Optimization method and system based on massage sofa control system

Abstract

The embodiment of the invention provides an optimization method and system based on a massage sofa control system, and high individuation and optimization of massage control are realized by combining a basic control optimization network and a customized control optimization network. The sample massage control data and the comfort feedback data of the sample massage control data are used for supervised training, the parameter optimization effect of a machine learning network is effectively improved, and then an accurate comfort evaluation network is generated. The comfort assessment network can accurately assess the massage instruction generated by the customized control optimization network, performs reinforcement learning on the customized control optimization network based on the assessment parameters, and finally outputs the optimized target control optimization network, so that the comfort and user experience of the massage sofa are remarkably improved, and the user experience is improved. And intelligent and personalized customization of massage control is realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an optimization method and system based on a massage sofa control system. Background Art

[0002] With the accelerated pace of life and the increasing demand for a healthy and comfortable life, massage sofas, as a type of smart furniture that can provide a comfortable massage experience, have gradually become widely popular in the market. Through the built-in control system, massage sofas can achieve a variety of massage functions and modes according to user needs, thereby helping users relieve fatigue and improve blood circulation.

[0003] However, existing massage sofa control systems still have some shortcomings in providing personalized massage experience. Traditional massage sofa control systems often use fixed massage programs and parameters, which cannot be flexibly adjusted according to individual differences and comfort preferences of users. As a result, some users may feel uncomfortable or have poor results when using massage sofas because the massage intensity, rhythm or mode do not meet their personal preferences.

[0004] In order to overcome this problem, related technologies have begun to try to apply advanced technologies such as machine learning to massage sofa control systems to personalize and optimize massage programs. By learning and optimizing massage control data through machine learning networks, massage programs that are more in line with user comfort preferences can be generated. However, existing methods often only rely on basic massage control networks and lack the ability to customize and optimize for individual user differences. Summary of the invention

[0005] In view of this, an object of the embodiments of the present invention is to provide an optimization method and system based on a massage sofa control system.

[0006] According to one aspect of an embodiment of the present invention, there is provided an optimization method based on a massage sofa control system, the method comprising: Acquire a basic control optimization network and a customized control optimization network of a massage sofa control system, wherein the customized control optimization network is obtained by customizing the basic control optimization network by setting a massage instruction set; Generate sample massage control data using the basic control optimization network and / or the customized control optimization network, and add training supervision data to the sample massage control data based on comfort feedback data of the sample massage control data, so as to optimize parameters of a set machine learning network by using the sample massage control data after adding the training supervision data as massage program learning data, and generate a comfort evaluation network; Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, and the customized control optimization network after the enhanced parameter learning is output as a target control optimization network.

[0007] In a possible implementation manner of the first aspect, obtaining a basic control optimization network and a customized control optimization network includes: Obtaining an initial control optimization network; Performing initialization parameter learning on the initial control optimization network based on a preset sample massage control optimization data sequence to generate a basic control optimization network, wherein the sample massage control optimization data sequence includes unsupervised sample massage control optimization data; The basic control optimization network is customized based on the set massage instruction set to generate a customized control optimization network.

[0008] In a possible implementation manner of the first aspect, the basic control optimization network is customized based on a set massage instruction set to generate a customized control optimization network, including: Setting a plurality of massage mode labels, wherein the massage mode labels include massage technique mode labels and / or massage area mode labels; Clustering the set massage instruction set based on each of the massage mode labels to generate massage instruction clusters corresponding to each of the massage mode labels; The control optimization network is customized for the basic control optimization network based on each of the massage instruction clusters, and a customized control optimization network corresponding to each of the massage mode labels is generated.

[0009] In a possible implementation manner of the first aspect, generating sample massage control data using the basic control optimization network and / or the customized control optimization network includes: Get sample simulated massage instruction data; The sample simulated massage instruction data are loaded into the basic control optimization network and the customized control optimization network respectively, and the optimization data of the basic control optimization network and the optimization data of the customized control optimization network are used as sample massage control data respectively.

