Intelligent control circuit board and control method for treadmill

Abstract

The present application discloses an intelligent control circuit board and control method for a treadmill, which monitors and collects the running speed of the treadmill and the heart rate value of the athlete in real time through a sensor group, and inputs these data into the intelligent control circuit board, so as to analyze and interactively process the running speed time series data and the heart rate time series data by using data processing and analysis algorithms based on artificial intelligence and deep learning in the intelligent control circuit board, so as to learn and capture the time series correlation relationship and interactive influence between the running speed and the athlete's heart rate, so as to adaptively control the slope of the treadmill by using the interactive fusion semantics of the two. In this way, not only can a more intelligent sports experience be provided to the user according to the user's ability to meet the user's personalized needs, but also the user can be prevented from excessive exercise and causing safety accidents, thereby increasing the safety of the treadmill.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to an intelligent control circuit board and a control method for a treadmill. Background Art

[0002] Running, as an efficient aerobic exercise, is becoming more and more popular around the world. With the accelerated pace of life and the pursuit of a healthy lifestyle, treadmills, as a convenient indoor exercise equipment, have gradually become an indispensable part of homes and gyms. The use of indoor treadmills is not restricted by weather, time, or location. Users can run at any time according to their own schedule and exercise needs, providing users with a stable running environment.

[0003] At present, the main functions of treadmills are concentrated on providing a sports platform, a drive device, and simple data monitoring. Some high-end treadmills are also equipped with touch screens that can display sports data, play music or videos, etc. However, existing treadmills can usually only provide a few fixed sports modes, such as jogging, sprinting, etc., and the interaction and adjustment of the treadmill are completed through physical buttons or touch screens, which cannot meet the personalized sports needs of users, nor can they provide a smarter and more personalized sports experience. In addition, in some cases, traditional treadmills may not be able to respond to abnormal conditions of users in a timely manner, such as excessive fatigue or abnormally high heart rate, thereby increasing the risk of exercise.

[0004] Therefore, an optimized control scheme for a treadmill is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent control circuit board and control method for a treadmill, which monitors and collects the running speed of the treadmill and the heart rate value of the athlete in real time through a sensor group, and inputs these data into the intelligent control circuit board, so as to use the data processing and analysis algorithm based on artificial intelligence and deep learning in the intelligent control circuit board to analyze and interactively process the running speed time series data and the heart rate time series data, so as to learn and capture the time series correlation relationship and interactive influence between the running speed and the athlete's heart rate, so as to use the interactive fusion semantics of the two to adaptively control the slope of the treadmill. In this way, it can not only provide users with a more intelligent sports experience according to their abilities to meet the personalized needs of users, but also avoid safety accidents caused by excessive user exercise, thereby increasing the safety of the treadmill.

[0006] According to one aspect of the present application, a control method for an intelligent control circuit board for a treadmill is provided, comprising:

[0007] Obtain the time queue of the running speed and the time queue of the heart rate value collected by the sensor group;

[0008] Performing time series coding on the time series of the running speed and the time series of the heart rate value to obtain a sequence of running speed local time series associated implicit feature vectors and a sequence of heart rate local time series associated implicit feature vectors;

[0009] Inputting the sequence of the running speed local time series associated implicit feature vectors and the sequence of the heart rate local time series associated implicit feature vectors into a node feature propagation network based on a node energy decay mechanism to obtain a running speed time series propagation aggregate representation vector and a heart rate time series propagation aggregate representation vector;

[0010] Performing feature interaction response processing on the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector to obtain a significant running speed-heart rate time series interaction fusion representation vector;

[0011] Based on the significant running speed-heart rate time series interaction fusion representation vector, a control instruction is generated, where the control instruction is used to indicate increasing the treadmill slope value, decreasing the treadmill slope value, and maintaining the treadmill slope value.

[0012] Compared with the prior art, the present application provides an intelligent control circuit board and control method for a treadmill, which monitors and collects the running speed of the treadmill and the heart rate value of the athlete in real time through a sensor group, and inputs these data into the intelligent control circuit board, so as to use the data processing and analysis algorithm based on artificial intelligence and deep learning in the intelligent control circuit board to analyze and interactively process the running speed time series data and the heart rate time series data, so as to learn and capture the time series correlation relationship and interactive influence between the running speed and the athlete's heart rate, so as to use the interactive fusion semantics of the two to adaptively control the slope of the treadmill. In this way, not only can a more intelligent sports experience be provided to the user according to the user's ability to meet the user's personalized needs, but also the user can be prevented from excessive exercise and causing safety accidents, thereby increasing the safety of the treadmill. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 It is a flow chart of a control method of an intelligent control circuit board for a treadmill according to an embodiment of the present application;

[0015] Figure 2 A data flow diagram of a control method for an intelligent control circuit board for a treadmill according to an embodiment of the present application;

[0016] Figure 3 The present invention is a flowchart of the training phase of the control method of the intelligent control circuit board for a treadmill according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0018] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0019] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0020] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0021] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0022] Existing treadmills usually only provide a few fixed exercise modes, such as jogging, sprinting, etc., and the interaction and adjustment of the treadmill are completed through physical buttons or touch screens, which cannot meet the personalized exercise needs of users, nor can they provide a smarter and more personalized exercise experience. In addition, in some cases, traditional treadmills may not be able to respond to abnormal conditions of users in a timely manner, such as excessive fatigue or abnormally high heart rate, thereby increasing exercise risks. Therefore, an optimized control solution for treadmills is desired.

