An AI-based intelligent motion information processing method and system

Abstract

The present invention discloses an AI-based intelligent motion information processing method and system, the method comprising: collecting motion data of different parts of the user's body, extracting features from the motion data; inputting the extracted motion features into an AI-based motion detection model, and outputting the user's motion pattern; calculating the user's motion stability score based on the motion data, and sending the motion pattern and motion stability score to the user terminal for display; encrypting the motion data, nonlinearly slicing the encrypted motion data, and assigning a unique periodic identifier to each slicing motion data; and distributing the slicing motion data to different nodes of the blockchain. By using the embodiments of the present invention, the analysis, encryption and intelligent storage of motion data can be realized through the combination of AI technology and blockchain, and a safe and efficient motion information processing solution can be provided to meet the user's needs for health management.

Description

Technical Field

[0001] The present invention belongs to the field of AI technology, and in particular to an AI-based intelligent motion information processing method and system. Background Art

[0002] With the improvement of health awareness and the advancement of science and technology, intelligent sports monitoring has gradually become an important part of the sports and fitness field. Traditional sports data collection methods mostly rely on manual recording, lack of real-time and accuracy, and it is difficult to provide users with personalized feedback and suggestions. In recent years, the rapid development of artificial intelligence technology has provided new opportunities for the intelligent processing and analysis of sports data.

[0003] Existing sports monitoring equipment can only collect data in real time, but cannot deeply analyze the user's sports pattern and stability. In addition, the security and privacy protection of sports data are becoming increasingly prominent, and traditional data storage methods are prone to the risk of information leakage. Summary of the invention

[0004] The purpose of the present invention is to provide an AI-based intelligent motion information processing method and system to address the deficiencies in the prior art. By combining AI technology with blockchain, it is possible to achieve analysis, encryption and intelligent storage of motion data, and provide a safe and efficient motion information processing solution to meet users' needs for health management.

[0005] An embodiment of the present application provides an AI-based intelligent motion information processing method, the method comprising:

[0006] Collecting motion data of different parts of the user's body and performing feature extraction on the motion data;

[0007] The extracted motion features are input into a pre-trained AI-based motion detection model to output the user’s motion pattern;

[0008] Calculating the user's motion stability score based on the motion data, and sending the user's motion pattern and motion stability score to the user terminal for display;

[0009] Encrypt the user's motion data, perform nonlinear slicing on the encrypted motion data, and assign a unique periodic identifier to each slicing piece of motion data. The periodic identifier will be recalculated and updated after each time period.

[0010] The sharded motion data is distributed and stored in different nodes of the blockchain.

[0011] Optionally, inputting the extracted motion features into a pre-trained AI-based motion detection model to output the user's motion pattern includes:

[0012] The extracted original motion feature set is standardized to eliminate the feature dimension difference and convert each motion feature into a standard normal distribution;

[0013] Map the standardized motion feature set to a low-dimensional continuous space to generate a feature embedding vector;

[0014] The feature embedding vector is input into a multi-channel convolutional neural network, where each channel extracts the corresponding deep features through convolution kernels of different sizes to generate the corresponding multi-channel feature map;

[0015] Map each multi-channel feature map to a feature vector of fixed length, and concatenate feature vectors of different scales to form an aggregated feature vector;

[0016] The aggregated feature vector is input into a bidirectional long short-term memory network to capture the temporal dynamic dependencies of motion features and generate a sequence feature representation containing temporal information;

[0017] Processing the sequence feature representation to identify key motion pattern features and generate a weighted feature representation;

[0018] The weighted feature representation is input into the fully connected layer to perform motion mode classification prediction and output the user's motion mode category.

[0019] Optionally, the non-linear slicing of the encrypted motion data to assign a unique periodic identifier to each slicing piece of motion data includes:

[0020] Divide the encrypted motion data into a number of initial data segments according to a fixed size, and generate a unique hash identifier for each initial data segment;

[0021] Apply a nonlinear mapping function to transform each initial data segment, and randomly divide the transformed data segment into a number of fragments, wherein the size of each fragment data segment is randomly generated according to Poisson distribution;

[0022] Based on the hash identifier, a periodic identifier is generated for each shard data segment, where the identifier is reused within a predefined time period. After each period, the shard's periodic identifier is recalculated and updated, and the old periodic identifier is recycled.

[0023] Optionally, the sharded motion data is distributed and stored in different nodes of the blockchain, including:

[0024] Generate shard metadata corresponding to each shard data segment, wherein the shard metadata includes the size of the shard data segment, the generation timestamp of the shard data segment, and a periodicity identifier;

[0025] The metadata of each shard is stored in the main chain node of the blockchain respectively, and the shard data segment corresponding to the shard metadata is dynamically stored in the side chain or cloud chain node corresponding to the main chain node according to the access frequency.

[0026] Another embodiment of the present application provides an AI-based intelligent motion information processing system, the system comprising:

[0027] A collection module, used to collect motion data of different parts of the user's body and perform feature extraction on the motion data;

[0028] An output module, which is used to input the extracted motion features into a pre-trained AI-based motion detection model and output the user's motion pattern;

[0029] A calculation module, used to calculate the user's motion stability score based on the motion data, and send the user's motion pattern and motion stability score to the user terminal for display;

[0030] The sharding module is used to encrypt the user's motion data, perform nonlinear sharding on the encrypted motion data, and assign a unique periodic identifier to each piece of the sharded motion data. The periodic identifier will be recalculated and updated after each time period.