[0010] In a possible implementation of the first aspect, the step of using the sample massage control data after adding the training supervision data as the massage program learning data to optimize the parameters of the set machine learning network to generate the comfort evaluation network includes: The number of training samples is set, and the sample massage control data after adding the training supervision data is used as the massage program learning data; If the number of massage program learning data is not less than the number of training samples, the parameters of the set machine learning network are optimized based on the massage program learning data to generate a comfort assessment network, wherein the comfort assessment network is used to extract a massage instruction description vector from the massage instruction optimization data and generate comfort assessment parameters based on the massage instruction description vector.

[0011] In a possible implementation of the first aspect, performing comfort evaluation on the massage instruction optimization data generated by the customized control optimization network based on the comfort evaluation network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, comprises: Loading the same set massage instruction data into the basic control optimization network and the customized control optimization network respectively, generating massage instruction optimization data corresponding to the basic control optimization network and massage instruction optimization data corresponding to the customized control optimization network; Based on the comfort evaluation network, respectively, the massage instruction optimization data corresponding to the basic control optimization network and the massage instruction optimization data corresponding to the customized control optimization network are evaluated for comfort, and a comfort evaluation parameter corresponding to the basic control optimization network and a comfort evaluation parameter corresponding to the customized control optimization network are generated; Determine a comfort difference value between the basic control optimization network and the customized control optimization network based on the comfort evaluation parameter corresponding to the basic control optimization network and the comfort evaluation parameter generated by the customized control optimization network; Parameters of the customized control optimization network are optimized according to the comfort difference value.

[0012] In a possible implementation of the first aspect, performing comfort evaluation on the massage instruction optimization data generated by the customized control optimization network based on the comfort evaluation network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, comprises: Setting a threshold value and acquiring cyclic learning data, wherein the comfort evaluation parameter corresponding to the cyclic learning data meets the threshold value; Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network to generate a comfort evaluation parameter corresponding to the customized control optimization network; If the comfort evaluation parameter is less than a threshold value, cyclic parameter learning is performed on the customized control optimization network based on the cyclic learning data until the comfort evaluation parameter corresponding to the customized control optimization network meets the threshold value.

[0013] In a possible implementation of the first aspect, the method further includes: Based on the target control optimization network, any input target massage control data is optimized, and based on the optimized massage control optimization data, the massage sofa control system is controlled.

[0014] Another aspect of the embodiment of the invention provides an optimization system based on a massage sofa control system, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; the memory is used to store a computer program; the processor is used to implement any one of the above steps of the optimization method based on the massage sofa control system when executing the computer program.

[0015] According to another aspect of an embodiment of the present invention, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned optimization method based on the massage sofa control system can be executed.

[0016] In order to make the above-mentioned purposes, features and advantages of the embodiments of the present invention more obvious and easy to understand, the embodiments will be described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 A schematic diagram of components of an optimization system based on a massage sofa control system provided by an embodiment of the present invention is shown; Figure 2 A schematic flow chart of an optimization method based on a massage sofa control system provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The terms "first", "second", "third", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0021] Figure 1 An exemplary component diagram of an optimization system 100 based on a massage sofa control system is shown. The optimization system 100 based on a massage sofa control system may include one or more processors 104, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The optimization system 100 based on a massage sofa control system may also include any storage medium 106, which is used to store any kind of information such as code, settings, data, etc. Non-limitingly, for example, the storage medium 106 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any storage medium may use any technology to store information. Further, any storage medium may provide volatile or non-volatile retention of information. Further, any storage medium may represent a fixed or removable component of the optimization system 100 based on a massage sofa control system. In one case, when the processor 104 executes instructions with dependencies stored in any storage medium or combination of storage media, the optimization system 100 based on the massage sofa control system may perform any operation of the associated instructions. The optimization system 100 based on the massage sofa control system further includes one or more drive units 108 for interacting with any storage medium, such as a hard disk drive unit, an optical disk drive unit, and the like.

[0022] The optimization system 100 based on the massage sofa control system also includes an input / output 110 (I / O) for receiving various inputs (via an input unit 112) and for providing various outputs (via an output unit 114). A specific output mechanism may include a presentation device 116 and a graphical user interface (GUI) 118 with dependencies. The optimization system 100 based on the massage sofa control system may also include one or more network interfaces 120, which are used to exchange data with other devices via one or more communication units 122. One or more communication buses 124 couple the components described above together.

