Multifunctional body characteristic monitoring method and system based on intelligent wearable device
By combining multi-source physiological data fusion algorithms and dynamic calibration models with support vector machine algorithms and intelligent data processing scheduling algorithms, the problem of insufficient monitoring accuracy of smart wearable devices under motion interference is solved, achieving efficient and safe physiological data monitoring and personalized guidance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN BRANDSOUND TECH CO LTD
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-10
AI Technical Summary
Existing smart wearable devices are easily affected by motion interference, have insufficient accuracy and stability in blood pressure monitoring, lack the ability to monitor biochemical indicators, have insufficient algorithm intelligence, and suffer from problems such as transmission delay, privacy leakage risk, and low computing efficiency.
A multi-source physiological data fusion algorithm and a dynamic calibration model are used, combined with a support vector machine algorithm for motion pattern classification and anomaly alerts. A data processing intelligent scheduling algorithm optimizes computational efficiency, and an encrypted transmission verification algorithm ensures data security.
It improves the accuracy of heart rate, blood glucose, and blood pressure monitoring, optimizes the recognition of arrhythmia features, realizes intelligent prompts and personalized guidance for physiological data, enhances the ability to recognize exercise patterns, and ensures data security and user experience adaptability in different scenarios.
Smart Images
Figure CN121817824A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of intelligent wearable devices and health monitoring, and particularly relates to a multifunctional body feature monitoring method and system based on an intelligent wearable device. BACKGROUND
[0002] Intelligent wearable devices provide a "portable, continuous, non-invasive" collection portal for health monitoring, which gives intelligent wearable devices core practical value, and together realizes a closed loop of "daily monitoring-risk early warning-health management", extending health management from a hospital scene to a life scene.
[0003] However, the prior art has the following disadvantages: susceptible to motion interference and non-invasive biochemistry monitoring, insufficient accuracy and stability of blood pressure monitoring; lack of key biochemical indicator monitoring capability, insufficient algorithm intelligence, no health risk prediction and personalized recommendations; transmission delay, privacy leakage risk and low algorithm operation efficiency.
[0004] Therefore, a new method is urgently needed. SUMMARY
[0005] The purpose of the present application is to provide a multifunctional body feature monitoring method and system based on an intelligent wearable device, which is based on a multi-source physiological data fusion algorithm and a dynamic calibration model to improve the accuracy of heart rate, blood glucose and blood pressure related data monitoring and optimize the recognition effect of arrhythmia features, and simultaneously realizes intelligent prompting and personalized adaptive guidance of physiological data abnormalities through a multi-dimensional physiological parameter intelligent analysis algorithm, completes the recognition of multiple motion modes relying on a motion feature extraction and classification algorithm, in addition, adopts a data processing intelligent scheduling algorithm to optimize operation efficiency, a cryptographic transmission verification algorithm to guarantee data security, and an algorithm adaptation strategy to improve experience adaptability in different use scenarios.
[0006] To achieve the above purpose, the present application provides a multifunctional body feature monitoring method and system based on an intelligent wearable device, comprising the following steps: S1, collecting multi-dimensional body feature data of a human body through a sensor of an intelligent wearable device, performing weighted fusion and dynamic calibration processing on the collected data to obtain calibrated accurate physiological data and feature recognition results; S2, receiving the accurate physiological data and feature recognition results output by S1, performing body feature abnormality recognition based on the data through an abnormality judgment model, and simultaneously realizing motion mode classification using a support vector machine algorithm to obtain abnormality prompt information, personalized guidance scheme and motion mode recognition results; S3 receives the abnormal prompts, personalized guidance schemes and motion pattern recognition results output by S2, uses a data processing intelligent scheduling algorithm to allocate and schedule resources for the tasks of S1 and S2, optimizes task processing efficiency through objective function, and ensures data security with encrypted transmission verification algorithm to obtain efficient and secure final data results. S4. Receive the efficient and secure final data result output by S3, extract the scene feature vector, perform algorithm adaptation optimization based on the mapping relationship between the scene feature vector and the parameter set of the preceding core algorithm, and use root mean square error to evaluate the adaptation effect.