[0023] In the technical solution of the present application, a control method for an intelligent control circuit board for a treadmill is proposed. Figure 1 It is a flow chart of a control method of an intelligent control circuit board for a treadmill according to an embodiment of the present application;

[0024] Figure 2 1 is a data flow diagram of a control method for an intelligent control circuit board for a treadmill according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to an embodiment of the present application, the control method for an intelligent control circuit board for a treadmill includes the following steps: S1, obtaining a time queue of a running speed and a time queue of a heart rate value collected by a sensor group; S2, performing time series encoding on the time queue of the running speed and the time queue of the heart rate value to obtain a sequence of local time series associated implicit feature vectors of the running speed and a sequence of local time series associated feature vectors of the heart rate; S3, inputting the sequence of local time series associated implicit feature vectors of the running speed and the sequence of local time series associated feature vectors of the heart rate into a node feature propagation network based on a node energy decay mechanism to obtain a running speed time series propagation aggregate representation vector and a heart rate time series propagation aggregate representation vector; S4, performing feature interaction response processing on the running speed time series propagation aggregate representation vector and the heart rate time series propagation aggregate representation vector to obtain a significant running speed-heart rate time series interaction fusion representation vector; S5, generating a control instruction based on the significant running speed-heart rate time series interaction fusion representation vector, the control instruction being used to indicate increasing the treadmill slope value, decreasing the treadmill slope value and maintaining the treadmill slope value.

[0025] In particular, the S1 and S2 obtain the time queue of the running speed and the time queue of the heart rate value collected by the sensor group; and perform time series encoding on the time queue of the running speed and the time queue of the heart rate value to obtain a sequence of local time series associated implicit feature vectors of the running speed and a sequence of local time series implicit associated feature vectors of the heart rate. In a specific example of the present application, the time queue of the running speed and the time queue of the heart rate value are input into a sequence encoder based on a 1D-CNN model to obtain a sequence of local time series associated implicit feature vectors of the running speed and a sequence of local time series implicit associated feature vectors of the heart rate. That is, the time queue of the running speed and the time queue of the heart rate value are input into a sequence encoder based on a 1D-CNN model for feature mining, so as to extract the local time series implicit associated feature information of the running speed and the heart rate value in the time dimension respectively, thereby obtaining a sequence of local time series associated implicit feature vectors of the running speed and a sequence of local time series implicit associated feature vectors of the heart rate. It is worth mentioning that 1D-CNN is a one-dimensional convolutional neural network suitable for processing sequence data, such as time series data or text data.

[0026] In particular, in S3, the sequence of the running speed local time series associated implicit feature vectors and the sequence of the heart rate local time series associated feature vectors are input into the node feature propagation network based on the node energy decay mechanism to obtain the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector. It should be understood that for the running speed, its change in the time dimension is a time series process, which has time continuity and correlation. The local time series associated implicit feature vectors of each running speed capture the local mode and trend of speed change in the local time period during the running process, and there are correlations and features based on the global time series between the running speed time series features in these different local time periods. Therefore, in order to be able to fuse these running speed local time series features to form a more comprehensive representation of the running speed time series change, in the technical solution of the present application, the sequence of the running speed local time series associated implicit feature vectors is further input into the node feature propagation network based on the node energy decay mechanism to obtain the running speed time series propagation aggregation representation vector. Through the processing of the node feature propagation network based on the node energy decay mechanism, the node energy decay process between the local time series features of the running speed in each local time period can be simulated, and the time-space aggregation representation of the node features can be realized, thereby improving the model's ability to understand the full time domain characteristics and change patterns of the running speed.

[0027] Specifically, the node feature propagation network based on the node energy decay mechanism uses the implicit feature vectors associated with each local running speed time series as the node feature vector to calculate the energy statistical paradigm value of the node feature vector sequence, wherein the energy statistical paradigm value is related to the maximum value, average value and variance of the node feature vector, which is used to quantify the energy level of the node feature. Subsequently, by counting the propagation space span values ​​between different node feature vectors, the topological structure and spatial relationship between nodes, that is, the node association characteristics between the implicit feature vectors associated with each local running speed time series, are captured. These span values ​​are combined with the node energy statistical paradigm values ​​of the historical running speed local time series to determine the node energy propagation attenuation coefficient of the running speed local time series, which is inversely correlated with the historical node energy of the running speed local time series, simulating the attenuation characteristics of the node energy of each running speed local time series over time and space. Then, using these running speed local time series attenuation coefficients as weights (that is, considering the energy attenuation effect), the weighted sum of the node feature vector sequence is further calculated based on the energy statistical paradigm value of the running speed local time series to generate the historical node energy attenuation time series aggregation feature vector of the running speed local time series to aggregate the historical node feature information of the running speed local time series. Finally, by fusing the energy statistical paradigm value of the current node of the running speed local time series, the weighted sum of the historical running speed local time series aggregation feature vector and the current running speed local time series node feature vector is calculated to generate the node energy attenuation time series spatial aggregation representation vector of the running speed local time series, which integrates the information of time series dynamics and spatial structure. In summary, the node feature propagation network based on the node energy decay mechanism dynamically evaluates the energy level of each running speed local time series node feature, captures the topological structure between nodes, simulates the energy propagation attenuation, and aggregates the time series spatial features. This method has significant advantages in improving the model's sensitivity to the time series dynamics of the running speed local time series node features, enhancing feature expression, and adapting to complex network structures. In addition, by dynamically adjusting the node energy weights and aggregating temporal spatial features, the model can more accurately identify and process the key local temporal information of running speed in the data, thereby achieving better performance in subsequent treadmill adaptive control tasks.

