Foot tub parameter intelligent adjusting method and system based on machine learning

Abstract

The invention discloses a foot tub parameter intelligent adjustment method and system based on machine learning. The method comprises the steps of data collection, data screening, foot tub parameter prediction model establishment, global parameter optimization and foot tub parameter adjustment. The invention relates to the technical field of smart homes, in particular to a foot tub parameter intelligent adjusting method and system based on machine learning, according to the scheme, data screening is carried out by dividing data neighborhoods and defining data complexity and local density, and the accuracy and reliability of abnormal value recognition are improved; a model is established by designing an excitation function, a front double transformation layer, a residual adjustment layer and a loss function, so that the expression ability, prediction accuracy and prediction error control ability of the model are improved; parameters are optimized by designing a parameter search function and a search rule, and the search direction and step length are dynamically adjusted, so that the search process is more efficient, the anti-interference capability of the search process is enhanced, and local optimum can be jumped out.

Description

Technical Field

[0001] The present invention relates to the field of smart home technology, and in particular to a method and system for intelligently adjusting parameters of a footbath basin based on machine learning. Background Art

[0002] The intelligent adjustment method and system of footbath parameters based on machine learning use machine learning technology and big data analysis technology to intelligently optimize and automatically adjust the operating parameters of the footbath, which can improve the intelligence level of the footbath and enhance the comfort of users. However, the traditional data screening method has the problem of being unable to effectively identify outliers and noise data in complex data sets; the traditional footbath parameter prediction model has the problems of simple model structure, single incentive mechanism, and lack of flexibility; the traditional global parameter optimization method has the problems of low search efficiency, easy to fall into local optimality, and high complexity of parameter adjustment. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method and system for intelligently adjusting footbath parameters based on machine learning. In view of the problem that traditional data screening methods cannot effectively identify outliers and noise data in complex data sets, this scheme performs data screening by dividing data neighborhoods, defining data complexity, defining local density, calculating anomalies and abnormal data detection, taking into account the spatial distance between data points and the density distribution of the neighborhood around the data points, thereby improving the accuracy and reliability of outlier identification. In view of the problems that traditional footbath parameter prediction models have simple model structures, single incentive mechanisms, and lack of flexibility, this scheme designs a model framework, an incentive function, a front double transformation layer, a hidden layer, a residual adjustment layer, an input The model is established by out-of-layer and loss function, which improves the model's adaptability and processing ability to different input situations, improves the model's expression ability, prediction accuracy, and ability to adjust and optimize features, and improves the model's output quality and ability to control prediction errors; in view of the problems of low search efficiency, easy to fall into local optimality and high parameter adjustment complexity of traditional global parameter optimization methods, this scheme designs search evolution factors, defines mapping functions, designs anti-interference factors, designs parameter search functions and search rules, and dynamically adjusts the search direction and step size, making the search process more efficient, realizing the randomness and continuity of the search direction, improving the search efficiency, and enhancing the anti-interference ability of the search process, so that the search process can jump out of the local optimality.

[0004] The technical solution adopted by the present invention is as follows: a method and system for intelligently adjusting parameters of a footbath basin based on machine learning, the method comprising the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data screening;

[0007] Step S3: Establishing a footbath parameter prediction model;

[0008] Step S4: global parameter optimization;

[0009] Step S5: footbath parameters adjustment.

[0010] Further, in step S1, the data collection collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users; the physical fatigue level includes three levels: high, medium and low; the physical reaction data refers to the skin temperature, heart rate and breathing rate when using the footbath; the footbath parameters include water volume, water temperature, heating speed and massage intensity.