[0007] Preferably, in S1, the fusion processing and dynamic calibration specifically include: The multi-source raw physiological data, including heart rate data, blood pressure data, exercise acceleration data, and body temperature data, were denoised. The preprocessed data from multiple sources of raw physiological data are integrated using a weighted fusion algorithm. The calculation formula is as follows: ; In the formula, For the first The data after road preprocessing To integrate weights, satisfy ; An error correction model is established based on historical calibration data, and the fusion results are adjusted in real time. The calibration formula is as follows: ; In the formula, For reference benchmark value, For adaptive calibration coefficients, the range of values is... .
[0008] Preferably, in S2, the anomaly detection model is based on the mean of the data. and standard deviation Build, when data deviates The interval is considered abnormal; The classification function of the support vector machine is: ; In the formula, The motion feature vector to be classified; For kernel functions; For bias terms; Label the sports mode category; It is a Lagrange multiplier.
[0009] Preferably, in S3, the objective function of the intelligent data processing scheduling algorithm is: ; In the formula, For the number of data processing tasks, For the first The weight coefficient of each task. For task processing time; The task weight coefficient corresponds to the data type priority, including exception message data. Personalized guidance data Motion recognition data .
[0010] Preferably, the constraints of the intelligent scheduling algorithm include: ; ; ; In the formula, This is the maximum tolerance time for a single round of data processing, with a default value of 500ms. For the first Minimum processing time for each task; The task consumes computing resources; Total computing resources for the device.
[0011] Preferably, in S4, the parameter set is represented as follows: ; In the formula, The fusion weights are the fusion weights of the weighted fusion algorithm; The mean and standard deviation of the anomaly detection model; These are the model parameters for the Support Vector Machine algorithm; These are the adaptive calibration coefficients for the dynamic calibration model; For mapping weights; This is the bias vector.
[0012] Preferably, the mapping relationship is expressed as follows: ; In the formula, for dimensional weight matrix, for Number of parameters For scene feature dimensions; It is the bias vector; Triggered only when the change in scene features is not less than a preset value. Adjustment, represented as: ; In the formula, This represents the change in the scene feature vector; This represents the total number of dimensions of the scene features; For the first The current new value of each scene feature; For the first Historical old values of scene features; The preset threshold for the amount of change, ; The adaptation benchmark value of RMSE The average value of monitoring data from 300 sets of standard laboratory equipment in the same scenario is calculated using the following formula: ; In the formula, To verify the data sample size, The algorithm after parameter adjustment One output value, These are the adaptation baseline values for the corresponding scenarios.
[0013] The present invention also provides a multifunctional body feature monitoring system based on a smart wearable device, comprising: The data fusion calibration module is used to collect multi-dimensional human body feature data through sensors; The intelligent analysis and recognition module is connected to the data fusion and calibration module and is used to perform weighted fusion and dynamic calibration on the collected data, and output accurate physiological data and feature recognition results. The data optimization and security module is connected to the intelligent analysis and recognition module. It is used to receive the output data of the fusion calibration module, identify abnormal body features based on the anomaly judgment model, classify movement patterns through the SVM algorithm, and output anomaly prompt information, personalized guidance plan and movement pattern recognition results. An anomaly identification module, connected to the data optimization and security module, is used to receive the output data of the fusion calibration module, identify abnormal body features based on the anomaly judgment model, classify movement patterns through the SVM algorithm, and output anomaly prompts, personalized guidance plans, and movement pattern identification results. The intelligent scheduling module, connected to the anomaly identification module, is used to receive the output data of the anomaly identification module, allocate computing resources and schedule processing tasks using an intelligent scheduling algorithm, and ensure data security with an encrypted transmission verification algorithm, and output efficient and secure final data results. The algorithm adaptation module, connected to the intelligent scheduling module, is used to receive the output data of the intelligent scheduling module, extract scene feature vectors, map relationships to optimize the parameter set, and evaluate the adaptation effect through RMSE.