[0028] In an embodiment of the present application, the sequence of the running speed local time series associated implicit feature vectors and the sequence of the heart rate local time series associated implicit feature vectors are input into a node feature propagation network based on a node energy decay mechanism to obtain a running speed time series propagation aggregate representation vector and a heart rate time series propagation aggregate representation vector, including: based on the maximum value, average value and variance of each running speed local time series associated implicit feature vector in the sequence of the running speed local time series associated implicit feature vectors, the node energy statistical paradigm value of each running speed local time series associated implicit feature vector is calculated to obtain the running speed local time series associated implicit feature vector. The energy statistical paradigm value of the local time series node of the running speed is obtained by calculating the energy statistical paradigm value of the local time series node of the running speed, wherein the energy statistical paradigm value of the local time series node of the running speed corresponding to the implicit feature vector of the current running speed local time series in the sequence of the energy statistical paradigm value of the local time series node of the running speed is used as the energy statistical paradigm value of the current node, and the energy statistical paradigm value of other local time series nodes of the running speed is used as the historical node energy statistical paradigm value to obtain the sequence of the energy statistical paradigm value of the local time series node of the current running speed and the energy statistical paradigm value of the historical running speed local time series node; and the energy statistical paradigm value of each other running speed in the sequence of the implicit feature vector of the running speed local time series is counted. The node propagation space span value between the local time series associated implicit feature vector of the running speed and the current local time series associated implicit feature vector of the running speed is determined to obtain a sequence of local time series node propagation space span values ​​of the running speed; based on the sequence of local time series node propagation space span values ​​of the running speed and the sequence of historical local time series node energy statistical paradigm values ​​of the running speed, the node energy propagation attenuation coefficient values ​​of other local time series associated implicit feature vectors of the running speed in the sequence of local time series associated implicit feature vectors of the running speed are determined to obtain a sequence of local time series node energy propagation attenuation coefficient values ​​of the running speed; and the node propagation space span value between the local time series associated implicit feature vector of the running speed and the current local time series associated implicit feature vector of the running speed is determined to obtain a sequence of local time series node energy propagation attenuation coefficient values ​​of the running speed. The sequence of the running speed local time series node energy propagation attenuation coefficient values ​​is used as a weight sequence, and the weighted sum of all other running speed local time series associated implicit feature vectors in the sequence of the running speed local time series associated implicit feature vectors is calculated to obtain a historical running speed local time series node energy attenuation time series aggregated feature vector; based on the current running speed local time series node energy statistical paradigm value, the weighted sum of the historical running speed local time series node energy attenuation time series aggregated feature vector and the current running speed local time series associated implicit feature vector is calculated to obtain the running speed time series propagation aggregated representation vector.

[0029] The process of calculating the node energy statistical paradigm value of each running speed local time series associated implicit feature vector based on the maximum value, average value and variance of each running speed local time series associated implicit feature vector in the sequence of the running speed local time series associated implicit feature vector to obtain a sequence of running speed local time series node energy statistical paradigm values ​​includes: calculating the maximum value, average value and variance of the running speed local time series associated implicit feature vector to obtain the running speed local time series maximum value, the running speed local time series average value and the running speed local time series variance; calculating the sum of the running speed local time series average value and the running speed local time series variance and then multiplying the sum by a constant 4 to obtain the first running speed local time series node energy statistical factor; respectively calculating the product of the running speed local timing average value and the running speed local timing variance and a constant 2 to obtain the double-modulated running speed local timing average value and the double-modulated running speed local timing variance; calculating the square of the difference between the running speed local timing maximum value and the running speed local timing average value, and then adding it to the double-modulated running speed local timing average value and the double-modulated running speed local timing variance to obtain a second running speed local timing node energy statistical factor; calculating the division between the first running speed local timing node energy statistical factor and the second running speed local timing node energy statistical factor to obtain a running speed local timing node energy statistical paradigm value. And, based on the sequence of the running speed local time series node propagation space span values ​​and the sequence of the historical running speed local time series node energy statistical paradigm values, determining the node energy propagation attenuation coefficient values ​​of each other running speed local time series associated implicit feature vector in the sequence of the running speed local time series associated implicit feature vector to obtain the sequence of the running speed local time series node energy propagation attenuation coefficient values ​​includes: using each running speed local time series node propagation space span value in the sequence of the running speed local time series node propagation space span value as an exponential power, calculating the natural constant e as the base The exponential function value is used to obtain the sequence of space span values ​​supported by the running speed local time series node propagation class; the positional sum of the sequence of space span values ​​supported by the running speed local time series node propagation class and the sequence of space span values ​​of the running speed local time series node propagation is calculated to obtain the sequence of space span modulation coefficients of the running speed local time series node propagation; the positional division between the sequence of energy statistical paradigm values ​​of the historical running speed local time series node and the sequence of space span modulation coefficients of the running speed local time series node propagation is calculated to obtain the sequence of energy propagation attenuation coefficient values ​​of the running speed local time series node.