[0031] The storage module is used to distribute and store the sharded motion data in different nodes of the blockchain.

[0032] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when running.

[0033] Yet another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the methods described above.

[0034] Compared with the prior art, the present invention provides an AI-based intelligent motion information processing method, which collects motion data of different parts of the user's body and extracts features from the motion data; inputs the extracted motion features into a pre-trained AI-based motion detection model, and outputs the user's motion pattern; calculates the user's motion stability score based on the motion data, and sends the user's motion pattern and motion stability score to the user terminal for display; encrypts the user's motion data, performs nonlinear slicing on the encrypted motion data, and assigns a unique periodic identifier to each piece of motion data after slicing, which is recalculated and updated after each time period; and stores the slicing motion data in a distributed manner in different nodes of the blockchain, so that through the combination of AI technology and blockchain, motion data analysis, encryption and intelligent storage can be achieved, providing a safe and efficient motion information processing solution to meet users' needs for health management. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A hardware structure block diagram of a computer terminal for an AI-based intelligent motion information processing method provided by an embodiment of the present invention;

[0036] Figure 2 A flowchart of an AI-based intelligent motion information processing method provided by an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of the structure of an AI-based intelligent motion information processing system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0039] The embodiment of the present invention first provides an AI-based intelligent motion information processing method, which can be applied to electronic devices such as computer terminals, specifically ordinary computers, etc.

[0040] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal of an AI-based intelligent motion information processing method provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the computer terminal may include one or more ( Figure 1Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0041] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / modules corresponding to the AI-based intelligent motion information processing method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implementing the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0042] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0043] See also Figure 2 , an embodiment of the present invention provides an AI-based intelligent motion information processing method, which may include the following steps:

[0044] S201, collecting motion data of different parts of the user's body, and performing feature extraction on the motion data;

[0045] In this process, it is first necessary to collect motion data of different parts of the user's body during exercise through various wearable devices (such as smart watches, sensors or sports equipment). These data may include but are not limited to joint angles, acceleration, heart rate, pace, muscle activity, etc. The diversity of motion data provides a rich information basis for subsequent feature extraction, which helps to fully understand the user's motion status. The purpose of feature extraction is to extract key and representative features from the original motion data to facilitate subsequent motion pattern analysis and identification. This step is crucial to improving the classification accuracy of the motion detection model. One implementation method may include:

[0046] 1. Data collection: By installing multiple sensors on key parts of the user's body (such as wrists, ankles, waist, etc.), motion data can be collected in real time. For example, a three-axis accelerometer and gyroscope can be used to simultaneously obtain the user's acceleration and angular velocity data.

[0047] 2. Data preprocessing: Noise filtering and missing value completion are performed on the collected raw data. The Kalman filter algorithm is used to smooth the data to reduce the interference of the external environment on the data. At the same time, the data is divided into time windows according to the type of exercise. For example, for running exercise, each time window is set to 1 second.

[0048] 3. Key point extraction Through kinematic analysis, the movement trajectory of the key points of the user's body (such as knee joints, ankle joints, shoulder joints, etc.) during exercise can be identified. In this process, the dynamic time warping (DTW) algorithm can be applied to align the time series data to ensure the accuracy of the feature extraction process.

[0049] 4. Feature generation Based on the extracted key point data, a series of motion features are calculated from each time window, such as:

[0050] - Velocity: Calculated by time-differentiating the displacement data.

[0051] - Acceleration rate of change: calculated by differentiating the acceleration data, the acceleration rate of change can reflect the smoothness of the movement.

[0052] - Power output: Combine heart rate data and exercise intensity calculation to obtain the user's energy consumption and output in each time window.

[0053] 5. Extraction of high-order statistical features For the basic features generated in each time window, further calculate the high-order statistical features such as mean, standard deviation, maximum value, minimum value, and combine them into a feature vector. In this way, the final feature vector not only contains basic motion information, but also reflects the changing trend during the motion process.

[0054] 6. Feature selection and dimensionality reduction In order to improve the efficiency of subsequent models, the principal component analysis (PCA) or linear discriminant analysis (LDA) method can be used to reduce the dimensionality of the extracted high-dimensional features, select features that are highly correlated with the motion pattern, and form a more compact feature set.

[0055] Through the above steps, representative features can be extracted from the user's motion data in a comprehensive and detailed manner, laying the foundation for subsequent motion pattern recognition. This not only improves the accuracy of the model, but also enhances the model's ability to recognize different types of motion.

[0056] S202, inputting the extracted motion features into a pre-trained AI-based motion detection model to output the user's motion pattern;

[0057] In this process, the extracted motion features are input into a pre-trained AI motion detection model. This AI model is usually based on machine learning or deep learning algorithms, which can identify and classify the user's motion patterns based on the input feature data. The motion patterns may include different categories such as running, walking, cycling, swimming, etc.

