Intelligent fitness exercise intensity adjusting system for old people

Abstract

The invention discloses an intelligent fitness exercise intensity adjusting system for old people. The system comprises a data acquisition module, a data preprocessing module, an old people physiology monitoring module, an exercise intensity recommendation model building module and an exercise intensity adjusting module. The invention relates to the technical field of fitness and health monitoring of old people, in particular to an intelligent fitness exercise intensity adjusting system for old people. A data preprocessing method of data cleaning, data coding and data normalization is adopted; preliminarily classifying the physiological health data by adopting an improved K center point clustering model, fully mining individualized health features, and converting the individualized health features into an accurate exercise intensity adjustment basis; a bidirectional long-short-term memory network model is adopted as an exercise intensity recommendation model, so that the time dependence of physiological reactions in the exercise process can be captured, and more accurate and continuous recommendation meeting individual requirements is provided.

Description

Technical Field

[0001] The invention relates to the technical field of fitness and health monitoring for the elderly, and in particular to an intelligent fitness exercise intensity adjustment system for the elderly. Background Art

[0002] The elderly fitness exercise intensity adjustment system is a method to scientifically adjust and optimize exercise intensity according to the individual health status, physical fitness level and physiological needs of the elderly. It aims to improve exercise effects while avoiding sports injuries. By evaluating indicators such as heart rate and respiratory rate, combined with real-time monitoring by smart devices, personalized exercise plans are designed to provide the elderly with safer and more effective exercise plans.

[0003] However, traditional methods of regulating exercise intensity for the elderly usually ignore individual health differences and physiological changes, and are often unable to adapt to the dynamic health status of the elderly group in real time, which can easily lead to unreasonable exercise intensity settings and increase technical problems such as excessive or too light exercise. Traditional methods of regulating exercise intensity for the elderly lack personalized adjustment mechanisms, fail to reflect the dynamic changes in the health status of the elderly in real time, and lack in-depth mining and analysis of historical exercise data, and cannot effectively optimize exercise intensity and improve the effectiveness of exercise intervention. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an intelligent system for adjusting the intensity of fitness exercise for the elderly. In view of the fact that the traditional method for adjusting the intensity of exercise for the elderly usually ignores the health differences and physiological changes of individuals, often fails to adapt to the dynamic health status of the elderly group in real time, and easily leads to unreasonable exercise intensity settings, and increases the technical problems of excessive or too light exercise, this solution creatively adopts improved K center point clustering to preliminarily classify the physiological health data of the elderly, avoids the influence of noise and falls into the local optimal solution, and fully explores the individual health characteristics, and converts them into accurate exercise intensity adjustment basis; in view of the fact that the traditional method for adjusting the intensity of exercise for the elderly lacks a personalized adjustment mechanism, fails to reflect the dynamic changes of the health status of the elderly in real time, lacks in-depth mining and analysis of historical exercise data, and cannot effectively optimize the exercise intensity and improve the effect of exercise intervention, this solution creatively adopts a bidirectional long short-term memory network model as an exercise intensity recommendation model, which can capture the time dependence of physiological reactions during exercise, and takes into account the long-term changes in the health status of the elderly and the timeliness of historical exercise performance, so as to provide more accurate, continuous and individualized recommendations.

[0005] The technical solution adopted by the present invention is as follows: the present invention provides an intelligent elderly fitness exercise intensity adjustment system, including a data acquisition module, a data preprocessing module, an elderly physiological monitoring module, an exercise intensity recommendation model construction module and an exercise intensity adjustment module;

[0006] The data collection module obtains the intensity adjustment original data set by collecting data from the health record management system;

[0007] The data preprocessing module performs preprocessing by data cleaning, data encoding and data normalization to obtain a preliminary physiological health data set and a preliminary non-physiological health data set;

[0008] The elderly physiological monitoring module constructs an improved K-center clustering model and performs preliminary classification of physiological health data to obtain an intensity adjustment data set;

[0009] The exercise intensity recommendation model construction module constructs a bidirectional long short-term memory network model as an exercise intensity recommendation model;

[0010] The exercise intensity adjustment module makes exercise intensity recommendations through the exercise intensity recommendation model to obtain fitness exercise intensity recommendation results, and comprehensively adjusts the fitness exercise intensity of the elderly based on the fitness exercise intensity recommendation results to formulate a personalized fitness exercise plan.