[0030] In summary, in the above embodiment, the sequence of the running speed local temporal association implicit feature vectors and the sequence of the heart rate local temporal association implicit feature vectors are input into the node feature propagation network based on the node energy decay mechanism to obtain the running speed temporal propagation aggregate representation vector and the heart rate temporal propagation aggregate representation vector, including: inputting the sequence of the running speed local temporal association implicit feature vectors into the node feature propagation network based on the node energy decay mechanism and processing them with the following node feature propagation formula to obtain the running speed temporal propagation aggregate representation vector; wherein the node feature propagation formula is:

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] in, is a sequence of implicit feature vectors associated with the local temporal sequence of running speed, is the sequence of the running speed local temporal association implicit feature vectors. running speed local temporal correlation implicit feature vector, For the The running speed local temporal correlation latent feature vector Position feature values, For the The length of the hidden feature vector of the local temporal association of running speed, For the The mean of the latent feature vectors of the local temporal association of running speed, For the The variance of the latent eigenvector of the local temporal correlation of running speed, is the regularization hyperparameter, Indicates The maximum value of the hidden feature vector of the local temporal association of running speed, For the The node energy statistical norm value of the running speed local time series associated implicit feature vector, Indicates the node propagation space span value, , , and is a trainable hyperparameter, A representation vector is aggregated for the running speed temporal propagation.

[0037] Similarly, the sequence of the implicitly associated feature vectors of the local time series of heart rate is also input into the node feature propagation network based on the node energy decay mechanism to obtain the heart rate time series propagation aggregate representation vector, relying on capturing the feature propagation aggregate representation information between the local time series features of the heart rate.

[0038] In particular, the S4 performs feature interaction response processing on the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector to obtain a significant running speed-heart rate time series interaction fusion representation vector. Since running speed and heart rate are two important physiological indicators, they respectively include the response of exercise intensity and cardiovascular system, and they can reflect the user's exercise intensity, endurance level and physical condition. Therefore, in order to provide more comprehensive user exercise state information and physiological state information, so as to adaptively adjust the treadmill slope according to the user's personalized needs and abilities, in the technical solution of the present application, the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector are further input into a feature interaction response module based on an adaptive distinguishable mechanism to obtain a significant running speed-heart rate time series interaction fusion representation vector. Through the processing of the feature interaction response module based on the adaptive distinguishable mechanism, the expression ability of the feature and the overall performance of the model can be significantly improved through fine analysis and adaptive weight adjustment of each position. This can realize the feature interaction between running speed and heart rate to discover the potential temporal association and influence between the two. This helps reveal the dynamic relationship between running speed and heart rate, providing a more in-depth and detailed data analysis basis for intelligent control systems.

[0039] Specifically, firstly, by calculating the position-by-position response between the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector, a running speed-heart rate time series position-by-position response feature vector is generated. This step effectively captures the local temporal feature interaction between the running speed time series aggregation information and the heart rate time series aggregation information. Subsequently, the running speed-heart rate time series position-by-position responses are normalized using the Softmax function to form a probability distribution, ensuring the standardization of the feature response and providing preparation for the subsequent gating function weight screening. The normalization process not only balances the scale of the feature response, but also makes the expression of the running speed-heart rate time series features in the model more balanced. By inputting the normalized running speed-heart rate time series position-by-position response feature vector into the learnable gating function, the model can adaptively learn and output the running speed-heart rate time series response screening weight mask vector. This dynamic weighting process realizes the temporal feature selection and reinforcement between running speed and heart rate through the gating mechanism, and dynamically adjusts the importance of the feature position. This step is the core of feature optimization because it directly affects the sensitivity and distinguishing ability of the model to the running speed-heart rate timing feature response. Further, by calculating the positional dot product between the running speed-heart rate timing response screening weight mask vector and the normalized running speed-heart rate timing positional response feature vector, a running speed-heart rate timing positional response distinguishable weight mask vector is obtained. This operation further refines the weight assignment and provides precise weight control for the final running speed-heart rate timing interaction fusion saliency characterization and optimization. This position-by-position weight refinement ensures that the model can assign the most appropriate weight to the position of each running speed-heart rate timing responsiveness feature, thereby more effectively capturing useful information in the data. Finally, by performing a dot product of the running speed-heart rate timing positional response distinguishable weight mask vector and the running speed-heart rate timing positional response feature vector, a significant running speed-heart rate timing interaction fusion representation vector is generated. This step integrates the feature responses after weight adjustment and generates the final running speed-heart rate timing interaction fusion representation necklace with enhanced significance expression, which significantly improves the model's ability to capture key running speed-heart rate timing interaction responsiveness information and provides a basis for subsequent treadmill slope adaptive control.