[0023] The communication unit 122 can be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication unit 122 can include any combination of hardwired links, wireless links, routers, gateway functions, the name based on the optimization system 100 of the massage sofa control system, etc., governed by any protocol or combination of protocols.

[0024] Figure 2 The schematic diagram of the process of the optimization method and system based on the massage sofa control system provided by the embodiment of the present invention is shown. The optimization method and system based on the massage sofa control system can be Figure 1 The optimization system 100 based on the massage sofa control system shown in the figure is executed, and the detailed steps of the optimization method based on the massage sofa control system are introduced as follows.

[0025] Step S110, obtaining a basic control optimization network and a customized control optimization network of a massage sofa control system, wherein the customized control optimization network is obtained by customizing the basic control optimization network by setting a massage instruction set.

[0026] Step S120, using the basic control optimization network and / or the customized control optimization network to generate sample massage control data, and adding training supervision data to the sample massage control data based on the comfort feedback data of the sample massage control data, so as to use the sample massage control data after adding the training supervision data as massage program learning data to optimize the parameters of the set machine learning network, and generate a comfort evaluation network.

[0027] Step S130, based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, and the customized control optimization network after the enhanced parameter learning is output as a target control optimization network.

[0028] In this embodiment, the server first obtains the initial control optimization network at the beginning of the entire optimization process of the massage sofa control system. This initial control optimization network can be regarded as a basic framework, which may contain some basic massage control logic and parameters. For example, the initial network may specify the initial range of basic parameters such as massage intensity and massage frequency, such as the initial range of massage intensity is set to 1-10 levels, and the initial range of massage frequency is set to 10-30 times per minute. This is like setting an initial template for subsequent optimization, which may be initially constructed based on some basic ergonomic principles and common massage experience.

[0029] The sample massage control optimization data sequence contains unsupervised sample massage control optimization data. This data may come from the collection of operation records of different users using massage sofas without specific guidance. For example, the server collects the operation data of 100 users when they freely select massage functions. One of the users may first set the massage intensity to level 5 and the frequency to 20 times / minute when using the massage sofa, and then gradually adjust the intensity to level 7 and the frequency to 25 times / minute during the massage process. Such operation data is part of the unsupervised sample massage control optimization data for the server.

[0030] The server inputs these sample data into the initial control optimization network. Taking massage intensity as an example, the initial control optimization network may initially believe that it is reasonable to evenly distribute the intensity between 1 and 10 levels. However, through learning from a large amount of sample data, it is found that most users prefer to choose intensity between 3 and 7 levels in the initial setting, and there is also a certain regularity in adjusting the amplitude of the intensity during the massage process. For example, few users jump directly from level 1 to level 10. Therefore, the server adjusts the parameters in the initial control optimization network based on these data, such as adjusting the probability distribution of the intensity parameters, so that there is a higher probability weight in the selection of intensity between 3 and 7 levels. After such a learning process, the basic control optimization network is generated. This basic control optimization network has optimized the initial massage control logic and parameters, which is more in line with the general usage habits of users.

[0031] The server sets multiple massage mode tags, including massage technique mode tags and / or massage area mode tags. For example, massage technique mode tags may include kneading, tapping, and massage, etc.; massage area mode tags may include neck, shoulder, waist, etc. These tags are for more detailed classification and customization of massage functions.

[0032] Assume that the massage instruction set contains 1,000 different massage instructions, and each instruction contains detailed information such as strength, frequency, and massage head movement trajectory. For the kneading technique in the massage technique mode tag, the server will cluster the instructions with these features according to the kneading-related features in the instructions, such as the massage head's movement mode is a circular trajectory, the strength is moderate, and the frequency is relatively stable, to form a massage instruction cluster corresponding to the kneading technique. Similarly, for the clustering of massage instructions for the neck area, the server will select those instructions that are mainly targeted at the neck area, such as those where the massage head's range of action is concentrated within the neck coordinate range, and cluster them.