[0011] Furthermore, in the data acquisition module, the intensity adjustment original data set specifically includes a physiological health original data set and a non-physiological health original data set. The physiological health original data set specifically includes basic physiological data of the elderly, metabolic physiological data of the elderly, cardiovascular health data of the elderly, respiratory health data of the elderly and musculoskeletal health data of the elderly. The basic physiological data of the elderly specifically includes age, gender, weight, height and disease information. The non-physiological health original data set specifically includes exercise history data of the elderly, environmental data and emotional state data of the elderly. The exercise history data of the elderly specifically includes daily exercise volume data, exercise type preference data and exercise performance data.

[0012] Furthermore, in the data preprocessing module, the data cleaning is specifically to remove missing values ​​and duplicate values, the data encoding is specifically to encode the original data using the one-hot encoding method, and the data normalization is specifically to normalize the original data using the minimum-maximum normalization method. Through the data cleaning, the data encoding and the data normalization, the physiological health original data set and the non-physiological health original data set are preprocessed to obtain the physiological health preliminary data set and the non-physiological health preliminary data set.

[0013] Further, in the elderly physiological monitoring module, specifically, by constructing an improved K center point clustering model, a physiological monitoring model for the elderly is obtained, and the physiological monitoring model for the elderly is used to preliminarily classify the physiological health preliminary data set, and the improved K center point clustering model is specifically a K center point clustering model combined with a harmonious search algorithm;

[0014] The physiological monitoring module for the elderly includes initializing the search algorithm, updating the harmony solution, iterative updating and preliminary classification;

[0015] The initialization search algorithm is used to initialize the harmony search algorithm, and the content includes:

[0016] Determine an iteration termination condition, wherein the iteration termination condition specifically includes reaching a maximum number of iterations and a fitness function value being less than a preset threshold;

[0017] Initialize the harmony search parameters, including the harmony library size Ha, the number of clusters K, and the maximum number of harmony search iterations , the probability pn of adopting a new harmonious solution and the probability pc of changing the solution;

[0018] Initializing the harmony library is specifically to randomly select Ha harmony solutions from the physiological health preliminary data set and initialize the harmony library matrix. The harmony solution is specifically a group of cluster centers of K center point clustering. The harmony library matrix is ​​expressed as follows:

[0019] ;

[0020] Where Hm represents the harmony library matrix, represents the first harmonious solution, represents the second harmonious solution, represents the hath harmonious solution, represents the first dimension of the first harmonious solution, represents the Kth dimension of the first harmonious solution, represents the first dimension of the Hath harmonious solution, represents the Kth dimension of the Hath harmonious solution;

[0021] Determine the fitness function, specifically, select the mean of the average distances from the sample data points in each cluster to the cluster center after clustering as the fitness function value of the harmonious solution. The formula used is as follows:

[0022] ;

[0023] In the formula, represents the fitness function, represents the hath harmonious solution, represents the total number of sample data points in the kth cluster, represents the Euclidean distance calculation function, represents the ath sample data point in the kth cluster, represents the kth cluster center;

[0024] The updated harmonious solution is used to update the harmonious solution of the harmonious search algorithm, and the content includes:

[0025] Generate candidate harmonious solutions, specifically, generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pn of adopting the new harmonious solution. If the generated random value is greater than pn, the original harmonious solution is adopted as the candidate harmonious solution. If the generated random value is less than pn, randomly select the value of the same dimension of other harmonious solutions to replace the value of this dimension of the original harmonious solution as the candidate harmonious solution;

[0026] Generate a new harmonious solution, specifically generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pc of the changed solution. If the generated random value is greater than pc, the candidate harmonious solution is used as the new harmonious solution. If the generated random value is less than pc, a new harmonious solution is generated based on the reference bandwidth and the candidate harmonious solution. The formula used is as follows:

[0027] ;

[0028] In the formula, represents the kth dimension of the hath new harmonious solution, represents the kth dimension of the hath candidate harmonious solution, BW represents the reference bandwidth, Represents a random value in the range [-1,1];