[0058] Inputting the extracted motion features into the AI ​​model can achieve real-time analysis and feedback of the user's motion status. This not only helps users understand the type of exercise they are doing, but also helps provide personalized health advice and training plans based on the exercise pattern, thereby optimizing the user's exercise effect and improving exercise safety. In addition, accurate motion pattern recognition can provide reliable data support for subsequent data analysis, health monitoring and other applications, and provide a scientific basis for decision-making in the fields of fitness and medical care.

[0059] Specifically, the extracted original motion feature set can be standardized to eliminate feature dimension differences and convert each motion feature into a standard normal distribution;

[0060] The feature normalization process converts all extracted motion features into the same dimension to eliminate the dimensional differences between different features, so that the impact of each feature on the results is balanced during the model training process. This is an important prerequisite for ensuring model performance.

[0061] Standardization can improve the training efficiency of the model and reduce the impact of features of different dimensions on the model learning process, thereby improving the model's sensitivity and expressiveness to features and enhancing the model's convergence speed and stability.

[0062] Map the standardized motion feature set to a low-dimensional continuous space to generate a feature embedding vector;

[0063] Feature embedding aims to map high-dimensional feature vectors to low-dimensional space to retain their important features and structural information. This process is usually achieved through some dimensionality reduction technology or embedding method, such as through deep learning-based embedding models such as Word2Vec.

[0064] Low-dimensional embedding makes subsequent calculations more efficient, reduces computational complexity, and retains important information in the data. This lays the foundation for model learning and classification and helps reduce the risk of overfitting.

[0065] The feature embedding vector is input into a multi-channel convolutional neural network, where each channel extracts the corresponding deep features through convolution kernels of different sizes to generate the corresponding multi-channel feature map;

[0066] Multi-channel convolutional neural networks use multiple convolution kernels to perform convolution on feature embedding vectors, which can effectively extract feature information at different scales and perspectives. This multi-channel structure helps capture the diversity of data and improve the representativeness of features.

[0067] By extracting deep features through convolution kernels of different sizes, the model can learn richer motion pattern information and enhance its ability to recognize different motion categories.

[0068] Map each multi-channel feature map to a feature vector of fixed length, and concatenate feature vectors of different scales to form an aggregated feature vector;

[0069] The feature map generated by each convolution channel is mapped into a feature vector of fixed length, and then the feature information of different scales is integrated together by splicing to form a comprehensive feature representation.

[0070] Aggregated feature vectors combine features from multiple scales, allowing the model to use more information for decision making and improve the accuracy of motion pattern classification.

[0071] The aggregated feature vector is input into a bidirectional long short-term memory network to capture the temporal dynamic dependencies of motion features and generate a sequence feature representation containing temporal information;

[0072] The bidirectional long short-term memory network (Bi-LSTM) can simultaneously consider the time series information of the past and the future, and capture the dynamic relationship and change trend between motion features. By capturing time information, the model can better understand the continuity of motion patterns and improve the ability to judge complex motion behaviors.

[0073] Processing the sequence feature representation to identify key motion pattern features and generate a weighted feature representation;

[0074] Through the weighting mechanism, the model can highlight the features that have a greater impact on the classification results, filter out the features that have a smaller impact on the decision, and enhance the model's discrimination ability. The generated weighted feature representation can help the model improve accuracy and stability when classifying motion patterns.

[0075] The weighted feature representation is input into the fully connected layer to perform motion mode classification prediction and output the user's motion mode category.

[0076] The final weighted feature representation is input into the fully connected layer to classify the movement pattern and derive the type of movement the user is performing. Through the classification of the fully connected layer, the model can accurately predict the user's movement pattern and provide real-time feedback and guidance to the user.

[0077] Through the above steps, the AI ​​model can efficiently identify and classify users' movement patterns, providing reliable support for intelligent sports management.

[0078] S203, calculating the user's motion stability score according to the motion data, and sending the user's motion pattern and motion stability score to the user terminal for display;

[0079] In this step, the system calculates the user's motion stability score through an algorithm based on the motion data collected from different parts of the user. This score reflects the stability of the user during the exercise. A higher score means a more stable motion posture, and vice versa. The calculation of the stability score is usually combined with the time series information of the motion data to accurately capture the dynamic changes of the user.

[0080] The purpose of calculating the motion stability score is to help users understand their motion status and health level. This can provide users with clear feedback to help them improve their sports skills and safety during exercise. At the same time, by displaying the motion mode and stability score in real time, it can motivate users to continue to pay attention to their exercise results and enhance their participation in exercise. The calculation steps of a motion stability score include:

[0081] 1. Define various basic parameters:

[0082] Stability score over a period of time.

[0083] Acceleration data of the i-th sensor at time t (the t-th acquisition time point).

[0084] The velocity data of the i-th sensor at time t.

[0085] The position data of the i-th sensor at time t.

[0086] The total number of sensors.

[0087] The length of the time window for acquisition.

[0088] The shock absorption coefficient reflects an individual's ability to withstand impact.

[0089] 2. Data preprocessing and standardization:

[0090] - Data from all sensors Standardization is performed to remove bias and dimension effects.