[0033] Taking the massage instruction cluster corresponding to the kneading technique as an example, the server inputs the instruction information in this cluster into the basic control optimization network. The basic control optimization network was originally a general network. Now, for the kneading technique, it will adjust its own parameters according to the instructions in the cluster. For example, in the kneading technique, the server found that the instructions in the cluster generally require the massage intensity to be between 4 and 6 levels, the frequency to be between 15 and 20 times / minute, and the movement trajectory of the massage head has a specific radius range. Therefore, the basic control optimization network customizes a special customized control optimization network for the kneading technique. When processing the massage instructions related to the kneading technique, this network will optimize according to these specific parameter ranges to better achieve the massage effect of the kneading technique.

[0034] Next, the server obtains sample simulated massage instruction data from the simulated test environment of the massage sofa. For example, the simulation environment can set some virtual user demand scenarios, such as simulating the user demand of neck and shoulder fatigue after a day of work. In this scenario, a series of massage instruction data may be generated, such as first performing a kneading massage on the neck with an intensity of level 5 and a frequency of 20 times / minute for 3 minutes, and then performing a tapping massage on the shoulder with an intensity of level 6 and a frequency of 18 times / minute for 2 minutes. These simulated massage instruction data are part of the sample simulated massage instruction data.

[0035] The server loads the acquired sample simulated massage command data into the basic control optimization network and the customized control optimization network respectively. For example, for the above-mentioned simulated neck and shoulder massage command data, when loaded into the basic control optimization network, the network may optimize the instructions according to the previously learned user habits and basic massage logic. For example, the intensity of the neck massage may be adjusted between 4.5 and 5.5 levels, and the frequency may be adjusted between 18 and 22 times / minute. This is the optimization data of the basic control optimization network. Similarly, when loaded into the control optimization network customized for neck and shoulder massage, this network may further optimize the instructions according to its specially customized parameters, such as adjusting the intensity of the neck massage more accurately to 4.8 levels and the frequency to 20 times / minute. This is the optimization data of the customized control optimization network. These optimization data are respectively used as sample massage control data.

[0036] The server will collect comfort feedback data from users who actually use the massage sofa. For example, after the user performs a massage according to the above sample massage control data, the server will pop up a feedback interface to ask the user to rate the comfort of the massage. The rating can be from 1 to 5 points, with 1 point being very uncomfortable and 5 points being very comfortable. At the same time, the user can also fill in some specific feedback information, such as "the massage force is a bit strong in the shoulders" or "the frequency of the neck massage feels very appropriate".

[0037] Based on the user's comfort feedback data, the server adds training supervision data to the sample massage control data. If the user rates the comfort of the neck massage as 3 points, and the feedback strength is a bit strong, the server will add supervision data for the neck massage strength in the sample massage control data, such as marking that the strength value needs to be appropriately reduced. If the user rates the comfort of the shoulder massage as 4 points, but says that the massage area can be expanded a little, the server will add supervision data to the relevant sample massage control data for the shoulder massage, marking the range of the massage area that needs to be adjusted.

[0038] The server sets the number of training samples to 1000. When enough sample massage control data with added training supervision data is collected, these data are used as massage program learning data. For example, after a period of collection, the server has collected 1000 sets of sample massage control data with added training supervision data, which covers different massage techniques, different massage areas, and different user feedback.

[0039] Assume that the set machine learning network is a neural network, and its initial parameters are randomly set. The server inputs 1,000 sets of massage program learning data into this neural network. Taking the weight parameters in the neural network as an example, during the learning process, if it is found that in a set of massage program learning data, the neck massage intensity is too high and is associated with the discomfort reported by the user, then the neural network will adjust the weight parameters related to the neck massage intensity, so that in the subsequent evaluation, it can more accurately identify that the excessive intensity may cause discomfort. After a large number of such learning processes, the neural network continuously adjusts its parameters and finally generates a comfort evaluation network. This comfort evaluation network can extract the massage instruction description vector from the massage instruction optimization data and generate a comfort evaluation parameter based on this vector. For example, when a new set of massage instruction optimization data is input, the comfort evaluation network can analyze the massage technique, intensity, frequency, massage area and other information in this set of data, convert this information into a massage instruction description vector, and then calculate a comfort evaluation parameter based on the previously learned parameters. This parameter can be a value between 0 and 1, where 0 means very uncomfortable and 1 means very comfortable.