[0011] Furthermore, in step S2, the data screening specifically includes the following steps:

[0012] Step S21: Divide the data neighborhood, select the N data points closest to each data point, calculate the Euclidean distance between each data point and the selected N data points, select the maximum value, set it as the neighborhood width of the data point, and set the data points within the neighborhood width as the neighborhood data point set;

[0013] Step S22: define data complexity, as follows:

[0014] ;

[0015] Where i represents the index of the data point, represents the position of the i-th data point, represents the data complexity of the i-th data point, d represents the index of the data dimension, D represents the maximum data dimension, n represents the index of the data point in the neighborhood data point set, and N represents the total number of data points in the neighborhood data point set. represents the eigenvalue of the dth dimension of the ith data point, represents the eigenvalue of the dth dimension of the nth data point, Indicates taking the absolute value, represents the maximum Euclidean distance between data point i and the data points in the neighborhood data point set, Represents the minimum Euclidean distance between data point i and the data points in the neighborhood data point set;

[0016] Step S23: define the local density, which is expressed as follows:

[0017] ;

[0018] in, represents the local density function, represents the local density of the i-th data point, j represents the index of the data point, M represents the total number of data points, Indicates taking the maximum value, represents the neighborhood width of the ith data point, Indicates the Euclidean distance. represents the Euclidean distance between the i-th data point and the j-th data point;

[0019] Step S24: Calculate the abnormality, expressed as follows:

[0020] ;

[0021] in, represents the abnormality of the i-th data point, represents the density adjustment factor, represents the local density average of all data points in the neighborhood data point set of the jth data point, represents the local density average of all data points in the neighborhood data point set of the i-th data point, represents the abnormality balance factor;

[0022] Step S25: abnormal data detection, setting an abnormality threshold, calculating the abnormality of all data points, and removing data with an abnormality greater than the abnormality threshold.

[0023] Furthermore, in step S3, the establishment of the footbath parameter prediction model specifically includes the following steps:

[0024] Step S31: designing a model framework, the footbath parameter prediction model includes a front double transformation layer, a hidden layer, a residual adjustment layer, a random inactivation layer and an output layer;

[0025] Step S32: Design an activation function, which is expressed as follows:

[0026] ;

[0027] in, represents the input of the domain activation function, represents the domain activation function, e represents the natural constant, represents the logarithmic function with a natural constant as base, Indicates square root;

[0028] Step S33: Design the front double conversion layer, which is expressed as follows:

[0029] ;

[0030] in, represents the output of the previous double transform layer, represents the leaky rectified linear unit function, and denote the weights and biases of the first layer of the preceding double transformation layer, and denote the weights and biases of the second layer of the preceding double transformation layer, represents the input of the previous double transform layer;

[0031] Step S34: Design the hidden layer, and use the domain activation function to excite the first and last layers of the hidden layer, which is expressed as follows:

[0032] ;

[0033] in, represents the output of the hidden layer, p represents the number of hidden layers, , and denote the weights of the pth, p-1th, and 1st layers of the hidden layer, respectively. , and They represent the bias of the 1st, p-1th and pth hidden layers respectively, Represents the input of the first hidden layer;

[0034] Step S35: Design a residual adjustment layer, which is expressed as follows:

[0035] ;

[0036] in, represents the output of the residual adjustment layer, , and Represent the weight, input and bias of the residual adjustment layer respectively;

[0037] Step S36: Design the output layer, which is expressed as follows:

[0038] ;

[0039] in, represents the output of the output layer, represents the normalized exponential function, , and Represent the weight, input and bias of the output layer respectively;

[0040] Step S37: Design a loss function, which is expressed as follows:

[0041] ;

[0042] in, represents the loss value of the model, j represents the index of the data sample for training the model, K represents the total number of data samples, log(·) represents the logarithmic operation, represents the true value of the jth sample, represents the predicted value of the jth sample, represents the edge loss weight, Represents the sensitivity factor.

[0043] Furthermore, in step S4, the global parameter optimization specifically includes the following steps:

[0044] Step S41: Optimization preparation, determining optimization parameters, including density adjustment factor, abnormality balance factor, number of hidden layers and sensitivity factor, creating a search space for optimization parameters, setting the inverse of the model's loss value as the fitness of the parameter, randomly generating an initial parameter search point cluster in the parameter search space, and calculating the fitness of the initial parameter search point cluster;

[0045] Step S42: Design and search evolution factors, as shown below:

[0046] ;

[0047] Among them, t represents the number of parameter searches, g represents the parameter dimension index of the parameter search point, represents the search evolution factor of the gth parameter dimension during the tth parameter search, T represents the maximum number of parameter searches, represents the average value of the g-th parameter dimension of all parameter search points, represents the maximum value of the g-th parameter dimension among all parameter search points, Represents the minimum value of the g-th parameter dimension among all parameter search points;

