An intelligent sofa control method and an intelligent sofa based on user habits

Abstract

The present invention provides an intelligent sofa control method and an intelligent sofa based on user habits, relating to the technical field of sofa control. The method includes: collecting the maximum depression amount of the sofa seat cushion in the user's usage state; establishing a sofa parameter recommendation model with the maximum depression amount of the sofa seat cushion as the state vector and the sofa parameters as the control variables; collecting the maximum depression amount of the sofa seat cushion and the corresponding sofa parameters with a maintenance duration greater than a preset maintenance duration; training a support vector regression model using a training set to obtain an optimal hyperplane that makes the prediction error less than a preset prediction error; collecting the real-time maximum depression amount of the sofa seat cushion; inputting the real-time maximum depression amount of the sofa seat cushion into the trained support vector regression model to output predicted sofa parameters; and controlling the sofa according to the predicted sofa parameters. The present invention can recommend sofa parameters that ensure the user's sitting posture is healthy and meet the comfort level and control the sofa.

Description

Technical Field

[0001] The present invention relates to the technical field of sofa control, and particularly to an intelligent sofa control method and an intelligent sofa based on user habits. Background Art

[0002] An intelligent sofa is a high-tech sofa that combines sensors and a control system. It monitors data such as the user's weight distribution, seat cushion pressure, and backrest angle through intelligent algorithms, and automatically adjusts various parameters to optimize the user's sitting comfort. This kind of sofa can achieve personalized sitting posture adjustment, providing higher comfort and health protection.

[0003] The intelligent sofa control method based on user habits usually involves using sensors to monitor data such as the user's sitting posture, usage time, and frequency. The system analyzes this data, learns the user's habits and preferences, and then automatically adjusts the hardness, angle, and other settings of the sofa to meet the user's comfort needs. It is very necessary to perform intelligent sofa control based on user habits because it can improve personal comfort and health. The intelligent adjustment that adapts to the user's habits can help reduce back and neck pain, improve the sitting posture quality, thereby enhancing the quality of life and work efficiency.

[0004] However, in the initial stage of using the sofa, the existing intelligent sofa control methods need to obtain the prior knowledge of user habits in advance, and then learn how to adjust the sofa to the posture of the user's habit according to the prior knowledge. However, when the user first uses the sofa, they are not sure which sofa parameter settings can achieve the best comfort. Some people just set the parameters by themselves and will not waste time adjusting anymore when they feel that the comfort is appropriate. In this state, it may not be the sofa parameter settings that enable the user to achieve the best comfort. If the data collected in this way is used as the user's habit for learning and automatic adjustment, it not only increases the user's burden of use, but also cannot ensure the accuracy of the sofa adjustment only based on the user's habit. The adjustment method is too single to improve the use experience of the sofa, and the practicability is insufficient, and the adjustable performance of the sofa cannot be fully exerted. Summary of the Invention

[0005] In order to solve the technical problems existing in the existing intelligent sofa control methods, which increase the user's burden of use, and cannot ensure the accuracy of the sofa adjustment only based on the user's habit, the adjustment method is too single to improve the use experience of the sofa, the practicability is insufficient, and the adjustable performance of the sofa cannot be fully exerted, the present invention provides an intelligent sofa control method and an intelligent sofa based on user habits.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First Aspect

[0008] An intelligent sofa control method based on user habits provided by an embodiment of the present invention includes:

[0009] S1: Collect the maximum depression of the sofa seat cushion in the user's usage state;

[0010] S2: Establish a sofa parameter recommendation model with the maximum depression of the sofa seat cushion as the state vector and the sofa parameters as the control variables, taking the ergonomic standard as the adjustment constraint condition for the H point of the sofa. The sofa parameters include the sofa backrest angle, the sofa seat cushion height, the sofa seat cushion depth, and the sofa seat cushion tilt;

[0011] S3: Collect the maximum depression of the sofa seat cushion and the corresponding sofa parameters whose maintenance duration is greater than the preset maintenance duration. The collected maximum depression of the sofa seat cushion and the sofa parameters form a training set;

[0012] S4: Combine the radial basis kernel function with an adaptive width adjustment parameter and the weight proportional to the maintenance duration, and use the training set to train the support vector regression model to obtain the optimal hyperplane that makes the prediction error less than the preset prediction error, where the prediction error is the difference between the predicted sofa parameters and the true sofa parameters;

[0013] S5: Collect the real-time maximum depression of the sofa seat cushion;

[0014] S6: Input the real-time maximum depression of the sofa seat cushion into the trained support vector regression model and output the predicted sofa parameters;

[0015] S7: Control the sofa according to the predicted sofa parameters.