[0040] Next, the server selects a set of massage instruction data, such as massage instruction data for waist fatigue, including first performing a massage with an intensity of level 6 and a frequency of 20 times / minute for 5 minutes, and then performing a kneading massage with an intensity of level 5 and a frequency of 18 times / minute for 3 minutes. This set of data is loaded into the basic control optimization network and the customized control optimization network respectively. After the basic control optimization network optimizes the instructions according to its own parameters, the massage instruction optimization data corresponding to the basic control optimization network is obtained. For example, the intensity of the massage may be adjusted to between 5.5 and 6.5 levels, and the frequency may be adjusted to between 18 and 22 times / minute. The customized control optimization network also optimizes the instructions to obtain the massage instruction optimization data corresponding to the customized control optimization network, which may more accurately adjust the intensity of the massage to level 5.8 and the frequency to 20 times / minute.

[0041] The server inputs the massage instruction optimization data corresponding to the basic control optimization network into the comfort evaluation network. The comfort evaluation network calculates the comfort evaluation parameter corresponding to the basic control optimization network based on the previously learned parameters and algorithms, assuming it is 0.6. Then the massage instruction optimization data corresponding to the customized control optimization network is also input into the comfort evaluation network, and the comfort evaluation parameter corresponding to the customized control optimization network is calculated, assuming it is 0.7.

[0042] The server calculates the comfort difference between the basic control optimization network and the customized control optimization network. From the above example, we can see that the comfort evaluation parameter corresponding to the basic control optimization network is 0.6, and the comfort evaluation parameter corresponding to the customized control optimization network is 0.7, so the comfort difference is 0.7 - 0.6 = 0.1.

[0043] The server optimizes the parameters of the customized control optimization network based on this comfort difference value. If the comfort difference value is positive, it means that the customized control optimization network has a higher comfort level, but it can be further optimized. For example, in the customized control optimization network, parameters related to comfort may include parameters such as massage intensity, frequency, and massage head motion trajectory. If it is found that the adjustment of massage intensity within a certain range has a greater impact on the improvement of comfort, the server will further refine the parameter adjustment in this range. For example, the intensity of waist massage in the original customized control optimization network was adjusted between 5.5 and 6.5 levels. Now, based on the analysis of the comfort difference value, it is found that controlling the intensity more accurately between 5.8 and 6.2 levels may further improve comfort, so the server adjusts the relevant parameters in the customized control optimization network.

[0044] After the above-mentioned enhanced parameter learning, the parameters of the customized control optimization network have been optimized to better meet the requirements of improving comfort. The server outputs the customized control optimization network after the enhanced parameter learning as a target control optimization network. This target control optimization network can be directly applied to the massage sofa control system to optimize any input target massage control data, and then control the massage sofa control system based on the optimized massage control optimization data. For example, when the user inputs a new massage demand, the target control optimization network will process this demand according to its own optimized parameters and generate the optimal massage control plan, so that the massage sofa can provide users with more comfortable massage services according to this plan.

[0045] Based on the above steps, the embodiment of the present application realizes a high degree of personalization and optimization of massage control by combining the basic control optimization network and the customized control optimization network. The supervised training using sample massage control data and its comfort feedback data effectively improves the parameter optimization effect of the machine learning network, thereby generating an accurate comfort evaluation network. The comfort evaluation network can accurately evaluate the massage instructions generated by the customized control optimization network, and perform enhanced learning on the customized control optimization network based on the evaluation parameters, and finally output the optimized target control optimization network, thereby significantly improving the comfort and user experience of the massage sofa, and realizing the intelligent and personalized customization of massage control.

[0046] In a possible implementation manner, obtaining a basic control optimization network and a customized control optimization network includes: Get the initial control optimization network.

[0047] Initialization parameters of the initial control optimization network are learned based on a preset sample massage control optimization data sequence to generate a basic control optimization network, wherein the sample massage control optimization data sequence includes unsupervised sample massage control optimization data.