[0091] 3. Calculation of dynamic characteristic parameters:

[0092] - RMS acceleration: used to measure fluctuations in motion:

[0093]

[0094] - Speed ​​coefficient of variation: used to quantify speed fluctuations during motion:

[0095]

[0096] in, is the standard deviation of the velocity during the acquisition period, is the mean velocity.

[0097] - Position drift average: reflects the overall drift degree of the position during the movement:

[0098]

[0099] in, is the initial position data of the i-th sensor in the collection time period.

[0100] 4. Stability scoring formula:

[0101] - Based on the dynamic characteristic parameters calculated above, considering the comprehensive influence of acceleration, speed and position during motion, the stability scoring formula is designed:

[0102]

[0103] - In this formula:

[0104] - : is a damping factor that reflects the negative impact of high acceleration fluctuations on stability.

[0105] - :Stability is evaluated by the consistency of relative movement speed. The lower the coefficient of variation, the higher the stability.

[0106] - : Stability is measured by position drift, the less drift, the higher the ratio.

[0107] Through this formula, the user's movement stability score can be obtained based on considering different biomechanical parameters.

[0108] S204, encrypting the user's motion data, performing nonlinear slicing on the encrypted motion data, and assigning a unique periodic identifier to each piece of the slicing motion data. The periodic identifier will be recalculated and updated after each time period ends;

[0109] First of all, the encryption of sports data is to protect the privacy and data security of users. In the current digital age, sports data usually contains important personal physiological information and behavioral habits, which may be used by criminals, so it is crucial to encrypt the data. Through encryption, only authorized users or systems can access and interpret this data, thereby effectively preventing potential leakage and abuse, and enhancing users' trust in smart sports systems.

[0110] Secondly, nonlinear sharding divides the encrypted motion data into multiple independent and randomized data fragments. This process not only improves the security of the data, but also enhances the fault tolerance and scalability of the data. Since each data fragment is stored independently, the system can process data more flexibly and reduce the impact of single data corruption on the entire system. In addition, the randomized data fragment storage makes the data structure more complex and difficult for attackers to directly analyze and reconstruct, thereby providing additional security.

[0111] Each piece of motion data is assigned a unique periodic identifier so that each data fragment can be clearly identified and managed throughout the system. These identifiers not only ensure the validity and integrity of the data fragment, but also make data tracking and management more efficient. The design of periodic identifiers can be flexibly updated according to time attributes, which can adapt to data changes and user needs, thereby providing convenience in data recovery and verification.

[0112] Finally, the periodic identifier is recalculated and updated after each time period, which enhances the dynamics and flexibility of data management. This mechanism can adjust the identifier in real time according to the changes in the data, provide the latest information for each data fragment, and ensure its accuracy and timeliness in the system. This design method not only improves the efficiency of data management, but also provides a good foundation for subsequent data analysis and mining.

[0113] The implementation of this series of steps not only improves the security, privacy protection and management flexibility of user motion data, but also lays a solid foundation for the reliable operation of the intelligent motion system and promotes the development and application of technology.

[0114] Specifically, the encrypted motion data may be divided into a number of initial data segments according to a fixed size, and a unique hash identifier may be generated for each initial data segment;

[0115] Dividing the encrypted motion data into fixed-size data segments helps manage the data, while generating hash identifiers to ensure that each data segment is unique in the system, facilitating subsequent reference and tracking. Ensure data integrity and security, prevent data overlap or loss, and prepare for further processing.

[0116] First, the encrypted motion data is obtained and a fixed segment size (e.g. 512 bytes) is set. Then, the data is divided into multiple initial data segments according to this size. At the same time, the SHA-256 algorithm is used to generate a unique hash identifier for each initial data segment, which ensures that even a slight change in the data will result in a change in the hash value, thereby ensuring uniqueness.

[0117] A nonlinear mapping function is applied to each initial data segment for transformation, and the transformed data segment is randomly divided into a number of fragments, wherein the size of each fragment data segment is randomly generated according to a Poisson distribution; a nonlinear mapping function can be:

[0118]

[0119] Or:

[0120]

[0121] in, is a nonlinear mapping function, is the transformation parameter, is the data point in the initial data segment. Through the nonlinear mapping function, the transformed data segment has nonlinear characteristics, which can resist possible pattern recognition attacks.

[0122] Nonlinear mapping functions can transform data, making the structure of data more complex, thereby improving security. The randomness of Poisson distribution can ensure the diversity of shard sizes and prevent attackers from analyzing based on static patterns of data fragments. Randomized sharding not only enhances data security, but also improves the flexibility and adaptability of the system. In this way, data can be better distributed in the network, reducing the risk of attacks.

[0123] Use a nonlinear mapping function, such as the above formula, to apply the mapping to the data points in each initial data segment. After the mapping is completed, the data size of each shard is randomly generated according to the Poisson distribution (for example, the lambda value in the Poisson distribution is selected as 5), which will create a non-uniform, randomized data sharding and increase the complexity of the data.

[0124] Based on the hash identifier, a periodic identifier is generated for each shard data segment, where the identifier is reused within a predefined time period. After each period, the shard periodic identifier is recalculated and updated, and the old periodic identifier is recycled. A method for generating a periodic identifier can be:

[0125]

[0126] in, is the periodic identifier of the i-th fragment data segment in the current time period, is the current time period, is the hash identifier of the i-th shard data segment, is the time increment factor, is the total number of identifiers. This mechanism uses time dynamics to enhance the confidentiality and reuse of shard identifiers.