[0040] In summary, in the above embodiment, the running speed timing propagation aggregate representation vector and the heart rate timing propagation aggregate representation vector are input into a feature interaction response module based on an adaptive distinguishable mechanism to obtain the significant running speed-heart rate timing interaction fusion representation vector, including: inputting the running speed timing propagation aggregate representation vector and the heart rate timing propagation aggregate representation vector into the feature interaction response module based on an adaptive distinguishable mechanism to process them with the following feature interaction response formula to obtain the significant running speed-heart rate timing interaction fusion representation vector; wherein the feature interaction response formula is:

[0041]

[0042]

[0043]

[0044]

[0045]

[0046] in, and represent the heart rate time series propagation aggregation representation vector and the running speed time series propagation aggregation representation vector respectively, is the position-by-position response feature vector of the running speed-heart rate time series, express function, is the normalized running speed-heart rate time series position-by-position response feature vector, is the value of the natural exponential function, Filter the weight mask vector for the running speed-heart rate timing response, It means point multiplication by position. is the distinguishable weight mask vector for the position-by-position response of the running speed-heart rate time series, The representation vector of the significant running speed-heart rate time series interaction fusion.

[0047] In particular, the S5 generates a control instruction based on the significant running speed-heart rate timing interactive fusion representation vector, and the control instruction is used to indicate increasing the treadmill slope value, reducing the treadmill slope value and maintaining the treadmill slope value. In particular, in a specific example of the present application, the significant running speed-heart rate timing interactive fusion representation vector is input into a slope controller based on a classifier to obtain a control instruction, and the control instruction is used to indicate increasing the treadmill slope value, reducing the treadmill slope value and maintaining the treadmill slope value. That is, the running speed and heart rate timing interactive fusion features optimized by the significant expression are used for classification processing, so as to adaptively control the slope of the treadmill. In this way, not only can a more intelligent sports experience be provided to the user according to the user's ability to meet the user's personalized needs, but also the user can be prevented from exercising excessively and causing safety accidents, thereby increasing the safety of the treadmill.

[0048] It should be understood that before using the above-mentioned neural network model for inference, it is necessary to train the sequence encoder based on the 1D-CNN model, the node feature propagation network based on the node energy decay mechanism, the feature interaction response module based on the adaptive distinguishable mechanism, and the slope controller based on the classifier. That is to say, in the control method of the intelligent control circuit board for a treadmill of the present application, a training phase is also included for training the sequence encoder based on the 1D-CNN model, the node feature propagation network based on the node energy decay mechanism, the feature interaction response module based on the adaptive distinguishable mechanism, and the slope controller based on the classifier.

[0049] Figure 3 FIG. 1 is a flow chart of the training phase of the control method of the intelligent control circuit board for a treadmill according to an embodiment of the present application. Figure 3As shown, the control method of the intelligent control circuit board for a treadmill according to the embodiment of the present application includes: a training stage, including: S110, obtaining training data, wherein the training data includes a time queue of training running speed and a time queue of training heart rate values ​​collected by a sensor group; S120, inputting the time queue of training running speed and the time queue of training heart rate values ​​into a sequence encoder based on a 1D-CNN model to obtain a sequence of local temporal association implicit feature vectors of training running speed and a sequence of local temporal association implicit feature vectors of training heart rate; S130, inputting the sequence of local temporal association implicit feature vectors of training running speed and the sequence of local temporal association implicit feature vectors of training heart rate into a node feature propagation network based on a node energy decay mechanism to obtain a training running speed temporal propagation aggregate representation vector and a training heart rate temporal propagation aggregate representation vector; S140, inputting the training running speed temporal propagation aggregate representation vector and the training heart rate timing propagation aggregation representation vector are input into the feature interaction response module based on the adaptive distinguishable mechanism to obtain the training significant running speed-heart rate timing interaction fusion representation vector; S150, the training significant running speed-heart rate timing interaction fusion representation vector is passed through the classifier-based slope controller to obtain the classification loss function value; S160, based on the significant running speed-heart rate timing interaction fusion representation loss term, the significant running speed-heart rate timing interaction fusion loss function value is calculated; S170, the weighted sum of the classification loss function value and the significant running speed-heart rate timing interaction fusion loss function value is calculated to obtain the final loss function value; S180, based on the final loss function value, the sequence encoder based on the 1D-CNN model, the node feature propagation network based on the node energy decay mechanism, the feature interaction response module based on the adaptive distinguishable mechanism and the slope controller based on the classifier are trained.