[0048] Step S43: define a mapping function, which is expressed as follows:

[0049] ;

[0050] in, represents the mapping function value of the gth parameter dimension during the tth parameter search, r1 represents a random number ranging from 3.3 to 3.9, Represents the mapping function value of the gth parameter dimension during the t-1th parameter search; set ;

[0051] Step S44: Design an anti-interference factor, which is expressed as follows:

[0052] ;

[0053] in, represents the anti-interference factor of the g-th parameter dimension during the t-th parameter search, Represents the value of the g-th dimension of the parameter search point with the highest parameter fitness in the initial parameter search point cluster, Represents the value of the g-th dimension of the parameter search point with the lowest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the highest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the lowest parameter fitness in the initial parameter search point cluster;

[0054] Step S45: Design parameter search function, expressed as follows:

[0055] ;

[0056] in, Indicates the position of the g-th parameter dimension of the parameter point obtained by the t-th parameter search. shows the position of the gth parameter dimension of the parameter point obtained by the t-1th parameter search, Indicates the position of the gth parameter dimension of the parameter point with the highest parameter fitness in the entire search process, Indicates the position of the gth parameter dimension of the parameter point with the lowest parameter fitness in the entire search process, represents the average value of the position of the gth parameter dimension of the initial parameter search point cluster, r2 represents a random number ranging from 0 to 1, Represents a random number between -0.5 and 0.5. and Respectively represent the upper and lower boundaries of the search space of the g-th parameter dimension;

[0057] Step S46: Design search rules, set the maximum number of parameter searches and the qualified value of parameter fitness, use the parameter search function to perform parameter search, and calculate the parameter performance value of the parameter point obtained by the search; if the fitness of the searched parameter point is greater than the qualified value of fitness, stop the search, and set the parameter of the parameter point with the highest parameter fitness at this time as the optimal parameter; if the maximum number of parameter searches is reached, search again; otherwise, continue the search.

[0058] Furthermore, in step S5, the footbath parameters are adjusted by collecting the age, physical fatigue level and physical reaction data of the footbath user in real time, inputting the data into a footbath parameter prediction model, the model predicts the parameters of the footbath, and adjusting the footbath parameters in real time according to the predicted parameters.

[0059] The intelligent adjustment system for footbath parameters based on machine learning provided by the present invention comprises a data acquisition module, a data screening module, a module for establishing a footbath parameter prediction model, a global parameter optimization module and a footbath parameter adjustment module;

[0060] The data collection module collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users, and sends the data to the data screening module;

[0061] The data screening module receives the data sent by the data acquisition module, performs data screening, and sends the screened data to the foot bath parameter prediction model establishment module and the global parameter optimization module;

[0062] The footbath parameter prediction model establishment module receives the data sent by the data screening module, establishes a footbath parameter prediction model, and sends the model data to the global parameter optimization module and the footbath parameter adjustment module;

[0063] The global parameter optimization module receives the data sent by the data screening module and the footbath parameter prediction model establishment module, performs global parameter optimization, and sends the data to the footbath parameter adjustment module;

[0064] The footbath parameter adjustment module receives data sent by the footbath parameter prediction model establishment module and the global parameter optimization module, and adjusts the parameters of the footbath in real time.

[0065] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0066] (1) To address the problem that traditional data screening methods cannot effectively identify outliers and noise data in complex data sets, this scheme performs data screening by dividing data neighborhoods, defining data complexity, defining local density, calculating anomaly and detecting abnormal data. It also considers the spatial distance between data points and the density distribution of the neighborhood around the data points, thereby improving the accuracy and reliability of outlier identification.

[0067] (2) Aiming at the problems of simple model structure, single incentive mechanism and lack of flexibility in the traditional footbath parameter prediction model, this scheme establishes the model by designing the model framework, incentive function, front double transformation layer, hidden layer, residual adjustment layer, output layer and loss function, which improves the model's adaptability and processing ability to different input conditions, improves the model's expression ability, prediction accuracy and ability to adjust and optimize features, and improves the model's output quality and ability to control prediction errors.

