their responsibilities for regional jurisdiction.

[0115] This section will use a specific campus labor scenario to fully demonstrate the entire process from data collection to evaluation output.

[0116] Scene definition and parameter initialization: Scenario Description: From 4:00 PM to 4:20 PM in a middle school, student A (wearing a smart bracelet) is mopping the floor in the corridor on the second floor of the teaching building. This area is divided into 20 grid cells numbered Grid_01 to Grid_20.

[0117] System parameter settings: Sampling frequency = 50Hz; sliding time window length = 256 (approximately 5.12 seconds); step size = 128 (50% overlap); task mode: determined by the ID issued by the academic affairs system as a periodic mode; effective action frequency range: [0.5Hz, 3.0Hz].

[0118] See attached document Figure 5 - Appendix Figure 7 Step 1: Signal slicing and frequency domain extraction At that time, the system intercepted the first Data for each time window.

[0119] Time-domain performance: The original acceleration signal exhibits sinusoidal fluctuation characteristics in the Y-axis direction (arm swing direction), with a peak value of approximately 1.2g.

[0120] Frequency domain analysis: After FFT transformation, the spectrum shows that the energy is highly concentrated at 1.1 Hz, which corresponds to student A's mop pushing and pulling motion of about 1.1 times per second.

[0121] Calculation result: The motion efficiency value calculated by integration is 0.85 (after normalization).

[0122] Step 2: Displacement flux and coupling verification Displacement data: Synchronously captured UWB positioning data shows that student A's trajectory coordinates moved from (12.5, 4.0) to (14.2, 4.2) within 5 seconds.

[0123] Flux calculation: After fretting suppression, the calculated displacement flux is 1.7m.

[0124] Coupling calculation: Substitute the system into the coupling coefficient formula. Due to stable motion and smooth trajectory, the penalty term in the denominator approaches 1.

[0125] ; Judgment result: (Set to 0.3), and motion displacement ratio The frame is within a reasonable range. This frame is determined to be a valid work frame.

[0126] Counterexample: If student A stands still and chats with someone without consciously waving their arms: The motion performance value may still be 0.4 (with motion energy).

[0127] However, the displacement flux dropped sharply to 0.1m (basically no displacement).

[0128] Calculated If the frame is below the threshold, the system determines it as an invalid frame, effectively preventing the fake work from being done.

[0129] Final evaluation output: After the task is completed, the indicator mapping evaluation module 135 outputs: Saturation: The effective duration is 18 minutes, accounting for 90% of the total duration of 20 minutes, and the score is 95 points.

[0130] Responsibility (Entropy): The trajectory covers 18 grids, and the calculated spatial distribution entropy is close to the theoretical maximum value, with a score of 92.

[0131] Final assessment: Work performed with ease and comprehensive coverage; excellent.

Claims

1. A data analysis method for the effectiveness of campus labor education integrated with moral education scenarios, characterized in that, Includes the following steps: Acquire inertial sensor data sequences and spatial positioning data sequences of the target object within a preset working period; Based on the attributes of the labor tasks performed by the target object, a feature extraction mode is determined, which includes periodic modes and discrete modes; Based on the aforementioned feature extraction mode, adaptive feature extraction is performed on the inertial sensor data sequence to generate motion performance values; Based on the spatial positioning data sequence, calculate the displacement flux value of the target object within the corresponding time window; Construct motion topology coupling coefficients, which characterize the degree of correlation between the motion efficiency value and the displacement flux value; The motion topology coupling coefficient is compared with the preset valid labor verification conditions to filter out valid labor frames; Based on the selected effective work frames, moral education effectiveness evaluation results are generated, including indicators of focus or sense of responsibility.

2. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The step of generating motion performance values ​​includes: When the labor task attribute corresponds to continuous sweeping or mopping, it is determined to be a periodic mode, and the action efficiency value is extracted using frequency domain analysis. When the labor task attribute corresponds to a sudden operation such as picking up or wiping, it is determined to be a discrete mode, and the action efficiency value is extracted using the time domain analysis method.

3. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 2, characterized in that, The steps for extracting motion performance values ​​using frequency domain analysis specifically include: Calculate the resultant acceleration sequence after removing the gravitational component; Perform a Fast Fourier Transform on the combined acceleration sequence to obtain the spectral distribution; Select a preset action frequency range, calculate the spectral energy density integral within the preset action frequency range, and use the integral result as the action performance value.

4. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 2, characterized in that, The steps for extracting motion performance values ​​using time-domain analysis methods specifically include: Calculate the resultant acceleration sequence after removing the gravitational component; The first derivative of the resultant acceleration sequence is obtained to obtain the stiffness sequence, which represents the rate of change of acceleration; The cumulative energy of the stiffness sequence within the time window is calculated, and the cumulative result is used as the action effectiveness value.

5. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The step of calculating the displacement flux value of the target object within the corresponding time window specifically includes: When it is determined to be a discrete mode, the calculation range of the current time window is extended backward, and the spatial positioning data sequence is extracted by applying the extended time window; The cumulative Euclidean distance of the spatial positioning coordinates within the extended time window is calculated as the displacement flux value.

6. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The steps for constructing motion topology coupling coefficients specifically include: Calculate the statistical variance of the inertial sensor data sequence and the statistical variance of the spatial positioning data sequence within the current time window; The product of the action performance value and the displacement flux value is taken as the main term, and the statistical variance of the inertial sensor data sequence and the statistical variance of the spatial positioning data sequence are taken as the penalty term. The motion topology coupling coefficient is constructed such that it is positively correlated with the main term and negatively correlated with the penalty term. The motion topology coupling coefficient is used to suppress noise interference caused by data dispersion.

7. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The step of filtering out valid labor frames specifically includes: The ratio of the motion efficiency value to the displacement flux value is calculated using a manifold constraint algorithm. Determine whether the motion topology coupling coefficient is greater than a preset minimum threshold, and simultaneously determine whether the ratio falls within a preset valid range; The current time window is marked as a valid labor frame if and only if both of the above conditions are met simultaneously.

8. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The steps for generating moral education effectiveness evaluation results that include attention index or responsibility index include: The total duration of the time window marked as a valid labor frame is recorded. Calculate the ratio of the total duration of the effective work frame time window to the total duration of the preset work period, and map the ratio to a focus index.

9. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, The step of generating moral education effectiveness evaluation results that include focus indicators or responsibility indicators also includes: Obtain the set of spatial positioning coordinates corresponding to all valid labor frames; The preset work area is divided into several grid units, and the probability distribution of the set of spatial positioning coordinate points falling into each grid unit is statistically analyzed. The information entropy value is calculated based on the probability distribution, and the information entropy value is mapped to a responsibility index, wherein the higher the information entropy value, the more uniform the regional coverage.

10. The method for analyzing the effectiveness of campus labor education in a moral education context according to claim 1, characterized in that, After acquiring the inertial sensor data sequence and spatial positioning data sequence of the target object within a preset working period, the method further includes: Using the sampling timestamps of the inertial sensor data sequences as a reference, linear interpolation is performed on spatial positioning data sequences with low sampling frequencies. Generate a time-aligned synchronization state data vector, and perform sliding slicing processing on the synchronization state data vector according to a preset step size.

Images