[0127] This approach provides a dynamically managed identifier for each data fragment, providing effective uniqueness and timeliness in data storage and retrieval. The periodic update mechanism ensures the validity and traceability of data in long-term storage. During the storage and processing of data, the life cycle of data can be effectively managed, which increases the management efficiency of the system and reduces resource usage through the reuse mechanism.

[0128] Generate a periodic identifier for each shard, recalculate the identifier at regular intervals (such as every 5 minutes), and mark the old identifier as invalid after a new identifier is generated to ensure dynamic management of the identifier.

[0129] The transformed data segment is randomly divided into several shards, and the size of each shard is randomly generated according to the Poisson distribution to increase the unpredictability of the shards. These shards are reordered according to the custom priority rules to ensure the integrity of the data during the storage process.

[0130] When a user requests data, the current valid identifier set is restored through reverse calculation based on the requested time period and the identification information of the original data. The dynamic hash table query mechanism is used to quickly lock the storage node and restore the original data, and the complete data is reconstructed through the reverse nonlinear transformation formula. Through the above steps, the entire data processing process not only enhances the security and privacy protection of motion data, but also improves the system's ability to manage and persist data.

[0131] - Improved data security: Nonlinear mapping increases the randomness and unpredictability of data sharding, preventing pattern recognition and side-channel attacks, thereby improving the security of data storage.

[0132] - Storage efficiency and privacy protection: The periodic identifier reuse mechanism effectively reduces the number of stored identifiers and improves storage efficiency, while concealing the true identity information of the data and enhancing privacy protection.

[0133] - Enhanced data availability: Distributed storage and redundancy verification strategies are used to ensure that the storage system can still provide high availability and data recovery capabilities even if some nodes fail.

[0134] - Network resource optimization: Dynamic identification update and recycling mechanism can reduce identification resource waste and improve the overall utilization and response speed of network resources.

[0135] S205, the sharded motion data is distributed and stored in different nodes of the blockchain.

[0136] This process aims to utilize the decentralized, tamper-proof, and transparent characteristics of blockchain technology to make the management and storage of motion data more secure and reliable. The motion data processed by sharding can not only effectively reduce the storage pressure of a single node, but also improve the efficiency and reliability of data access. Specifically, storing the sharded motion data in the blockchain can ensure the integrity of the data chain, and any modification to the data can be traced, thereby effectively preventing malicious tampering of the data. In addition, using the consensus mechanism of the blockchain, participants can ensure the consistency and accuracy of the data, thereby providing users with higher trust and data security. This method can effectively enhance users' trust in motion data while ensuring data security and privacy protection in a multi-user environment.

[0137] Specifically, shard metadata corresponding to each shard data segment may be generated, wherein the shard metadata includes the size of the shard data segment, a generation timestamp of the shard data segment, and a periodicity identifier;

[0138] In this step, for each sharded data segment, the system will generate corresponding metadata, which includes the size of the shard (i.e., the number of bytes), the creation timestamp, and the corresponding periodic identifier. This information is crucial for subsequent data management and access, because it not only provides the basic properties of the data segment, but also helps the system track the historical version and storage status of the data.

[0139] Generating shard metadata helps achieve efficient query and retrieval. When you need to obtain a specific data segment, you can quickly rely on metadata to determine the size, generation time, access rights, etc. of the data. In addition, these metadata will also become the basic elements stored in the blockchain, which can improve the accessibility and traceability of data in the distributed network.

[0140] The metadata of each shard is stored in the main chain node of the blockchain respectively, and the shard data segment corresponding to the shard metadata is dynamically stored in the side chain or cloud chain node corresponding to the main chain node according to the access frequency.

[0141] This step involves storing the generated shard metadata in the main chain of the blockchain. The main chain is the core part of the blockchain and is responsible for recording all important transactions and data changes. This process ensures that each shard metadata has a persistent record on the blockchain. Storing shard metadata on the main chain ensures its immutability and transparency, making the metadata information accessible to all parties, thereby establishing a trust mechanism. In addition, all data on the main chain is publicly verifiable, which can effectively avoid the risk of data loss or tampering.

[0142] This step mainly monitors and analyzes the access frequency of motion data, and makes dynamic storage decisions for each shard data segment. Highly accessed data segments will be stored in the side chain first to increase the reading speed of access information; while infrequently accessed data segments can be transferred to the cloud chain to reduce storage costs.

[0143] This storage strategy can optimize resource utilization and storage costs while maintaining system performance. By storing high-frequency data in a more efficient sidechain environment, users can get a fast response when accessing it; while the storage of low-frequency data on the cloud chain ensures that even with lower access requirements, the data can still be saved and accessed on demand, maximizing the security and availability of the data.

[0144] You can design an access frequency monitoring module that can track and record the access of each shard data segment in real time. You can use an array or hash table to store the access count of each data segment and increment it every time it is accessed. Set a threshold for access frequency to distinguish high-frequency data from low-frequency data. The threshold can be adjusted based on historical data access trends to ensure that the appropriate data segment is selected for storage under changing load conditions.