[0014] Therefore, the present invention employs the above-mentioned multifunctional body feature monitoring method and system based on smart wearable devices. Compared with the prior art, the technical solution of the present invention has the following beneficial effects: (1) Based on the multi-source physiological data fusion algorithm and dynamic calibration model, improve the accuracy of monitoring heart rate, blood glucose and blood pressure related data, and optimize the identification effect of arrhythmia characteristics; (2) Through multi-dimensional physiological parameter intelligent analysis algorithm, intelligent prompts and personalized adaptation guidance for abnormal physiological data are realized, and the identification of various movement modes is completed by relying on motion feature extraction and classification algorithm; (3) The data processing intelligent scheduling algorithm is used to optimize the computing efficiency, and the encrypted transmission verification algorithm is used to ensure data security. The algorithm adaptation strategy is used to improve the experience adaptability in different usage scenarios.
[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0016] Figure 1 This is a flowchart of an embodiment of a multifunctional body feature monitoring method and system based on a smart wearable device according to the present invention. Detailed Implementation
[0017] 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. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used in the present invention should have the ordinary meaning understood by those skilled in the art.
[0018] Example 1 like Figure 1 As shown, this embodiment provides a multifunctional body feature monitoring method and system based on smart wearable devices. It should be understood that the specific parameters, models and protocols mentioned in this embodiment are merely examples to help those skilled in the art understand the present invention, and are not intended to limit the present invention.
[0019] The present invention provides a multifunctional body feature monitoring method and system based on a smart wearable device, comprising the following steps: S1. Collect multi-dimensional human body feature data through sensors of smart wearable devices, perform weighted fusion and dynamic calibration on the collected data, and obtain accurate physiological data and feature recognition results after calibration. In this step, the fusion processing and dynamic calibration of the collected raw physiological data specifically involve: For multi-source raw physiological data, denoted as , To reduce noise in the data sources, a moving average filter is used to eliminate random interference. The preprocessed data is integrated using a weighted fusion algorithm, and the calculation formula is as follows: ; In the formula, For the first The data after road preprocessing To integrate weights, satisfy The allocation is dynamically determined by the data's credibility. An error correction model is established based on historical calibration data, and the fusion results are adjusted in real time. The calibration formula is as follows: ; In the formula, For reference benchmark value, For adaptive calibration coefficients, the range of values is... It updates dynamically with data fluctuations; Time-domain and frequency-domain features were extracted from the calibrated data to obtain arrhythmia-related feature vectors. , For feature dimensions; Obtaining precise physiological data (Heart rate / blood glucose / blood pressure related), arrhythmia feature vector ; S2 receives the accurate physiological data and feature recognition results output by S1, and uses the anomaly judgment model to identify abnormal body features based on the data. At the same time, the support vector machine algorithm is used to classify the movement pattern to obtain abnormal prompt information, personalized guidance plan and movement pattern recognition results. In this step, the receiver of S1 and By standardizing the data scale through feature normalization, the changing trend characteristics of physiological parameters are calculated. ,in, For trend characteristics; An anomaly detection model is built based on statistical learning algorithms to detect data deviations. When the interval is considered abnormal, the data anomaly is determined by the following formula: ; In the formula, For the first The mean of a trend-like characteristic; The standard deviation is denoted as ; when An error message is triggered at any time. An adaptive threshold; Based on anomaly Based on user historical data, a rule-based reasoning algorithm is used to output adaptation guidance suggestions, specifically: Historical data, including users' physiological data, abnormal records, and guidance execution feedback from the past 30 days, was extracted. Invalid values were removed through data cleaning, and data dimensions were unified using normalization to obtain a standardized historical feature vector. ,in, Historical characteristics include anomaly frequency, feedback satisfaction, and compliance. Based on medical knowledge, health management guidelines, and user feedback, an IF-THEN structured rule base was established. Rules are graded according to their degree of abnormality. Example rules are as follows: Low anomaly rule: If And historical characteristics (Frequency of movement) If the frequency is calculated per week, the output will be "It is recommended to increase moderate-intensity exercise by 30 minutes daily and monitor changes in indicators regularly". Medium anomaly rule: If And historical characteristics (Sleep duration) If the time limit is 3 hours, the output will be "Prioritize ensuring 7-8 hours of sleep per day, reduce staying up late, and recheck relevant indicators in 3 days"; High anomaly rule: If And historical characteristics (Number of abnormal repetitions) If the error occurs again, the output will be "It is recommended to seek medical attention promptly, suspend high-intensity exercise, and monitor and record data regularly every day". Current anomaly With standardized historical feature vectors Input the rule base, match the rules that meet the conditions through the forward reasoning algorithm, calculate the rule priority according to "anomaly weight × historical fit", and select the top 3 high priority rules as candidates; The guidance suggestions corresponding to the candidate rules are deduplicated and logically validated (to avoid suggestion conflicts), and combined with historical user feedback. (Preference coefficient) Adjust the style of expression and difficulty of implementation of the suggestions, and generate a personalized adaptation guidance plan with 1 core suggestion and 2 auxiliary suggestions.