[0068] (3) In view of the problems of low search efficiency, easy to fall into local optimum and high complexity of parameter adjustment in traditional global parameter optimization methods, this scheme dynamically adjusts the search direction and step size by designing search evolution factors, defining mapping functions, designing anti-interference factors, designing parameter search functions and search rules, making the search process more efficient, achieving randomness and continuity of the search direction, improving search efficiency, and enhancing the anti-interference ability of the search process, so that the search process can jump out of the local optimum. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A schematic diagram of the process of the intelligent adjustment method of footbath parameters based on machine learning provided by the present invention;

[0070] Figure 2 A schematic diagram of the intelligent adjustment system for footbath parameters based on machine learning provided by the present invention;

[0071] Figure 3 is a schematic flow chart of step S2;

[0072] Figure 4 is a schematic flow chart of step S3;

[0073] Figure 5 It is a schematic diagram of the process of step S4.

[0074] 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

[0075] 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.

[0076] In the description of the present invention, it should be understood 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 should not be understood as limiting the present invention.

[0077] Example 1, see Figure 1 The present invention provides a method for intelligently adjusting parameters of a footbath basin based on machine learning, the method comprising the following steps:

[0078] Step S1: data collection, collecting the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users;

[0079] Step S2: Data screening, data screening is performed by dividing data neighborhoods, defining data complexity, defining local density, calculating anomaly and detecting abnormal data;

[0080] Step S3: Establishing a footbath parameter prediction model by designing a model framework, an excitation function, a front double transformation layer, a hidden layer, a residual adjustment layer, an output layer and a loss function;

[0081] Step S4: global parameter optimization, optimizing global parameters by designing search evolution factors, defining mapping functions, designing anti-interference factors, designing parameter search functions and search rules;

[0082] Step S5: footbath parameter adjustment, real-time collection of footbath user's age, physical fatigue level and physical reaction data, input of the data into the footbath parameter prediction model, the model predicts the footbath parameters, and the footbath parameters are adjusted in real time according to the predicted parameters.

[0083] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the data collection collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users; the physical fatigue level includes three levels: high, medium and low; the physical reaction data refers to the skin temperature, heart rate and breathing rate when using the footbath; the footbath parameters include water volume, water temperature, heating speed and massage intensity.

[0084] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, the data screening specifically includes the following steps:

[0085] Step S21: Divide the data neighborhood, select the N data points closest to each data point, calculate the Euclidean distance between each data point and the selected N data points, select the maximum value, set it as the neighborhood width of the data point, and set the data points within the neighborhood width as the neighborhood data point set;

[0086] Step S22: define data complexity, as follows:

[0087] ;

[0088] Where i represents the index of the data point, represents the position of the i-th data point, represents the data complexity of the i-th data point, d represents the index of the data dimension, D represents the maximum data dimension, n represents the index of the data point in the neighborhood data point set, and N represents the total number of data points in the neighborhood data point set. represents the eigenvalue of the dth dimension of the ith data point, represents the eigenvalue of the dth dimension of the nth data point, Indicates taking the absolute value, represents the maximum Euclidean distance between data point i and the data points in the neighborhood data point set, Represents the minimum Euclidean distance between data point i and the data points in the neighborhood data point set;

[0089] Step S23: define the local density, which is expressed as follows:

[0090] ;

[0091] in, represents the local density function, represents the local density of the i-th data point, j represents the index of the data point, M represents the total number of data points, Indicates taking the maximum value, represents the neighborhood width of the ith data point, Indicates the Euclidean distance. represents the Euclidean distance between the i-th data point and the j-th data point;

[0092] Step S24: Calculate the abnormality, expressed as follows:

[0093] ;

[0094] in, represents the abnormality of the i-th data point, represents the density adjustment factor, represents the local density average of all data points in the neighborhood data point set of the jth data point, represents the local density average of all data points in the neighborhood data point set of the i-th data point, represents the abnormality balance factor;

[0095] Step S25: abnormal data detection, setting an abnormality threshold, calculating the abnormality of all data points, and removing data with an abnormality greater than the abnormality threshold.