[0145] The storage strategy is dynamically determined based on the access frequency and the set threshold. If the access frequency of a shard data segment is higher than the threshold, the system recommends storing it in a more efficient side chain to maintain a fast access response; low-frequency data segments can be transferred to a lower-cost cloud chain. When a shard data segment is identified as high-frequency data, the system automatically routes its data storage request to the side chain for appropriate storage. At the same time, the metadata of the main chain is kept updated to ensure that all storage information can be reflected in real time. Exemplary, a specific implementation method may include:

[0146] 1. Access frequency statistics module: Design an access frequency statistics module, and use data structures (such as hash tables or counters) to record the number of accesses to each shard data segment in real time. Whenever a user accesses a shard, update the access count of the shard.

[0147] 2. Time window setting: Introduce the concept of sliding time window, for example, set a 7-day time window to monitor the access to shard data. The number of accesses within this time window determines its storage location in the side chain or cloud chain.

[0148] 3. Frequency threshold setting: Set a dynamically adjusted frequency threshold based on historical access data. For example, when a data segment is accessed more than 100 times within a time window, the system marks it as high-frequency access data. An adaptive algorithm can be implemented in the system to adjust this threshold based on the overall access pattern of the system.

[0149] 4. Storage decision: Whenever the access frequency is updated, the system will check the relationship between the current access count of the data segment and the threshold. If the access count is higher than the threshold, the data segment will be marked as high frequency and converted to sidechain storage. If it is lower than the threshold, it will be retained in the cloud chain. While dynamically updating the storage path, ensure that the metadata of the main chain is updated.

[0150] 5. Storage path management: In terms of storage path management, version control is used to track the movement of data segments. When a shard is migrated from the cloud chain to the side chain, its new storage location is recorded and the metadata is updated for subsequent access.

[0151] 6. Regular audit and data migration: Set up a regular audit mechanism to regularly check the access to the stored sharded data segments. For data segments that have not been accessed for a long time, the system can recommend moving them back to the cloud chain to save storage resources on the side chain. At the same time, upload frequently accessed data in the cloud chain to the side chain to ensure its access efficiency and ensure that the data is in the most suitable location for storage.

[0152] 7. Data recovery mechanism: Design a data recovery mechanism to ensure that the data segments stored in the cloud chain can be quickly restored to the side chain as needed. If a data segment stored in a cloud chain suddenly becomes frequently accessed, the system will automatically enable backup recovery and immediately transfer it to the side chain when the user needs it, ensuring the smoothness of user operations.

[0153] Through the above steps, the system can effectively manage the storage and access of motion data, optimize resource allocation, improve user access experience, and ensure the security and continuous availability of data.

[0154] It can be seen that the motion data of different parts of the user's body are collected, and features of the motion data are extracted; the extracted motion features are input into a pre-trained AI-based motion detection model to output the user's motion pattern; based on the motion data, the user's motion stability score is calculated, and the user's motion pattern and motion stability score are sent to the user terminal for display; the user's motion data is encrypted, and the encrypted motion data is non-linearly sliced, and each piece of motion data after slicing is assigned a unique periodic identifier, which will be recalculated and updated after each time period; the slicing motion data is distributed and stored in different nodes of the blockchain, so that through the combination of AI technology and blockchain, the analysis, encryption and intelligent storage of motion data can be realized, and a safe and efficient motion information processing solution can be provided to meet the user's needs for health management.

[0155] Another embodiment of the present invention provides an AI-based intelligent motion information processing system, see Figure 3 , the system may include:

[0156] The acquisition module 301 is used to collect motion data of different parts of the user's body and perform feature extraction on the motion data;

[0157] An output module 302 is used to input the extracted motion features into a pre-trained AI-based motion detection model and output the user's motion pattern;

[0158] A calculation module 303 is used to calculate the user's motion stability score according to the motion data, and send the user's motion pattern and motion stability score to the user terminal for display;

[0159] The slicing module 304 is used to encrypt the user's motion data, perform nonlinear slicing on the encrypted motion data, and assign a unique periodic identifier to each piece of the sliced ​​motion data. The periodic identifier will be recalculated and updated after each time period.

[0160] The storage module 305 is used to store the sharded motion data in a distributed manner in different nodes of the blockchain.

[0161] It can be seen that the motion data of different parts of the user's body are collected, and features of the motion data are extracted; the extracted motion features are input into a pre-trained AI-based motion detection model to output the user's motion pattern; based on the motion data, the user's motion stability score is calculated, and the user's motion pattern and motion stability score are sent to the user terminal for display; the user's motion data is encrypted, and the encrypted motion data is non-linearly sliced, and each piece of motion data after slicing is assigned a unique periodic identifier, which will be recalculated and updated after each time period; the slicing motion data is distributed and stored in different nodes of the blockchain, so that through the combination of AI technology and blockchain, the analysis, encryption and intelligent storage of motion data can be realized, and a safe and efficient motion information processing solution can be provided to meet the user's needs for health management.