[0020] Extract time-domain features (such as peak value and period) and frequency-domain features (such as dominant frequency) from motion-related data, and classify them using the Support Vector Machine (SVM) algorithm. The calculation formula is as follows: ; In the formula, The motion feature vector to be classified; For kernel functions; For bias terms; Label the sports mode category; For Lagrange multipliers; Obtain abnormal alerts, personalized adaptation guidance, and sports mode category tags; S3 receives the abnormal prompts, personalized guidance schemes and motion pattern recognition results output by S2, uses a data processing intelligent scheduling algorithm to allocate and schedule resources for the tasks of S1 and S2, optimizes task processing efficiency through objective function, and ensures data security with encrypted transmission verification algorithm to obtain efficient and secure final data results. In this step, based on the priority of data types (analysis conclusions / guidance plans / identification results), the computational process is optimized through a resource allocation algorithm, with the objective function being: ; The constraints are: ; ; ; In the formula, For the number of data processing tasks, For the first The weight coefficient of each task. To optimize computational efficiency and reduce task processing time; This is the maximum tolerance time for a single round of data processing, with a default value of 500ms. For the first The minimum processing time for a task is determined by hardware performance, such as heart rate data processing. ; For the first The number of CPU cores / memory used by each task; Total computing resources for the device; Data type priority is directly mapped to task weight coefficient. (Weight of exception alert data) Personalized guidance Motion recognition The higher the priority, the better. The larger the value, the higher the priority for allocating computing resources; By dynamically allocating computing resources, such as allocating more CPU cores to high-priority tasks and shortening task parallel scheduling time... And thus minimize the objective function For example, the anomaly detection task and the motion recognition task can be processed in parallel to avoid sequential waiting; The output data is segmented and encrypted, and a hash algorithm is used to generate a data digest. The encryption formula is expressed as: ; In the formula, For the data to be encrypted, For random salt values, For data summary; At the data receiving end, the digest is reconstructed using the same hash algorithm, and the consistency of the digests at both ends is compared. The verification formula is expressed as follows: ; In the formula, For the sender's digest, This is the receiver's digest. This indicates that the verification has passed; To obtain the final data after efficient and secure processing; S4. Receive the efficient and secure final data result output by S3, extract the scene feature vector, perform algorithm adaptation optimization based on the mapping relationship between the scene feature vector and the parameter set of the preceding core algorithm, and use root mean square error to evaluate the adaptation effect. In this step, the final data results from S3 are received, and relevant parameters for the usage scenario (such as usage duration, ambient brightness, and user operation frequency) are extracted to construct a scenario feature vector. ,in, For scene feature dimensions; Based on the scene feature vector, the key parameters of the preceding core algorithm are adjusted through a mapping algorithm. The mapping relationship is expressed as follows: ; In the formula, for dimensional weight matrix, for Number of parameters For scene feature dimensions; The adjusted set of algorithm parameters is represented as follows: ; In the formula, The fusion weights are the fusion weights of the weighted fusion algorithm; The mean and standard deviation of the anomaly detection model; These are the model parameters for the Support Vector Machine algorithm; These are the adaptive calibration coefficients for the dynamic calibration model; For mapping weights; It is the bias vector; Calculate the output error of the adjusted algorithm. If the error exceeds a preset range, iteratively optimize the mapping algorithm parameters, specifically as follows: The root mean square error (RMSE) is used as the error evaluation metric. The deviation between the adjusted algorithm output data and the scene adaptation baseline value is calculated using the following formula: ; In the formula, To verify the data sample size, This represents the i-th output value of the algorithm after parameter adjustment. To serve as the adaptation benchmark for the corresponding scenario, the average value of 300 sets of monitoring data from standard laboratory equipment in the same scenario was taken. Only when the scene feature vector Triggered when the change exceeds the threshold. Adjustments to avoid ineffective optimizations are represented as: ; In the formula, This represents the change in the scene feature vector; This represents the total number of