[0096] By performing the above operations, in order to address the problem that traditional data screening methods cannot effectively identify outliers and noise data in complex data sets, this solution performs data screening by dividing data neighborhoods, defining data complexity, defining local density, calculating anomalies and detecting abnormal data. It takes into account the spatial distance between data points and the density distribution of the neighborhood around the data points, thereby improving the accuracy and reliability of outlier identification.

[0097] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S3, the foot bath parameter prediction model is established, which specifically includes the following steps:

[0098] Step S31: designing a model framework, the footbath parameter prediction model includes a front double transformation layer, a hidden layer, a residual adjustment layer, a random inactivation layer and an output layer;

[0099] Step S32: Design an activation function, which is expressed as follows:

[0100] ;

[0101] in, represents the input of the domain activation function, represents the domain activation function, e represents the natural constant, represents a logarithmic function with a natural constant as base, Indicates square root;

[0102] Step S33: Design the front double conversion layer, which is expressed as follows:

[0103] ;

[0104] in, represents the output of the previous double transform layer, represents the leaky rectified linear unit function, and denote the weights and biases of the first layer of the preceding double transformation layer, and denote the weights and biases of the second layer of the preceding double transformation layer, represents the input of the previous double transform layer;

[0105] Step S34: Design the hidden layer, and use the domain activation function to excite the first and last layers of the hidden layer, which is expressed as follows:

[0106] ;

[0107] in, represents the output of the hidden layer, p represents the number of hidden layers, , and denote the weights of the pth, p-1th, and 1st layers of the hidden layer, respectively. , and They represent the bias of the 1st, p-1th and pth hidden layers respectively, Represents the input of the first hidden layer;

[0108] Step S35: Design a residual adjustment layer, which is expressed as follows:

[0109] ;

[0110] in, represents the output of the residual adjustment layer, , and Represent the weight, input and bias of the residual adjustment layer respectively;

[0111] Step S36: Design the output layer, which is expressed as follows:

[0112] ;

[0113] in, represents the output of the output layer, represents the normalized exponential function, , and Represent the weight, input and bias of the output layer respectively;

[0114] Step S37: Design a loss function, which is expressed as follows:

[0115] ;

[0116] in, represents the loss value of the model, j represents the index of the data sample for training the model, K represents the total number of data samples, log(·) represents the logarithmic operation, represents the true value of the jth sample, represents the predicted value of the jth sample, represents the edge loss weight, Represents the sensitivity factor.

[0117] By performing the above operations, in view of the problems of simple model structure, single incentive mechanism and lack of flexibility in the traditional footbath parameter prediction model, this scheme establishes the model by designing the model framework, incentive function, front double transformation layer, hidden layer, residual adjustment layer, output layer and loss function, which improves the model's adaptability and processing ability to different input conditions, improves the model's expression ability, prediction accuracy and ability to adjust and optimize features, and improves the model's output quality and ability to control prediction errors.

[0118] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the global parameter optimization specifically includes the following steps:

[0119] Step S41: Optimization preparation, determining optimization parameters, including density adjustment factor, abnormality balance factor, number of hidden layers and sensitivity factor, creating a search space for optimization parameters, setting the inverse of the model's loss value as the fitness of the parameter, randomly generating an initial parameter search point cluster in the parameter search space, and calculating the fitness of the initial parameter search point cluster;

[0120] Step S42: Design and search evolution factors, as shown below:

[0121] ;

[0122] Among them, t represents the number of parameter searches, g represents the parameter dimension index of the parameter search point, represents the search evolution factor of the gth parameter dimension during the tth parameter search, T represents the maximum number of parameter searches, represents the average value of the g-th parameter dimension of all parameter search points, represents the maximum value of the g-th parameter dimension among all parameter search points, Represents the minimum value of the g-th parameter dimension among all parameter search points;

[0123] Step S43: define a mapping function, which is expressed as follows:

[0124] ;

[0125] in, represents the mapping function value of the gth parameter dimension during the tth parameter search, r1 represents a random number ranging from 3.3 to 3.9, Represents the mapping function value of the gth parameter dimension during the t-1th parameter search; set ;

[0126] Step S44: Design an anti-interference factor, which is expressed as follows:

[0127] ;