[0162] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0163] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps:

[0164] S201, collecting motion data of different parts of the user's body, and performing feature extraction on the motion data;

[0165] S202, inputting the extracted motion features into a pre-trained AI-based motion detection model to output the user's motion pattern;

[0166] S203, calculating the user's motion stability score according to the motion data, and sending the user's motion pattern and motion stability score to the user terminal for display;

[0167] S204, encrypting the user's motion data, performing nonlinear slicing on the encrypted motion data, and assigning a unique periodic identifier to each piece of the slicing motion data. The periodic identifier will be recalculated and updated after each time period ends;

[0168] S205, the sharded motion data is distributed and stored in different nodes of the blockchain.

[0169] It can be seen that the motion data of different parts of the user's body are collected and features are extracted from the motion data;

[0170] The extracted motion features are input into a pre-trained AI-based motion detection model to output the user's motion pattern; based on the motion data, the user's motion stability score is calculated, and the user's motion pattern and motion stability score are sent to the user terminal for display; the user's motion data is encrypted, and the encrypted motion data is processed by nonlinear slicing, and each piece of sliced ​​motion data is assigned a unique periodic identifier, which will be recalculated and updated after each time period; the sliced ​​motion data is distributed and stored in different nodes of the blockchain, so that through the combination of AI technology and blockchain, the analysis, encryption and intelligent storage of motion data can be realized, and a safe and efficient motion information processing solution can be provided to meet the user's needs for health management.

[0171] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0172] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0173] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0174] S201, collecting motion data of different parts of the user's body, and performing feature extraction on the motion data;

[0175] S202, inputting the extracted motion features into a pre-trained AI-based motion detection model to output the user's motion pattern;

[0176] S203, calculating the user's motion stability score according to the motion data, and sending the user's motion pattern and motion stability score to the user terminal for display;

[0177] S204, encrypting the user's motion data, performing nonlinear slicing on the encrypted motion data, and assigning a unique periodic identifier to each piece of the slicing motion data. The periodic identifier will be recalculated and updated after each time period ends;

[0178] S205, the sharded motion data is distributed and stored in different nodes of the blockchain.

[0179] It can be seen that the motion data of different parts of the user's body are collected, and features of the motion data are extracted; the extracted motion features are input into a pre-trained AI-based motion detection model to output the user's motion pattern; based on the motion data, the user's motion stability score is calculated, and the user's motion pattern and motion stability score are sent to the user terminal for display; the user's motion data is encrypted, and the encrypted motion data is non-linearly sliced, and each piece of motion data after slicing is assigned a unique periodic identifier, which will be recalculated and updated after each time period; the slicing motion data is distributed and stored in different nodes of the blockchain, so that through the combination of AI technology and blockchain, the analysis, encryption and intelligent storage of motion data can be realized, and a safe and efficient motion information processing solution can be provided to meet the user's needs for health management.

[0180] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made according to the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. An AI-based intelligent motion information processing method, characterized in that: The method comprises: Collecting motion data of different parts of the user's body and performing feature extraction on the motion data; The extracted motion features are input into a pre-trained AI-based motion detection model to output the user’s motion pattern; According to the motion data, the motion stability score of the user is calculated, and the motion pattern and motion stability score of the user are sent to the user terminal for display; wherein the motion stability score of the user is: Among them, the is the shock absorption coefficient, which reflects the individual's ability to withstand impact. is the root mean square value of acceleration, which is used to measure the volatility in motion: in, is the acceleration data of the i-th sensor at the t-th acquisition time point, is the total number of sensors, is the length of the acquisition time window; Said is the velocity variation coefficient, which is used to quantify velocity fluctuations during motion: in, is the standard deviation of the speed during the acquisition period, is the mean value of speed; Said is the mean value of position drift, reflecting the overall drift degree of position during movement: in, : The position data of the i-th sensor at the t-th acquisition time point, is the initial position data of the i-th sensor in the collection time period; The user's motion data is encrypted, and the encrypted motion data is subjected to non-linear slicing processing, and each piece of the sliced ​​motion data is assigned a unique periodic identifier, and the periodic identifier is recalculated and updated after each time period ends; the encrypted motion data is subjected to non-linear slicing processing, and each piece of the sliced ​​motion data is assigned a unique periodic identifier, including: Divide the encrypted motion data into a number of initial data segments according to a fixed size, and generate a unique hash identifier for each initial data segment; A nonlinear mapping function is applied to each initial data segment for transformation, and the transformed data segment is randomly divided into a number of slices, wherein the size of each slice data segment is randomly generated according to a Poisson distribution; the nonlinear mapping function is: Or: in, is a nonlinear mapping function, , , , is the transformation parameter, is the data point in the initial data segment; Based on the hash identifier, a periodic identifier is generated for each shard data segment, wherein the identifier is reused within a predefined time period. After each period, the periodic identifier of the shard is recalculated and updated, and the old periodic identifier is recycled. The generation method of the periodic identifier is as follows: in, is the periodic identifier of the jth fragment data segment in the current time period, is the current time period, is the hash identifier of the jth shard data segment, is the time increment factor, is the total number of identifiers; The sharded motion data is distributed and stored in different nodes of the blockchain.