[0050] In particular, preferably, a new loss function value is further introduced in addition to the classification loss function value, wherein the construction of the new loss function value comprises the following steps:

[0051] Calculate the training significant running speed-heart rate timing interaction fusion sum matrix and the training significant running speed-heart rate timing interaction fusion difference matrix based on the training significant running speed-heart rate timing interaction fusion representation vector, wherein the training significant running speed-heart rate timing interaction fusion sum matrix is The characteristic value of the position is the first Eigenvalues ​​and The mean of the eigenvalues, and the training significant running speed-heart rate time series interaction fusion difference matrix The characteristic value of the position is the first Eigenvalues ​​and One-half the absolute value of the difference in eigenvalues;

[0052] Performing query-matrix multiplication of the training significant running speed-heart rate timing interaction fusion representation vector with the training significant running speed-heart rate timing interaction fusion sum matrix and the training significant running speed-heart rate timing interaction fusion difference matrix to obtain a training significant running speed-heart rate timing interaction fusion query sum vector and a training significant running speed-heart rate timing interaction fusion query difference vector;

[0053] Calculate the vector inner product of the training significant running speed-heart rate timing interactive fusion representation and the vector and the training significant running speed-heart rate timing interactive fusion query difference vector to obtain a first significant running speed-heart rate timing interactive fusion representation loss term;

[0054] The training significant running speed-heart rate timing interaction fusion sum matrix and the training significant running speed-heart rate timing interaction fusion difference matrix are matrix-multiplied, and the product matrix is ​​calculated. norm to obtain the second significant running speed-heart rate time series interaction fusion representation loss term; and

[0055] The first significant running speed-heart rate timing interaction fusion representation loss term is subtracted from the product of a predetermined weight hyperparameter and the second significant running speed-heart rate timing interaction fusion representation loss term to obtain the new loss function value.

[0056] The new loss function value, for example, the significant running speed-heart rate time series interaction fusion loss function value, is specifically expressed as:

[0057]

[0058] in

[0059]

[0060]

[0061] in, is the significant running speed-heart rate time series interaction fusion representation vector of the training, and They are respectively the training significant running speed-heart rate timing interaction fusion sum matrix and the training significant running speed-heart rate timing interaction fusion difference matrix, and The training significant running speed-heart rate timing interaction fusion sum matrix and the training significant running speed-heart rate timing interaction fusion difference matrix are respectively The eigenvalues ​​of the positions, and are the first and second representation vectors of the significant running speed-heart rate time series interaction fusion of the training Eigenvalues ​​and Eigenvalues, is matrix multiplication, To calculate the matrix norm, is the predetermined weight hyperparameter, is the loss function value of the significant running speed-heart rate time series interaction fusion.

[0062] Here, considering that the training running speed timing propagation aggregation representation vector and the training heart rate timing propagation aggregation representation vector respectively represent the timing node energy attenuation propagation characteristics of the one-dimensional local timing correlation characteristics of the running speed and heart rate values, therefore, when the training running speed timing propagation aggregation representation vector and the training heart rate timing propagation aggregation representation vector are subjected to feature interaction response based on an adaptive distinguishable mechanism, the training significant running speed-heart rate timing interaction fusion representation vector will also have classification regression recognition difficulties due to the differences in adaptive distinguishable interaction response weights caused by differences in the timing feature distribution of different data, thereby affecting the accuracy of the classification results.

[0063] Therefore, the applicant of the present application performs the query composition of the detail inner product space in the training significant running speed-heart rate time series interaction fusion representation vector through the structured feature representation of the short-range-long-range cross-scale detail link of the training significant running speed-heart rate time series interaction fusion representation vector, so as to approximate the low-rank independent observable composition of the link detail composition provided by the structured detail interaction of the training significant running speed-heart rate time series interaction fusion representation vector, so that by training with the significant running speed-heart rate time series interaction fusion loss function, the distributed detail group of the training significant running speed-heart rate time series interaction fusion representation vector can be used to decompose the detail group on the basis of detail complexity, so as to promote the classification regression decomposition recognition of the training significant running speed-heart rate time series interaction fusion representation vector based on complex feature representation, and improve the accuracy of the control instructions obtained by the slope controller based on the classifier. In this way, the adaptive controllability of the treadmill can be performed more accurately, thereby improving the intelligent degree of treadmill control, providing users with a more intelligent sports experience, so as to meet the personalized needs of users, and at the same time, it can avoid safety accidents caused by excessive user exercise, thereby increasing the safety of the treadmill.

[0064] In summary, the control method of the intelligent control circuit board for a treadmill according to the embodiment of the present application is explained, which monitors and collects the running speed of the treadmill and the heart rate value of the athlete in real time through a sensor group, and inputs these data into the intelligent control circuit board, so as to use the data processing and analysis algorithm based on artificial intelligence and deep learning in the intelligent control circuit board to analyze and interactively process the running speed time series data and the heart rate time series data, so as to learn and capture the time series correlation relationship and interactive influence between the running speed and the athlete's heart rate, so as to use the interactive fusion semantics of the two to adaptively control the slope of the treadmill. In this way, not only can a more intelligent exercise experience be provided to the user according to the user's ability to meet the user's personalized needs, but also it can avoid safety accidents caused by excessive exercise of the user, thereby increasing the safety of the treadmill.

[0065] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A control method for an intelligent control circuit board of a treadmill, characterized in that: include: Obtain the time queue of the running speed and the time queue of the heart rate value collected by the sensor group; Performing time series coding on the time series of the running speed and the time series of the heart rate value to obtain a sequence of running speed local time series associated implicit feature vectors and a sequence of heart rate local time series associated implicit feature vectors; Inputting the sequence of the running speed local time series associated implicit feature vectors and the sequence of the heart rate local time series associated implicit feature vectors into a node feature propagation network based on a node energy decay mechanism to obtain a running speed time series propagation aggregate representation vector and a heart rate time series propagation aggregate representation vector; Performing feature interaction response processing on the running speed time series propagation aggregation representation vector and the heart rate time series propagation aggregation representation vector to obtain a significant running speed-heart rate time series interaction fusion representation vector; Based on the significant running speed-heart rate time series interaction fusion representation vector, a control instruction is generated, where the control instruction is used to indicate increasing the treadmill slope value, decreasing the treadmill slope value, and maintaining the treadmill slope value.