[0029] Since the cluster center of the K-center clustering should be a known sample data point, and the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is an unknown sample data point, the sample data point in the cluster that is closest to the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is selected as the value of this dimension of the new harmonious solution to obtain the new harmonious solution;

[0030] The state parameters of the original harmonious solution and the new harmonious solution are calculated using the following formula:

[0031] ;

[0032] In the formula, represents the growth rate calculation function, represents the hath new harmonious solution, represents the hath original and harmonious solution, represents the calculation function of the state parameters of the new harmonious solution, It represents the calculation function of the original and harmonious solution state parameters, represents the fitness adjustment coefficient, represents the growth rate adjustment factor;

[0033] Update the harmony library matrix, specifically, sort the original harmony solution and the new harmony solution from low to high according to the state parameter, and select the first Ha harmony solutions to form a new harmony library matrix;

[0034] The iterative update specifically includes continuously iterating and updating the algorithm until the iterative termination condition is reached, at which time a group of cluster centers represented by a harmonious solution with the lowest fitness function value is the optimal cluster center combination, and a K center point clustering model is constructed with the optimal cluster center combination as a physiological monitoring model for the elderly;

[0035] The preliminary classification is specifically to perform preliminary classification on the physiological health preliminary data set through the physiological monitoring model for the elderly, and the obtained classification labels are combined with the non-physiological health preliminary data set to obtain the intensity adjustment data set.

[0036] Furthermore, in the exercise intensity recommendation model construction module, a bidirectional long short-term memory network model is specifically constructed as an exercise intensity recommendation model, and the specific contents include labeling and segmenting the data set, constructing a bidirectional long short-term memory layer, designing a cross-level attention mechanism, calculating the model output, and constructing and training the model;

[0037] The labeling and segmenting of the data set specifically includes labeling the data in the intensity adjustment data set as extremely low intensity, low intensity, medium intensity and high intensity, and using the data labels as data labels, and segmenting the labeled intensity adjustment data set into an intensity adjustment training set and an intensity adjustment test set;

[0038] The formula used to construct the bidirectional long short-term memory layer is as follows:

[0039] ;

[0040] In the formula, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the forward long short-term memory function, represents the model input at time step t, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t-1, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the backward long short-term memory function, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t+1, represents the hidden state of the l-th bidirectional long short-term memory layer at time step t, Represents a splicing operation;

[0041] The design of the cross-level attention mechanism includes:

[0042] To calculate the attention weight, the formula used is as follows:

[0043] ;

[0044] In the formula, Indicates Layer and The attention weight calculation function between the bidirectional long short-term memory layers, represents the similarity score calculation function, Indicates The bidirectional long short-term memory layer is at time step The hidden state of Indicates The bidirectional long short-term memory layer is at time step The hidden state of , T represents the total number of time steps, Indicates The bidirectional long short-term memory layer is at time step The hidden state of represents the softmax function, represents the learnable parameter matrix used to map features to the query space, represents a learnable parameter matrix, which is used to map features to the key space, and Tr represents the transposition operation;

[0045] Weighted feature fusion, the formula used is as follows:

[0046] ;

[0047] In the formula, Indicates Layer and The bidirectional long short-term memory layer is at time step The fusion characteristics of

[0048] The formula used for inter-layer feature fusion is as follows:

[0049] ;

[0050] In the formula, represents the final fusion feature, L represents the total number of bidirectional long short-term memory layers;

[0051] The calculation model output is as follows:

[0052] ;

[0053] In the formula, represents the predicted value of the model output, represents the output weight, represents the output bias term;

[0054] The model building and training are specifically to build a bidirectional long short-term memory network model by building a bidirectional long short-term memory layer, designing a cross-level attention mechanism and calculating the model output, and training the model based on the intensity adjustment training set, and verifying the model performance based on the intensity adjustment test set. The model loss function adopts a cross entropy loss function to obtain an exercise intensity recommendation model.

[0055] Furthermore, in the exercise intensity adjustment module, exercise intensity is recommended through the exercise intensity recommendation model to obtain fitness exercise intensity recommendation results, and fitness exercise intensity suitable for the elderly is comprehensively analyzed based on the fitness exercise intensity recommendation results to formulate a personalized fitness exercise plan.