dimensions of the scene features; For the first The current new value of each scene feature; For the first Historical old values of scene features; The preset threshold for the amount of change; When the scene is in "motion state", Adapted to the S2 support vector machine algorithm, adjustments were made. Improve the accuracy of motion classification; When the scenario is "static monitoring", Adapt to S1 weighted fusion algorithm and S2 anomaly detection model, and adjust Improve the accuracy of heart rate / blood pressure monitoring; Preset error threshold ,like The weight matrix of the mapping algorithm is then iteratively optimized using the gradient descent method. and bias The iterative update formula is as follows: ; ; In the formula, For the number of iterations, For learning rate Iterate to Or it may stop after reaching the maximum number of iterations; The algorithm output data after parameter optimization is format-standardized to unify data fields, precision, and transmission protocols, generating structured data that meets the application interface requirements. Invalid and redundant information is removed using a redundancy removal algorithm, retaining core valid data and reducing data transmission volume. The CRC32 check algorithm is used to verify the integrity of the output data, generating a check code that is output synchronously with the data to ensure that the data received by the application is complete and error-free. The standardized, redundancy-removed, and verified data is output to the application as usable data adapted to the scenario.
[0021] Therefore, this invention adopts the above-mentioned multifunctional body feature monitoring method and system based on smart wearable devices. This solution is based on multi-source physiological data fusion algorithms and dynamic calibration models to improve the accuracy of monitoring heart rate, blood glucose, and blood pressure related data and optimize the recognition effect of arrhythmia features. At the same time, it realizes intelligent prompts and personalized adaptation guidance for abnormal physiological data through multi-dimensional physiological parameter intelligent analysis algorithms, completes the recognition of various movement modes by relying on motion feature extraction and classification algorithms, optimizes the computing efficiency by using data processing intelligent scheduling algorithms, ensures data security by using encrypted transmission verification algorithms, and improves the experience adaptability in different usage scenarios through algorithm adaptation strategies.
[0022] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0023] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.
Claims
1. A multifunctional body feature monitoring method based on a smart wearable device, characterized in that, Includes the following steps: S1. Collect multi-dimensional human body feature data through sensors of smart wearable devices, perform weighted fusion and dynamic calibration on the collected data, and obtain accurate physiological data and feature recognition results after calibration. S2 receives the accurate physiological data and feature recognition results output by S1, and uses the anomaly judgment model to identify abnormal body features based on the data. At the same time, the support vector machine algorithm is used to classify the movement pattern to obtain abnormal prompt information, personalized guidance plan and movement pattern recognition results. S3 receives the abnormal prompts, personalized guidance schemes and motion pattern recognition results output by S2, uses a data processing intelligent scheduling algorithm to allocate and schedule resources for the tasks of S1 and S2, optimizes task processing efficiency through objective function, and ensures data security with encrypted transmission verification algorithm to obtain efficient and secure final data results. S4. Receive the efficient and secure final data result output by S3, extract the scene feature vector, perform algorithm adaptation optimization based on the mapping relationship between the scene feature vector and the parameter set of the preceding core algorithm, and use root mean square error to evaluate the adaptation effect.
2. The multifunctional body feature monitoring method based on a smart wearable device according to claim 1, characterized in that, In S1, the fusion processing and dynamic calibration specifically include: The multi-source raw physiological data, including heart rate data, blood pressure data, exercise acceleration data, and body temperature data, were denoised. The preprocessed data from multiple sources of raw physiological data are integrated using a weighted fusion algorithm. The calculation formula is as follows: ； In the formula, For the first The data after road preprocessing To integrate weights, satisfy ; An error correction model is established based on historical calibration data, and the fusion results are adjusted in real time. The calibration formula is as follows: ； In the formula, For reference benchmark value, For adaptive calibration coefficients, the range of values is... .