[0048] The basic control optimization network is customized based on the set massage instruction set to generate a customized control optimization network.

[0049] In a possible implementation, the basic control optimization network is customized based on the set massage instruction set to generate a customized control optimization network, including: A plurality of massage mode labels are set, wherein the massage mode labels include massage technique mode labels and / or massage area mode labels.

[0050] The set massage instruction set is clustered based on each of the massage mode labels to generate massage instruction clusters corresponding to each of the massage mode labels.

[0051] The control optimization network is customized for the basic control optimization network based on each of the massage instruction clusters, and a customized control optimization network corresponding to each of the massage mode labels is generated.

[0052] In a possible implementation, generating sample massage control data using the basic control optimization network and / or the customized control optimization network includes: Get sample simulated massage instruction data.

[0053] The sample simulated massage instruction data are loaded into the basic control optimization network and the customized control optimization network respectively, and the optimization data of the basic control optimization network and the optimization data of the customized control optimization network are used as sample massage control data respectively.

[0054] In a possible implementation, the steps of using the sample massage control data after adding the training supervision data as the massage program learning data to optimize the parameters of the set machine learning network and generate the comfort evaluation network include: The number of training samples is set, and the sample massage control data after adding the training supervision data is used as the massage program learning data.

[0055] If the number of massage program learning data is not less than the number of training samples, the parameters of the set machine learning network are optimized based on the massage program learning data to generate a comfort assessment network, wherein the comfort assessment network is used to extract a massage instruction description vector from the massage instruction optimization data and generate comfort assessment parameters based on the massage instruction description vector.

[0056] In a possible implementation, performing comfort evaluation on the massage instruction optimization data generated by the customized control optimization network based on the comfort evaluation network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, includes: The same set massage instruction data are loaded into the basic control optimization network and the customized control optimization network respectively, and massage instruction optimization data corresponding to the basic control optimization network and massage instruction optimization data corresponding to the customized control optimization network are generated.

[0057] Based on the comfort evaluation network, comfort evaluation is performed on the massage instruction optimization data corresponding to the basic control optimization network and the massage instruction optimization data corresponding to the customized control optimization network, respectively, to generate comfort evaluation parameters corresponding to the basic control optimization network and the comfort evaluation parameters corresponding to the customized control optimization network.

[0058] The comfort difference value between the basic control optimization network and the customized control optimization network is determined based on the comfort evaluation parameter corresponding to the basic control optimization network and the comfort evaluation parameter generated by the customized control optimization network.

[0059] Parameters of the customized control optimization network are optimized according to the comfort difference value.

[0060] In a possible implementation, performing comfort evaluation on the massage instruction optimization data generated by the customized control optimization network based on the comfort evaluation network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, includes: A threshold value is set, and cyclic learning data is acquired, wherein a comfort evaluation parameter corresponding to the cyclic learning data meets the threshold value.

[0061] Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network to generate comfort evaluation parameters corresponding to the customized control optimization network.

[0062] If the comfort evaluation parameter is less than a threshold value, cyclic parameter learning is performed on the customized control optimization network based on the cyclic learning data until the comfort evaluation parameter corresponding to the customized control optimization network meets the threshold value.

[0063] In a possible implementation, the method further includes: Based on the target control optimization network, any input target massage control data is optimized, and based on the optimized massage control optimization data, the massage sofa control system is controlled.

[0064] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0065] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.

Claims

1. An optimization method based on a massage sofa control system, characterized in that: The method comprises: Acquire a basic control optimization network and a customized control optimization network of a massage sofa control system, wherein the customized control optimization network is obtained by customizing the basic control optimization network by setting a massage instruction set; Generate sample massage control data using the basic control optimization network and / or the customized control optimization network, and add training supervision data to the sample massage control data based on comfort feedback data of the sample massage control data, so as to optimize parameters of a set machine learning network by using the sample massage control data after adding the training supervision data as massage program learning data, and generate a comfort evaluation network; Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, and the customized control optimization network after the enhanced parameter learning is output as a target control optimization network.