[0056] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0057] (1) Traditional methods for adjusting the exercise intensity of the elderly usually ignore individual health differences and physiological changes, and are often unable to adapt to the dynamic health status of the elderly in real time. This easily leads to unreasonable exercise intensity settings and increases technical problems such as excessive or light exercise. This scheme creatively uses improved K-center point clustering to perform preliminary classification of the physiological health data of the elderly, avoiding the influence of noise and falling into local optimal solutions, while fully exploring individual health characteristics and converting them into accurate basis for exercise intensity adjustment.

[0058] (2) In view of the technical problems that traditional methods for regulating exercise intensity for the elderly lack a personalized adjustment mechanism, fail to reflect the dynamic changes in the health status of the elderly in real time, lack in-depth mining and analysis of historical exercise data, and cannot effectively optimize exercise intensity and improve the effect of exercise intervention, this solution creatively adopts a bidirectional long short-term memory network model as an exercise intensity recommendation model, which can capture the time dependence of physiological responses during exercise, taking into account the long-term changes in the health status of the elderly and the timeliness of historical exercise performance, thereby providing more accurate, continuous and individualized recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic diagram of a module of an intelligent elderly fitness exercise intensity adjustment system provided by the present invention;

[0060] Figure 2 It is a flowchart of the data preprocessing module;

[0061] Figure 3This is a flow chart of the physiological monitoring module for the elderly;

[0062] Figure 4 Schematic diagram of the process of building modules for the exercise intensity recommendation model.

[0063] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0065] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention.

[0066] Example 1, see Figure 1 The present invention provides an intelligent elderly fitness exercise intensity adjustment system, which includes a data acquisition module, a data preprocessing module, an elderly physiological monitoring module, an exercise intensity recommendation model construction module and an exercise intensity adjustment module;

[0067] The data collection module obtains the intensity adjustment original data set by collecting data from the health record management system;

[0068] The data preprocessing module performs preprocessing by data cleaning, data encoding and data normalization to obtain a preliminary physiological health data set and a preliminary non-physiological health data set;

[0069] The elderly physiological monitoring module constructs an improved K-center clustering model and performs preliminary classification of physiological health data to obtain an intensity adjustment data set;

[0070] The exercise intensity recommendation model construction module constructs a bidirectional long short-term memory network model as an exercise intensity recommendation model;

[0071] The exercise intensity adjustment module makes exercise intensity recommendations through the exercise intensity recommendation model to obtain fitness exercise intensity recommendation results, and comprehensively adjusts the fitness exercise intensity of the elderly based on the fitness exercise intensity recommendation results to formulate a personalized fitness exercise plan.

[0072] Example 2, see Figure 1 In the data acquisition module, the intensity adjustment original data set specifically includes a physiological health original data set and a non-physiological health original data set. The physiological health original data set specifically includes basic physiological data of the elderly, metabolic physiological data of the elderly, cardiovascular health data of the elderly, respiratory health data of the elderly and musculoskeletal health data of the elderly. The basic physiological data of the elderly specifically includes age, gender, weight, height and disease information. The non-physiological health original data set specifically includes exercise history data, environmental data and emotional state data of the elderly. The exercise history data of the elderly specifically includes daily exercise volume data, exercise type preference data and exercise performance data.

[0073] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In the data preprocessing module, the data cleaning is specifically to remove missing values ​​and duplicate values, the data encoding is specifically to encode the original data using the one-hot encoding method, and the data normalization is specifically to normalize the original data using the minimum-maximum normalization method. Through the data cleaning, the data encoding and the data normalization, the physiological health original data set and the non-physiological health original data set are preprocessed to obtain the physiological health preliminary data set and the non-physiological health preliminary data set.

[0074] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In the elderly physiological monitoring module, an improved K center point clustering model is constructed to obtain an elderly physiological monitoring model, and the elderly physiological monitoring model is used to preliminarily classify the physiological health preliminary data set. The improved K center point clustering model is specifically a K center point clustering model combined with a harmonious search algorithm.