2. The control method of the intelligent control circuit board for a treadmill according to claim 1, characterized in that: The time queue of the running speed and the time queue of the heart rate value are time-series encoded to obtain a sequence of local time-series associated implicit feature vectors of the running speed and a sequence of local time-series associated implicit feature vectors of the heart rate, including: inputting the time queue of the running speed and the time queue of the heart rate value into a sequence encoder based on a 1D-CNN model to obtain a sequence of local time-series associated implicit feature vectors of the running speed and a sequence of local time-series associated implicit feature vectors of the heart rate.

3. The control method of the intelligent control circuit board for a treadmill according to claim 2, characterized in that: Inputting the sequence of the running speed local time series associated implicit feature vectors and the sequence of the heart rate local time series associated implicit feature vectors into a node feature propagation network based on a node energy decay mechanism to obtain a running speed time series propagation aggregate representation vector and a heart rate time series propagation aggregate representation vector, including: Based on the maximum value, average value and variance of each running speed local time series associated implicit feature vector in the sequence of running speed local time series associated implicit feature vectors, the node energy statistical paradigm value of each running speed local time series associated implicit feature vector is calculated to obtain a sequence of running speed local time series node energy statistical paradigm values, wherein the running speed local time series node energy statistical paradigm value corresponding to the current running speed local time series associated implicit feature vector in the sequence of running speed local time series node energy statistical paradigm values ​​is used as the current node energy statistical paradigm value, and other running speed local time series node energy statistical paradigm values ​​are used as historical node energy statistical paradigm values ​​to obtain a sequence of current running speed local time series node energy statistical paradigm value and historical running speed local time series node energy statistical paradigm values; Counting the node propagation space span values ​​between each other running speed local time series associated implicit feature vector in the sequence of running speed local time series associated implicit feature vectors and the current running speed local time series associated implicit feature vector to obtain a sequence of running speed local time series node propagation space span values; Based on the sequence of running speed local time series node propagation space span values ​​and the sequence of historical running speed local time series node energy statistical paradigm values, determining node energy propagation attenuation coefficient values ​​of other running speed local time series associated implicit feature vectors in the sequence of running speed local time series associated implicit feature vectors to obtain a sequence of running speed local time series node energy propagation attenuation coefficient values; Taking the sequence of energy propagation attenuation coefficient values ​​of the running speed local time series nodes as a weight sequence, calculating the weighted sum of all other running speed local time series associated implicit feature vectors in the sequence of the running speed local time series associated implicit feature vectors to obtain a historical running speed local time series node energy attenuation time series aggregated feature vector; The weighted sum of the historical running speed local time series node energy decay time series aggregation feature vector and the current running speed local time series associated implicit feature vector is calculated based on the current running speed local time series node energy statistical norm value to obtain the running speed time series propagation aggregation representation vector.

4. The control method of the intelligent control circuit board for a treadmill according to claim 3, characterized in that: Based on the maximum value, average value and variance of each running speed local time series associated implicit feature vector in the sequence of the running speed local time series associated implicit feature vector, the node energy statistical paradigm value of each running speed local time series associated implicit feature vector is calculated to obtain a sequence of running speed local time series node energy statistical paradigm values, including: Calculating the maximum value, average value and variance of the running speed local time series associated implicit feature vector to obtain the running speed local time series maximum value, the running speed local time series average value and the running speed local time series variance; Calculate the sum of the running speed local time series average value and the running speed local time series variance, and then multiply the sum by a constant 4 to obtain a first running speed local time series node energy statistical factor; Respectively calculating the product of the local time series average value of the running speed and the local time series variance of the running speed and a constant 2 to obtain the local time series average value of the double-modulated running speed and the local time series variance of the double-modulated running speed; After calculating the square of the difference between the running speed local time series maximum value and the running speed local time series average value, the square is added to the double-modulated running speed local time series average value and the double-modulated running speed local time series variance to obtain a second running speed local time series node energy statistical factor; The division between the first running speed local timing node energy statistics factor and the second running speed local timing node energy statistics factor is calculated to obtain a running speed local timing node energy statistics paradigm value.

5. The control method of the intelligent control circuit board for a treadmill according to claim 4, characterized in that: Based on the sequence of the running speed local time series node propagation space span values ​​and the sequence of the historical running speed local time series node energy statistical paradigm values, determining the node energy propagation attenuation coefficient values ​​of other running speed local time series associated implicit feature vectors in the sequence of the running speed local time series associated implicit feature vectors to obtain a sequence of running speed local time series node energy propagation attenuation coefficient values, including: Taking each running speed local time series node propagation space span value in the sequence of running speed local time series node propagation space span values ​​as an exponential power, calculating an exponential function value with a natural constant e as a base to obtain a sequence of running speed local time series node propagation class support space span values; Calculating the positional sum of the sequence of the running speed local time series node propagation class supported space span values ​​and the sequence of the running speed local time series node propagation space span values ​​to obtain a sequence of the running speed local time series node propagation space span modulation coefficients; The positional division between the sequence of the historical running speed local time series node energy statistical paradigm values ​​and the sequence of the running speed local time series node propagation space span modulation coefficients is calculated to obtain the sequence of the running speed local time series node energy propagation attenuation coefficient values.