[0128] in, represents the anti-interference factor of the g-th parameter dimension during the t-th parameter search, Represents the value of the g-th dimension of the parameter search point with the highest parameter fitness in the initial parameter search point cluster, Represents the value of the g-th dimension of the parameter search point with the lowest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the highest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the lowest parameter fitness in the initial parameter search point cluster;

[0129] Step S45: Design parameter search function, expressed as follows:

[0130] ;

[0131] in, Indicates the position of the g-th parameter dimension of the parameter point obtained by the t-th parameter search. shows the position of the gth parameter dimension of the parameter point obtained by the t-1th parameter search, Indicates the position of the gth parameter dimension of the parameter point with the highest parameter fitness in the entire search process, Indicates the position of the gth parameter dimension of the parameter point with the lowest parameter fitness in the entire search process, represents the average value of the position of the gth parameter dimension of the initial parameter search point cluster, r2 represents a random number ranging from 0 to 1, Represents a random number between -0.5 and 0.5. and Respectively represent the upper and lower boundaries of the search space of the g-th parameter dimension;

[0132] Step S46: Design search rules, set the maximum number of parameter searches and the qualified value of parameter fitness, use the parameter search function to perform parameter search, and calculate the parameter performance value of the parameter point obtained by the search; if the fitness of the searched parameter point is greater than the qualified value of fitness, stop the search, and set the parameter of the parameter point with the highest parameter fitness at this time as the optimal parameter; if the maximum number of parameter searches is reached, search again; otherwise, continue the search.

[0133] By performing the above operations, in order to address the problems of low search efficiency, easy to fall into local optimality and high parameter adjustment complexity in traditional global parameter optimization methods, this scheme designs search evolution factors, defines mapping functions, designs anti-interference factors, designs parameter search functions and search rules, and dynamically adjusts the search direction and step size to make the search process more efficient, achieve the randomness and continuity of the search direction, improve the search efficiency, enhance the anti-interference ability of the search process, and enable the search process to jump out of the local optimality.

[0134] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the footbath parameter adjustment is performed by collecting the age, physical fatigue level and physical reaction data of the footbath user in real time, inputting the data into the footbath parameter prediction model, and the model predicts the parameters of the footbath, and adjusts the parameters of the footbath in real time according to the predicted parameters.

[0135] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, and the intelligent adjustment system of footbath parameters based on machine learning provided by the present invention includes a data acquisition module, a data screening module, a module for establishing a footbath parameter prediction model, a global parameter optimization module and a footbath parameter adjustment module;

[0136] The data collection module collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users, and sends the data to the data screening module;

[0137] The data screening module receives the data sent by the data acquisition module, performs data screening, and sends the screened data to the foot bath parameter prediction model establishment module and the global parameter optimization module;

[0138] The footbath parameter prediction model establishment module receives the data sent by the data screening module, establishes a footbath parameter prediction model, and sends the model data to the global parameter optimization module and the footbath parameter adjustment module;

[0139] The global parameter optimization module receives the data sent by the data screening module and the footbath parameter prediction model establishment module, performs global parameter optimization, and sends the data to the footbath parameter adjustment module;

[0140] The footbath parameter adjustment module receives data sent by the footbath parameter prediction model establishment module and the global parameter optimization module, and adjusts the parameters of the footbath in real time.

[0141] 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.

[0142] 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.

[0143] 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. A method for intelligently adjusting footbath parameters based on machine learning, characterized in that: The method comprises the following steps: Step S1: data collection, collecting the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users; Step S2: Data screening, data screening is performed by dividing data neighborhoods, defining data complexity, defining local density, calculating anomaly and detecting abnormal data; Step S3: Establishing a footbath parameter prediction model by designing a model framework, an excitation function, a front double transformation layer, a hidden layer, a residual adjustment layer, an output layer and a loss function; Step S4: global parameter optimization, optimizing global parameters by designing search evolution factors, defining mapping functions, designing anti-interference factors, designing parameter search functions and search rules; Step S5: footbath parameter adjustment, real-time collection of footbath user's age, physical fatigue level and physical reaction data, input of the data into the footbath parameter prediction model, the model predicts the footbath parameters, and the footbath parameters are adjusted in real time according to the predicted parameters.