2. The optimization method based on the massage sofa control system according to claim 1, characterized in that: Get basic and custom control optimization networks, including: Obtaining an initial control optimization network; Performing initialization parameter learning on the initial control optimization network based on a preset sample massage control optimization data sequence to generate a basic control optimization network, wherein the sample massage control optimization data sequence includes unsupervised sample massage control optimization data; The basic control optimization network is customized based on the set massage instruction set to generate a customized control optimization network.

3. The optimization method based on the massage sofa control system according to claim 2, characterized in that: Customizing the basic control optimization network based on the set massage instruction set to generate a customized control optimization network includes: Setting a plurality of massage mode labels, wherein the massage mode labels include massage technique mode labels and / or massage area mode labels; Clustering the set massage instruction set based on each of the massage mode labels to generate massage instruction clusters corresponding to each of the massage mode labels; The control optimization network is customized for the basic control optimization network based on each of the massage instruction clusters, and a customized control optimization network corresponding to each of the massage mode labels is generated.

4. The optimization method based on the massage sofa control system according to claim 2, characterized in that: Generating sample massage control data using the basic control optimization network and / or the customized control optimization network includes: Get sample simulated massage instruction data; The sample simulated massage instruction data are loaded into the basic control optimization network and the customized control optimization network respectively, and the optimization data of the basic control optimization network and the optimization data of the customized control optimization network are used as sample massage control data respectively.

5. The optimization method based on the massage sofa control system according to claim 4, characterized in that: The steps of using the sample massage control data after adding the training supervision data as the massage program learning data to optimize the parameters of the set machine learning network and generate the comfort evaluation network include: The number of training samples is set, and the sample massage control data after adding the training supervision data is used as the massage program learning data; If the number of massage program learning data is not less than the number of training samples, the parameters of the set machine learning network are optimized based on the massage program learning data to generate a comfort assessment network, wherein the comfort assessment network is used to extract a massage instruction description vector from the massage instruction optimization data and generate comfort assessment parameters based on the massage instruction description vector.

6. The optimization method based on the massage sofa control system according to any one of claims 1 to 5, characterized in that: Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, including: Loading the same set massage instruction data into the basic control optimization network and the customized control optimization network respectively, generating massage instruction optimization data corresponding to the basic control optimization network and massage instruction optimization data corresponding to the customized control optimization network; Based on the comfort evaluation network, respectively, the massage instruction optimization data corresponding to the basic control optimization network and the massage instruction optimization data corresponding to the customized control optimization network are evaluated for comfort, and a comfort evaluation parameter corresponding to the basic control optimization network and a comfort evaluation parameter corresponding to the customized control optimization network are generated; Determine a comfort difference value between the basic control optimization network and the customized control optimization network based on the comfort evaluation parameter corresponding to the basic control optimization network and the comfort evaluation parameter generated by the customized control optimization network; Parameters of the customized control optimization network are optimized according to the comfort difference value.

7. The optimization method based on the massage sofa control system according to any one of claims 1 to 5, characterized in that: Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network, so as to perform enhanced parameter learning on the customized control optimization network based on the comfort evaluation parameters, including: Setting a threshold value and acquiring cyclic learning data, wherein the comfort evaluation parameter corresponding to the cyclic learning data meets the threshold value; Based on the comfort evaluation network, a comfort evaluation is performed on the massage instruction optimization data generated by the customized control optimization network to generate a comfort evaluation parameter corresponding to the customized control optimization network; If the comfort evaluation parameter is less than a threshold value, cyclic parameter learning is performed on the customized control optimization network based on the cyclic learning data until the comfort evaluation parameter corresponding to the customized control optimization network meets the threshold value.

8. The optimization method based on the massage sofa control system according to any one of claims 1 to 5, characterized in that: The method further comprises: Based on the target control optimization network, any input target massage control data is optimized, and based on the optimized massage control optimization data, the massage sofa control system is controlled.

9. An optimization system based on a massage sofa control system, characterized in that: include: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; The memory is used to store computer programs; the processor is used to implement the optimization method steps based on the massage sofa control system according to any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the optimization method based on the massage sofa control system described in any one of claims 1 to 8 are implemented.

Images