[0075] The physiological monitoring module for the elderly includes initializing the search algorithm, updating the harmony solution, iterative updating and preliminary classification;

[0076] The initialization search algorithm is used to initialize the harmony search algorithm, and the content includes:

[0077] Determine an iteration termination condition, wherein the iteration termination condition specifically includes reaching a maximum number of iterations and a fitness function value being less than a preset threshold;

[0078] Initialize the harmony search parameters, including the harmony library size Ha, the number of clusters K, and the maximum number of harmony search iterations , the probability pn of adopting a new harmonious solution and the probability pc of changing the solution;

[0079] Initializing the harmony library is specifically to randomly select Ha harmony solutions from the physiological health preliminary data set and initialize the harmony library matrix. The harmony solution is specifically a group of cluster centers of K center point clustering. The harmony library matrix is ​​expressed as follows:

[0080] ;

[0081] Where Hm represents the harmony library matrix, represents the first harmonious solution, represents the second harmonious solution, represents the hath harmonious solution, represents the first dimension of the first harmonious solution, represents the Kth dimension of the first harmonious solution, represents the first dimension of the Hath harmonious solution, represents the Kth dimension of the Hath harmonious solution;

[0082] Determine the fitness function, specifically, select the mean of the average distances from the sample data points in each cluster to the cluster center after clustering as the fitness function value of the harmonious solution. The formula used is as follows:

[0083] ;

[0084] In the formula, represents the fitness function, represents the hath harmonious solution, represents the total number of sample data points in the kth cluster, represents the Euclidean distance calculation function, represents the ath sample data point in the kth cluster, represents the kth cluster center;

[0085] The updated harmonious solution is used to update the harmonious solution of the harmonious search algorithm, and the content includes:

[0086] Generate candidate harmonious solutions, specifically, generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pn of adopting the new harmonious solution. If the generated random value is greater than pn, the original harmonious solution is adopted as the candidate harmonious solution. If the generated random value is less than pn, randomly select the value of the same dimension of other harmonious solutions to replace the value of this dimension of the original harmonious solution as the candidate harmonious solution;

[0087] Generate a new harmonious solution, specifically generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pc of the changed solution. If the generated random value is greater than pc, the candidate harmonious solution is used as the new harmonious solution. If the generated random value is less than pc, a new harmonious solution is generated based on the reference bandwidth and the candidate harmonious solution. The formula used is as follows:

[0088] ;

[0089] In the formula, represents the kth dimension of the hath new harmonious solution, represents the kth dimension of the hath candidate harmonious solution, BW represents the reference bandwidth, Represents a random value in the range [-1,1];

[0090] Since the cluster center of the K-center clustering should be a known sample data point, and the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is an unknown sample data point, the sample data point in the cluster that is closest to the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is selected as the value of this dimension of the new harmonious solution to obtain the new harmonious solution;

[0091] The state parameters of the original harmonious solution and the new harmonious solution are calculated using the following formula:

[0092] ;

[0093] In the formula, represents the growth rate calculation function, represents the hath new harmonious solution, represents the hath original and harmonious solution, represents the calculation function of the state parameters of the new harmonious solution, It represents the calculation function of the original and harmonious solution state parameters, represents the fitness adjustment coefficient, represents the growth rate adjustment factor;

[0094] Update the harmony library matrix, specifically, sort the original harmony solution and the new harmony solution from low to high according to the state parameter, and select the first Ha harmony solutions to form a new harmony library matrix;

[0095] The iterative update specifically includes continuously iterating and updating the algorithm until the iterative termination condition is reached, at which time a group of cluster centers represented by a harmonious solution with the lowest fitness function value is the optimal cluster center combination, and a K center point clustering model is constructed with the optimal cluster center combination as a physiological monitoring model for the elderly;

[0096] The preliminary classification is specifically to perform preliminary classification on the physiological health preliminary data set through the physiological monitoring model for the elderly, and the obtained classification labels are combined with the non-physiological health preliminary data set to obtain the intensity adjustment data set.

[0097] By performing the above operations, this solution creatively uses improved K-center point clustering to perform preliminary classification of the physiological health data of the elderly, avoiding the influence of noise and falling into local optimal solutions, while fully exploring individual health characteristics and converting them into accurate basis for exercise intensity adjustment.