2. The method according to claim 1, characterized in that The step of inputting the extracted motion features into a pre-trained AI-based motion detection model and outputting the user's motion pattern includes: The extracted original motion feature set is standardized to eliminate the feature dimension difference and convert each motion feature into a standard normal distribution; Map the standardized motion feature set to a low-dimensional continuous space to generate a feature embedding vector; The feature embedding vector is input into a multi-channel convolutional neural network, where each channel extracts the corresponding deep features through convolution kernels of different sizes to generate the corresponding multi-channel feature map; Map each multi-channel feature map to a feature vector of fixed length, and concatenate feature vectors of different scales to form an aggregated feature vector; The aggregated feature vector is input into a bidirectional long short-term memory network to capture the temporal dynamic dependencies of motion features and generate a sequence feature representation containing temporal information; Processing the sequence feature representation to identify key motion pattern features and generate a weighted feature representation; The weighted feature representation is input into the fully connected layer to perform motion mode classification prediction and output the user's motion mode category.

3. The method according to claim 2, characterized in that The sharded motion data is distributed and stored in different nodes of the blockchain, including: Generate shard metadata corresponding to each shard data segment, wherein the shard metadata includes the size of the shard data segment, the generation timestamp of the shard data segment, and a periodicity identifier; The metadata of each shard is stored in the main chain node of the blockchain respectively, and the shard data segment corresponding to the shard metadata is dynamically stored in the side chain or cloud chain node corresponding to the main chain node according to the access frequency.

4. An AI-based intelligent motion information processing system, characterized in that: The system comprises: A collection module, used to collect motion data of different parts of the user's body and perform feature extraction on the motion data; An output module, which is used to input the extracted motion features into a pre-trained AI-based motion detection model and output the user's motion pattern; The calculation module is used to calculate the user's motion stability score according to the motion data, and send the user's motion pattern and motion stability score to the user terminal for display; wherein the user's motion stability score is: Among them, the is the shock absorption coefficient, which reflects the individual's ability to withstand impact. is the root mean square value of acceleration, which is used to measure the volatility in motion: in, is the acceleration data of the i-th sensor at the t-th acquisition time point, is the total number of sensors, is the length of the acquisition time window; Said is the velocity variation coefficient, which is used to quantify velocity fluctuations during motion: in, is the standard deviation of the speed during the acquisition period, is the mean value of speed; Said is the mean value of position drift, reflecting the overall drift degree of position during movement: in, : The position data of the i-th sensor at the t-th acquisition time point, is the initial position data of the i-th sensor in the collection time period; The slicing module is used to encrypt the user's motion data, perform nonlinear slicing processing on the encrypted motion data, and assign a unique periodic identifier to each piece of the sliced ​​motion data. The periodic identifier will be recalculated and updated after each time period. The nonlinear slicing processing on the encrypted motion data and the unique periodic identifier for each piece of the sliced ​​motion data include: Divide the encrypted motion data into a number of initial data segments according to a fixed size, and generate a unique hash identifier for each initial data segment; A nonlinear mapping function is applied to each initial data segment for transformation, and the transformed data segment is randomly divided into a number of slices, wherein the size of each slice data segment is randomly generated according to a Poisson distribution; the nonlinear mapping function is: Or: in, is a nonlinear mapping function, , , , is the transformation parameter, is the data point in the initial data segment; Based on the hash identifier, a periodic identifier is generated for each shard data segment, wherein the identifier is reused within a predefined time period. After each period, the periodic identifier of the shard is recalculated and updated, and the old periodic identifier is recycled. The generation method of the periodic identifier is as follows: in, is the periodic identifier of the jth fragment data segment in the current time period, is the current time period, is the hash identifier of the jth shard data segment, is the time increment factor, is the total number of identifiers; The storage module is used to distribute and store the sharded motion data in different nodes of the blockchain.

5. The system according to claim 4, characterized in that The output module is specifically used for: The extracted original motion feature set is standardized to eliminate the feature dimension difference and convert each motion feature into a standard normal distribution; Map the standardized motion feature set to a low-dimensional continuous space to generate a feature embedding vector; The feature embedding vector is input into a multi-channel convolutional neural network, where each channel extracts the corresponding deep features through convolution kernels of different sizes to generate the corresponding multi-channel feature map; Map each multi-channel feature map to a feature vector of fixed length, and concatenate feature vectors of different scales to form an aggregated feature vector; The aggregated feature vector is input into a bidirectional long short-term memory network to capture the temporal dynamic dependencies of motion features and generate a sequence feature representation containing temporal information; Processing the sequence feature representation to identify key motion pattern features and generate a weighted feature representation; The weighted feature representation is input into the fully connected layer to perform motion mode classification prediction and output the user's motion mode category.

6. The system according to claim 5, characterized in that The storage module is specifically used for: Generate shard metadata corresponding to each shard data segment, wherein the shard metadata includes the size of the shard data segment, the generation timestamp of the shard data segment, and a periodicity identifier; The metadata of each shard is stored in the main chain node of the blockchain respectively, and the shard data segment corresponding to the shard metadata is dynamically stored in the side chain or cloud chain node corresponding to the main chain node according to the access frequency.

7. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 3 when executed.

8. An electronic device, comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 3.

Images