6. The control method of the intelligent control circuit board for a treadmill according to claim 5, characterized in that: The running speed timing propagation aggregate representation vector and the heart rate timing propagation aggregate representation vector are subjected to feature interaction response processing to obtain a significant running speed-heart rate timing interaction fusion representation vector, including: inputting the running speed timing propagation aggregate representation vector and the heart rate timing propagation aggregate representation vector into a feature interaction response module based on an adaptive distinguishable mechanism to obtain the significant running speed-heart rate timing interaction fusion representation vector.

7. The control method of the intelligent control circuit board for a treadmill according to claim 6, characterized in that: Inputting the running speed time series propagation aggregate representation vector and the heart rate time series propagation aggregate representation vector into a feature interaction response module based on an adaptive distinguishable mechanism to obtain the significant running speed-heart rate time series interaction fusion representation vector, including: Calculating the position-by-position response between the heart rate time series propagation aggregate representation vector and the running speed time series propagation aggregate representation vector to obtain a running speed-heart rate time series position-by-position response feature vector; Using a Softmax function to normalize the running speed-heart rate time series position-by-position response feature vector to obtain a normalized running speed-heart rate time series position-by-position response feature vector; Taking the negative number of each position characteristic value in the normalized running speed-heart rate time series position-by-position response characteristic vector as a power, calculating the value of a natural exponential function with a natural constant e as a base to obtain a normalized running speed-heart rate time series position-by-position response class support characteristic vector; Calculating the inverse of the sum of the eigenvalues ​​of each position in the normalized running speed-heart rate timing position-by-position response class support eigenvector and a constant one to obtain a running speed-heart rate timing response screening weight mask vector; Calculate the position point multiplication between the running speed-heart rate timing response screening weight mask vector and the normalized running speed-heart rate timing position-by-position response feature vector to obtain a running speed-heart rate timing position-by-position response distinguishable weight mask vector; The running speed-heart rate timing position-by-position response distinguishable weight mask vector is calculated and multiplied by the position point between the running speed-heart rate timing position-by-position response feature vector to obtain a significant running speed-heart rate timing interaction fusion representation vector.

8. The control method of the intelligent control circuit board for a treadmill according to claim 7, characterized in that: Based on the significant running speed-heart rate timing interaction fusion representation vector, a control instruction is generated, and the control instruction is used to indicate increasing the treadmill slope value, decreasing the treadmill slope value, and maintaining the treadmill slope value, including: inputting the significant running speed-heart rate timing interaction fusion representation vector into a classifier-based slope controller to obtain a control instruction, and the control instruction is used to indicate increasing the treadmill slope value, decreasing the treadmill slope value, and maintaining the treadmill slope value.

9. The control method of the intelligent control circuit board for a treadmill according to claim 8, characterized in that: It also includes a training step: for training the sequence encoder based on the 1D-CNN model, the node feature propagation network based on the node energy decay mechanism, the feature interaction response module based on the adaptive distinguishable mechanism, and the slope controller based on the classifier; Wherein, the training steps are: Acquire training data, wherein the training data includes a time queue of training running speed and a time queue of training heart rate values ​​collected by the sensor group; Input the time queue of the training running speed and the time queue of the training heart rate value into a sequence encoder based on the 1D-CNN model to obtain a sequence of local temporal association implicit feature vectors of the training running speed and a sequence of local temporal association implicit feature vectors of the training heart rate; Inputting the sequence of the training running speed local time series associated implicit feature vectors and the sequence of the training heart rate local time series associated implicit feature vectors into a node feature propagation network based on a node energy decay mechanism to obtain a training running speed time series propagation aggregate representation vector and a training heart rate time series propagation aggregate representation vector; Inputting the training running speed time series propagation aggregation representation vector and the training heart rate time series propagation aggregation representation vector into a feature interaction response module based on an adaptive distinguishable mechanism to obtain a training significant running speed-heart rate time series interaction fusion representation vector; Passing the training significant running speed-heart rate time series interaction fusion representation vector through a classifier-based slope controller to obtain a classification loss function value; Based on the significant running speed-heart rate time series interaction fusion representation loss term, calculate the significant running speed-heart rate time series interaction fusion loss function value; Calculating a weighted sum of the classification loss function value and the significant running speed-heart rate time series interaction fusion loss function value to obtain a final loss function value; Based on the final loss function value, the sequence encoder based on the 1D-CNN model, the node feature propagation network based on the node energy decay mechanism, the feature interaction response module based on the adaptive distinguishable mechanism and the classifier-based slope controller are trained.

10. An intelligent control circuit board for a treadmill, which is controlled by using the control method for an intelligent control circuit board for a treadmill as claimed in any one of claims 1 to 9.

Images