[0098] Example 5, see Figure 1 and Figure 4 , based on the above embodiment, in the exercise intensity recommendation model construction module, this embodiment specifically constructs a bidirectional long short-term memory network model as an exercise intensity recommendation model, and specifically includes labeling and segmenting the data set, constructing a bidirectional long short-term memory layer, designing a cross-level attention mechanism, calculating the model output, and constructing and training the model;

[0099] The labeling and segmenting of the data set specifically includes labeling the data in the intensity adjustment data set as extremely low intensity, low intensity, medium intensity and high intensity, and using the data labels as data labels, and segmenting the labeled intensity adjustment data set into an intensity adjustment training set and an intensity adjustment test set;

[0100] The formula used to construct the bidirectional long short-term memory layer is as follows:

[0101] ;

[0102] In the formula, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the forward long short-term memory function, represents the model input at time step t, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t-1, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the backward long short-term memory function, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t+1, represents the hidden state of the l-th bidirectional long short-term memory layer at time step t, Represents a splicing operation;

[0103] The design of the cross-level attention mechanism includes:

[0104] To calculate the attention weight, the formula used is as follows:

[0105] ;

[0106] In the formula, Indicates Layer and The attention weight calculation function between the bidirectional long short-term memory layers, represents the similarity score calculation function, Indicates The bidirectional long short-term memory layer is at time step The hidden state of Indicates The bidirectional long short-term memory layer is at time step The hidden state of , T represents the total number of time steps, Indicates The bidirectional long short-term memory layer is at time step The hidden state of represents the softmax function, represents the learnable parameter matrix used to map features to the query space, represents a learnable parameter matrix, which is used to map features to the key space, and Tr represents the transposition operation;

[0107] Weighted feature fusion, the formula used is as follows:

[0108] ;

[0109] In the formula, Indicates Layer and The bidirectional long short-term memory layer is at time step The fusion characteristics of

[0110] The formula used for inter-layer feature fusion is as follows:

[0111] ;

[0112] In the formula, represents the final fusion feature, L represents the total number of bidirectional long short-term memory layers;

[0113] The calculation model output is as follows:

[0114] ;

[0115] In the formula, represents the predicted value of the model output, represents the output weight, represents the output bias term;

[0116] The model building and training are specifically to build a bidirectional long short-term memory network model by building a bidirectional long short-term memory layer, designing a cross-level attention mechanism and calculating the model output, and training the model based on the intensity adjustment training set, and verifying the model performance based on the intensity adjustment test set. The model loss function adopts a cross entropy loss function to obtain an exercise intensity recommendation model.

[0117] By performing the above operations, this solution creatively adopts a bidirectional long short-term memory network model as an exercise intensity recommendation model to address the technical problems that traditional methods of regulating exercise intensity for the elderly lack a personalized adjustment mechanism, fail to reflect the dynamic changes in the health status of the elderly in real time, lack in-depth mining and analysis of historical exercise data, and cannot effectively optimize exercise intensity and improve the effect of exercise intervention. It can capture the time dependence of physiological responses during exercise, take into account the long-term changes in the health status of the elderly and the timeliness of historical exercise performance, and thus provide more accurate, continuous and individually tailored recommendations.

[0118] Example 6, see Figure 1 This embodiment is based on the above embodiment. In the exercise intensity adjustment module, the exercise intensity is recommended through the exercise intensity recommendation model to obtain fitness exercise intensity recommendation results, and the fitness exercise intensity suitable for the elderly is comprehensively analyzed based on the fitness exercise intensity recommendation results to formulate a personalized fitness exercise plan.

[0119] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0120] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0121] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. An intelligent system for adjusting the intensity of fitness exercise for the elderly, characterized by: The system includes a data acquisition module, a data preprocessing module, an elderly physiological monitoring module, an exercise intensity recommendation model building module and an exercise intensity adjustment module; The data acquisition module obtains the intensity-adjusted original data set through data acquisition; The data preprocessing module obtains a physiological health preliminary data set and a non-physiological health preliminary data set through data preprocessing; The elderly physiological monitoring module constructs an improved K-center clustering model and performs preliminary classification of physiological health data to obtain an intensity adjustment data set; The exercise intensity recommendation model construction module constructs a bidirectional long short-term memory network model as an exercise intensity recommendation model; The exercise intensity adjustment module obtains a fitness exercise intensity recommendation result by making an exercise intensity recommendation, and comprehensively adjusts the fitness exercise intensity of the elderly based on the fitness exercise intensity recommendation result; In the physiological monitoring module for the elderly, the physiological monitoring model for the elderly is obtained by constructing an improved K-center point clustering model, and the physiological monitoring model for the elderly is used to preliminarily classify the preliminary physiological health data set. The improved K-center point clustering model is specifically a K-center point clustering model combined with a harmonious search algorithm.