2. The method for intelligently adjusting footbath parameters based on machine learning according to claim 1 is characterized in that: In step S2, the data screening specifically includes the following steps: Step S21: Divide the data neighborhood, select the N data points closest to each data point, calculate the Euclidean distance between each data point and the selected N data points, select the maximum value, set it as the neighborhood width of the data point, and set the data points within the neighborhood width as the neighborhood data point set; Step S22: define data complexity, as follows: ； Where i represents the index of the data point, represents the position of the i-th data point, represents the data complexity of the i-th data point, d represents the index of the data dimension, D represents the maximum data dimension, n represents the index of the data point in the neighborhood data point set, and N represents the total number of data points in the neighborhood data point set. represents the eigenvalue of the dth dimension of the ith data point, represents the eigenvalue of the dth dimension of the nth data point, Indicates taking the absolute value, represents the maximum Euclidean distance between data point i and the data points in the neighborhood data point set, Represents the minimum Euclidean distance between data point i and the data points in the neighborhood data point set; Step S23: define the local density, which is expressed as follows: ； in, represents the local density function, represents the local density of the i-th data point, j represents the index of the data point, M represents the total number of data points, Indicates taking the maximum value, represents the neighborhood width of the ith data point, Indicates the Euclidean distance. represents the Euclidean distance between the i-th data point and the j-th data point; Step S24: Calculate the abnormality, expressed as follows: ； in, represents the abnormality of the i-th data point, represents the density adjustment factor, represents the local density average of all data points in the neighborhood data point set of the jth data point, represents the local density average of all data points in the neighborhood data point set of the i-th data point, represents the abnormality balance factor; Step S25: abnormal data detection, setting an abnormality threshold, calculating the abnormality of all data points, and removing data with an abnormality greater than the abnormality threshold.

3. The method for intelligently adjusting footbath parameters based on machine learning according to claim 1 is characterized in that: In step S3, the establishment of a footbath parameter prediction model specifically includes the following steps: Step S31: designing a model framework, the footbath parameter prediction model includes a front double transformation layer, a hidden layer, a residual adjustment layer, a random inactivation layer and an output layer; Step S32: Design an activation function, which is expressed as follows: ； in, represents the input of the domain activation function, represents the domain activation function, e represents the natural constant, represents a logarithmic function with a natural constant as base, Indicates square root; Step S33: Design the front double conversion layer, which is expressed as follows: ； in, represents the output of the previous double transform layer, represents the leaky rectified linear unit function, and denote the weights and biases of the first layer of the preceding double transformation layer, and denote the weights and biases of the second layer of the preceding double transformation layer, represents the input of the previous double transform layer; Step S34: Design the hidden layer, and use the domain activation function to excite the first and last layers of the hidden layer, which is expressed as follows: ； in, represents the output of the hidden layer, p represents the number of hidden layers, , and denote the weights of the pth, p-1th, and 1st layers of the hidden layer, respectively. , and They represent the bias of the 1st, p-1th and pth hidden layers respectively, Represents the input of the first hidden layer; Step S35: Design a residual adjustment layer, which is expressed as follows: ； in, represents the output of the residual adjustment layer, , and Represent the weight, input and bias of the residual adjustment layer respectively; Step S36: Design the output layer, which is expressed as follows: ； in, represents the output of the output layer, represents the normalized exponential function, , and Represent the weight, input and bias of the output layer respectively; Step S37: Design a loss function, which is expressed as follows: ； in, represents the loss value of the model, j represents the index of the data sample for training the model, K represents the total number of data samples, log(·) represents the logarithmic operation, represents the true value of the jth sample, represents the predicted value of the jth sample, represents the edge loss weight, Represents the sensitivity factor.