3. The multifunctional body feature monitoring method based on a smart wearable device according to claim 1, characterized in that, In S2, the anomaly detection model is based on the mean of the data. and standard deviation Build, when data deviates The interval is considered abnormal; The classification function of the support vector machine is: ； In the formula, The motion feature vector to be classified; For kernel functions; For bias terms; Label the sports mode category; It is a Lagrange multiplier.
4. The multifunctional body feature monitoring method based on a smart wearable device according to claim 1, characterized in that, In S3, the objective function of the intelligent scheduling algorithm for data processing is: ； In the formula, For the number of data processing tasks, For the first The weight coefficient of each task. For task processing time; The task weight coefficient corresponds to the data type priority, including exception message data. Personalized guidance data Motion recognition data .
5. The multifunctional body feature monitoring method based on a smart wearable device according to claim 1, characterized in that, The constraints of the intelligent scheduling algorithm include: ； ； ； In the formula, This is the maximum tolerance time for a single round of data processing, with a default value of 500ms. For the first Minimum processing time for each task; The task consumes computing resources; Total computing resources for the device.
6. The multifunctional body feature monitoring method based on a smart wearable device according to claim 1, characterized in that, In S4, the parameter set is represented as: ； In the formula, The fusion weights are the fusion weights of the weighted fusion algorithm; The mean and standard deviation of the anomaly detection model; These are the model parameters for the Support Vector Machine algorithm; These are the adaptive calibration coefficients for the dynamic calibration model; For mapping weights; This is the bias vector.
7. The multifunctional body feature monitoring method based on a smart wearable device according to claim 6, characterized in that, The mapping relationship is expressed as follows: ； In the formula, for dimensional weight matrix, for Number of parameters For scene feature dimensions; It is the bias vector; Triggered only when the change in scene features is not less than a preset value. Adjustment, represented as: ； In the formula, This represents the change in the scene feature vector; This represents the total number of dimensions of the scene features; For the first The current new value of each scene feature; For the first Historical old values of scene features; The preset threshold for the amount of change, ; The adaptation benchmark value of RMSE The average value of monitoring data from 300 sets of standard laboratory equipment in the same scenario is calculated using the following formula: ； In the formula, To verify the data sample size, The algorithm after parameter adjustment One output value, These are the adaptation baseline values for the corresponding scenarios.
8. A multifunctional body feature monitoring system based on a smart wearable device, applied to the multifunctional body feature monitoring method based on a smart wearable device as described in any one of claims 1-7, characterized in that, include: The data fusion calibration module is used to collect multi-dimensional human body feature data through sensors; The intelligent analysis and recognition module is connected to the data fusion and calibration module and is used to perform weighted fusion and dynamic calibration on the collected data, and output accurate physiological data and feature recognition results. The data optimization and security module is connected to the intelligent analysis and recognition module. It is used to receive the output data of the fusion calibration module, identify abnormal body features based on the anomaly judgment model, classify movement patterns through the SVM algorithm, and output anomaly prompt information, personalized guidance plan and movement pattern recognition results. An anomaly identification module, connected to the data optimization and security module, is used to receive the output data of the fusion calibration module, identify abnormal body features based on the anomaly judgment model, classify movement patterns through the SVM algorithm, and output anomaly prompts, personalized guidance plans, and movement pattern identification results. The intelligent scheduling module, connected to the anomaly identification module, is used to receive the output data of the anomaly identification module, allocate computing resources and schedule processing tasks using the intelligent scheduling algorithm described in claims 4-5, and ensure data security with the encrypted transmission verification algorithm, and output efficient and secure final data results. An algorithm adaptation module, connected to the intelligent scheduling module, is used to receive the output data of the intelligent scheduling module, extract scene feature vectors, optimize the parameter set based on the mapping relationship described in claims 6-7, and evaluate the adaptation effect through RMSE.
9. A computer device, characterized in that, include: A processor configured to be coupled to a memory, read and execute instructions and / or program code in the memory to perform the method as described in any one of claims 1-7.
10. A computer-readable medium, characterized in that, The computer-readable medium stores computer program code that, when executed on a computer, causes the computer to perform the method as described in any one of claims 1-7.