2. The intelligent elderly fitness exercise intensity adjustment system according to claim 1 is characterized by: The physiological monitoring module for the elderly includes initializing the search algorithm, updating the harmony solution, iterative updating and preliminary classification; The initialization search algorithm is used to initialize the harmony search algorithm, and the content includes: Determine an iteration termination condition, wherein the iteration termination condition specifically includes reaching a maximum number of iterations and a fitness function value being less than a preset threshold; Initialize the harmony search parameters, including the harmony library size Ha, the number of clusters K, and the maximum number of harmony search iterations , the probability pn of adopting a new harmonious solution and the probability pc of changing the solution; Initializing the harmony library is specifically to randomly select Ha harmony solutions from the physiological health preliminary data set and initialize the harmony library matrix. The harmony solution is specifically a group of cluster centers of K center point clustering. The harmony library matrix is ​​expressed as follows: ； Where Hm represents the harmony library matrix, represents the first harmonious solution, represents the second harmonious solution, represents the hath harmonious solution, represents the first dimension of the first harmonious solution, represents the Kth dimension of the first harmonious solution, represents the first dimension of the Hath harmonious solution, represents the Kth dimension of the Hath harmonious solution; Determine the fitness function, specifically, select the mean of the average distances from the sample data points in each cluster to the cluster center after clustering as the fitness function value of the harmonious solution. The formula used is as follows: ； In the formula, represents the fitness function, represents the hath harmonious solution, represents the total number of sample data points in the kth cluster, represents the Euclidean distance calculation function, represents the ath sample data point in the kth cluster, represents the kth cluster center; The updated harmonious solution is used to update the harmonious solution of the harmonious search algorithm, and the content includes: Generate candidate harmonious solutions, specifically, generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pn of adopting the new harmonious solution. If the generated random value is greater than pn, the original harmonious solution is adopted as the candidate harmonious solution. If the generated random value is less than pn, randomly select the value of the same dimension of other harmonious solutions to replace the value of this dimension of the original harmonious solution as the candidate harmonious solution; Generate a new harmonious solution, specifically generate a random value in the range [0,1] for each dimension of each harmonious solution, and compare it with the probability pc of the changed solution. If the generated random value is greater than pc, the candidate harmonious solution is used as the new harmonious solution. If the generated random value is less than pc, a new harmonious solution is generated based on the reference bandwidth and the candidate harmonious solution. The formula used is as follows: ； In the formula, represents the kth dimension of the hath new harmonious solution, represents the kth dimension of the hath candidate harmonious solution, BW represents the reference bandwidth, Represents a random value in the range [-1,1]; Since the cluster center of the K-center clustering should be a known sample data point, and the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is an unknown sample data point, the sample data point in the cluster that is closest to the value of the dimension of the new harmonious solution generated by the reference bandwidth and the candidate harmonious solution is selected as the value of this dimension of the new harmonious solution to obtain the new harmonious solution; The state parameters of the original harmonious solution and the new harmonious solution are calculated using the following formula: ； In the formula, represents the growth rate calculation function, represents the hath new harmonious solution, represents the hath original and harmonious solution, represents the calculation function of the state parameters of the new harmonious solution, It represents the calculation function of the original and harmonious solution state parameters, represents the fitness adjustment coefficient, represents the growth rate adjustment factor; Update the harmony library matrix, specifically, sort the original harmony solution and the new harmony solution from low to high according to the state parameter, and select the first Ha harmony solutions to form a new harmony library matrix; The iterative update specifically includes continuously iterating and updating the algorithm until the iterative termination condition is reached, at which time a group of cluster centers represented by a harmonious solution with the lowest fitness function value is the optimal cluster center combination, and a K center point clustering model is constructed with the optimal cluster center combination as a physiological monitoring model for the elderly; The preliminary classification is specifically to perform preliminary classification on the physiological health preliminary data set through the physiological monitoring model for the elderly, and the obtained classification labels are combined with the non-physiological health preliminary data set to obtain the intensity adjustment data set.