4. The method for intelligently adjusting footbath parameters based on machine learning according to claim 1, characterized in that: In step S4, the global parameter optimization specifically includes the following steps: Step S41: Optimization preparation, determining optimization parameters, including density adjustment factor, abnormality balance factor, number of hidden layers and sensitivity factor, creating a search space for optimization parameters, setting the inverse of the model's loss value as the fitness of the parameter, randomly generating an initial parameter search point cluster in the parameter search space, and calculating the fitness of the initial parameter search point cluster; Step S42: Design and search evolution factors, as shown below: ； Among them, t represents the number of parameter searches, g represents the parameter dimension index of the parameter search point, represents the search evolution factor of the gth parameter dimension during the tth parameter search, T represents the maximum number of parameter searches, represents the average value of the g-th parameter dimension of all parameter search points, represents the maximum value of the g-th parameter dimension among all parameter search points, Represents the minimum value of the g-th parameter dimension among all parameter search points; Step S43: define a mapping function, which is expressed as follows: ； in, represents the mapping function value of the gth parameter dimension during the tth parameter search, r1 represents a random number ranging from 3.3 to 3.9, Represents the mapping function value of the gth parameter dimension during the t-1th parameter search; set ; Step S44: Design an anti-interference factor, which is expressed as follows: ； in, represents the anti-interference factor of the g-th parameter dimension during the t-th parameter search, Represents the value of the g-th dimension of the parameter search point with the highest parameter fitness in the initial parameter search point cluster, Represents the value of the g-th dimension of the parameter search point with the lowest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the highest parameter fitness in the initial parameter search point cluster, The feature vector representing the parameter search point with the lowest parameter fitness in the initial parameter search point cluster; Step S45: Design parameter search function, expressed as follows: ； in, Indicates the position of the g-th parameter dimension of the parameter point obtained by the t-th parameter search. shows the position of the gth parameter dimension of the parameter point obtained by the t-1th parameter search, Indicates the position of the gth parameter dimension of the parameter point with the highest parameter fitness in the entire search process, Indicates the position of the gth parameter dimension of the parameter point with the lowest parameter fitness in the entire search process, represents the average value of the position of the gth parameter dimension of the initial parameter search point cluster, r2 represents a random number ranging from 0 to 1, Represents a random number between -0.5 and 0.

5. and Respectively represent the upper and lower boundaries of the search space of the g-th parameter dimension; Step S46: Design search rules, set the maximum number of parameter searches and the qualified value of parameter fitness, use the parameter search function to perform parameter search, and calculate the parameter performance value of the parameter point obtained by the search; if the fitness of the searched parameter point is greater than the qualified value of fitness, stop the search, and set the parameter of the parameter point with the highest parameter fitness at this time as the optimal parameter; if the maximum number of parameter searches is reached, search again; otherwise, continue the search.

5. The method for intelligently adjusting footbath parameters based on machine learning according to claim 1, characterized in that: In step S1, the data collection collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users; the physical fatigue level includes three levels: high, medium and low; the physical reaction data refers to the skin temperature, heart rate and breathing rate when using the footbath; the footbath parameters include water volume, water temperature, heating speed and massage intensity; In step S5, the footbath parameters are adjusted by real-time collection of the footbath user's age, physical fatigue level and physical reaction data, inputting the data into a footbath parameter prediction model, the model predicts the parameters of the footbath, and adjusting the footbath parameters in real time according to the predicted parameters.

6. A footbath parameter intelligent adjustment system based on machine learning, used to implement the footbath parameter intelligent adjustment method based on machine learning as described in any one of claims 1 to 5, characterized in that: It includes a data acquisition module, a data screening module, a footbath parameter prediction model establishment module, a global parameter optimization module and a footbath parameter adjustment module; The data collection module collects the age, physical fatigue level, physical reaction data and footbath parameters of historical footbath users, and sends the data to the data screening module; The data screening module receives the data sent by the data acquisition module, performs data screening, and sends the screened data to the foot bath parameter prediction model establishment module and the global parameter optimization module; The footbath parameter prediction model establishment module receives the data sent by the data screening module, establishes a footbath parameter prediction model, and sends the model data to the global parameter optimization module and the footbath parameter adjustment module; The global parameter optimization module receives the data sent by the data screening module and the footbath parameter prediction model establishment module, performs global parameter optimization, and sends the data to the footbath parameter adjustment module; The footbath parameter adjustment module receives data sent by the footbath parameter prediction model establishment module and the global parameter optimization module, and adjusts the parameters of the footbath in real time.

Images