3. The intelligent elderly fitness exercise intensity adjustment system according to claim 1 is characterized by: In the exercise intensity recommendation model construction module, a bidirectional long short-term memory network model is constructed as an exercise intensity recommendation model. The specific contents include labeling and segmenting the data set, building a bidirectional long short-term memory layer, designing a cross-level attention mechanism, calculating the model output, and building and training the model. The labeling and segmenting of the data set specifically includes labeling the data in the intensity adjustment data set as extremely low intensity, low intensity, medium intensity and high intensity, and using the data labels as data labels, and segmenting the labeled intensity adjustment data set into an intensity adjustment training set and an intensity adjustment test set; The formula used to construct the bidirectional long short-term memory layer is as follows: ； In the formula, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the forward long short-term memory function, represents the model input at time step t, represents the forward hidden state of the l-th bidirectional long short-term memory layer at time step t-1, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t, represents the backward long short-term memory function, represents the backward hidden state of the l-th bidirectional long short-term memory layer at time step t+1, represents the hidden state of the l-th bidirectional long short-term memory layer at time step t, Represents a splicing operation; The design of the cross-level attention mechanism includes: To calculate the attention weight, the formula used is as follows: ； In the formula, Indicates Layer and The attention weight calculation function between the bidirectional long short-term memory layers, represents the similarity score calculation function, Indicates The bidirectional long short-term memory layer is at time step The hidden state of Indicates The bidirectional long short-term memory layer is at time step The hidden state of , T represents the total number of time steps, Indicates The bidirectional long short-term memory layer is at time step The hidden state of represents the softmax function, represents the learnable parameter matrix used to map features to the query space, represents a learnable parameter matrix, which is used to map features to the key space, and Tr represents the transposition operation; Weighted feature fusion, the formula used is as follows: ； In the formula, Indicates Layer and The bidirectional long short-term memory layer is at time step The fusion characteristics of The formula used for inter-layer feature fusion is as follows: ； In the formula, represents the final fusion feature, L represents the total number of bidirectional long short-term memory layers; The calculation model output is as follows: ； In the formula, represents the predicted value of the model output, represents the output weight, represents the output bias term; The model building and training are specifically to build a bidirectional long short-term memory network model by building a bidirectional long short-term memory layer, designing a cross-level attention mechanism and calculating the model output, and training the model based on the intensity adjustment training set, and verifying the model performance based on the intensity adjustment test set. The model loss function adopts a cross entropy loss function to obtain an exercise intensity recommendation model.

4. The intelligent elderly fitness exercise intensity adjustment system according to claim 1 is characterized by: In the data acquisition module, the intensity adjustment original data set specifically includes a physiological health original data set and a non-physiological health original data set. The physiological health original data set specifically includes basic physiological data of the elderly, metabolic physiological data of the elderly, cardiovascular health data of the elderly, respiratory health data of the elderly and musculoskeletal health data of the elderly. The basic physiological data of the elderly specifically includes age, gender, weight, height and disease information. The non-physiological health original data set specifically includes exercise history data of the elderly, environmental data and emotional state data of the elderly. The exercise history data of the elderly specifically includes daily exercise volume data, exercise type preference data and exercise performance data.

5. The intelligent elderly fitness exercise intensity adjustment system according to claim 1 is characterized by: In the data preprocessing module, the data cleaning is specifically to remove missing values ​​and duplicate values, the data encoding is specifically to encode the original data using the one-hot encoding method, and the data normalization is specifically to normalize the original data using the minimum-maximum normalization method. Through the data cleaning, the data encoding and the data normalization, the physiological health original data set and the non-physiological health original data set are preprocessed to obtain the physiological health preliminary data set and the non-physiological health preliminary data set.

6. The intelligent elderly fitness exercise intensity adjustment system according to claim 1 is characterized by: In the exercise intensity adjustment module, the exercise intensity is recommended through the exercise intensity recommendation model to obtain fitness exercise intensity recommendation results, and based on the fitness exercise intensity recommendation results, a comprehensive analysis of the fitness exercise intensity suitable for the elderly is performed to formulate a personalized fitness exercise plan.

Images