[0016] In a second aspect

[0017] An intelligent sofa provided by an embodiment of the present invention is applied to the intelligent sofa control method based on user habits as described in the first aspect.

[0018] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0019] In the present invention, first, the maximum downward limit of the sofa seat cushion that can reflect the user's usage status is collected. Then, combined with ergonomic standards as the constraint for the sofa H-point, a sofa parameter recommendation model is determined. When the maximum downward limit of the sofa seat cushion is obtained, the sofa parameter range of the H-point that not only conforms to the healthy sitting posture of the human body but also enables the user to achieve sufficient comfort can be recommended, namely the sofa backrest angle, the sofa seat cushion height, the sofa seat depth, and the sofa seat inclination. And a set of sofa parameters is randomly selected according to this to adjust the seat. Then, the user feedback is obtained, and the sofa parameters with a duration meeting the requirements are retained to form a training set, and the support vector regression model is trained using the training set. By adaptively adjusting the width of the kernel function, the model can more flexibly adapt to the characteristics of the data, and the introduction of time weights ensures that the data used for a long time has a greater impact on the model training, thus more truly reflecting the impact of usage conditions on the performance of the sofa seat cushion. This not only improves the generalization ability of the model but also makes the sofa design more in line with the actual usage requirements. It enables the model to learn the complex and implicit correlation relationship between the maximum depression of the sofa seat cushion and the sofa parameters, improves the prediction accuracy and adaptability of the model, discovers more reasonable sofa parameters that meet the requirements of health and comfort, avoids the user burden caused by the user's self-adjustment during the initial use process, can also make reasonable recommendations considering the user's health and comfort and deeply explore more reasonable sofa parameters, enhances the user experience, and increases the product competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a schematic flowchart of an intelligent sofa control method based on user habits provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The technical solutions in the present invention will be described below with reference to the drawings.

[0023] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0024] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0025] Refer to the attached Figure 1 , which shows a schematic flow chart of an intelligent sofa control method provided by an embodiment of the present invention based on user habits.

[0026] The embodiment of the present invention provides an intelligent sofa control method based on user habits. The processing flow of the intelligent sofa control method based on user habits may include the following steps:

[0027] S1: Collect the maximum depression amount of the sofa seat cushion in the user's usage state.

[0028] Among them, the maximum depression amount of the sofa seat cushion refers to the sinking depth of the seat cushion from the uncompressed state to the maximum compression state when the user sits on the sofa. This measurement reflects the softness, load-bearing capacity, and elasticity of the sofa seat cushion material. Measuring the maximum depression amount is extremely important for evaluating the comfort and functionality of the sofa, as it affects sitting posture support and comfort during long-term use.

[0029] S2: Establish a sofa parameter recommendation model with the maximum depression amount of the sofa seat cushion as the state vector and the sofa parameters as the control variables, with the ergonomic standards as the adjustment constraint conditions for the H point of the sofa.

[0030] Among them, the sofa parameters include the sofa backrest angle, the sofa seat cushion height, the sofa seat cushion depth, and the sofa seat cushion inclination.

[0031] Specifically, the ergonomic standard can be the Ergonomic Requirements for Seats of GB / T 14755-1993. Performing self-recommendation of parameters at the initial stage, this function can help users quickly find relatively comfortable sofa parameters that meet the sitting posture health standards. Even if the user is not clear about the specific parameter settings, the system can give suitable recommendations to ensure the comfort of the first use.

[0032] Among them, the sofa H-point refers to the point on the sofa that comes into contact with the human hip. This point is a key position for determining how the seat supports the user's body and is particularly important in ergonomics as it relates to the comfort of sitting posture and its impact on health. Using ergonomic standards as constraints for adjusting the sofa H-point, an intelligent sofa parameter recommendation model is created. This model uses the maximum sag of the sofa seat cushion as the state vector and sofa parameters (such as backrest angle, seat cushion height, seat cushion depth, and seat cushion tilt) as control variables. In this way, the model can recommend sofa parameter settings based on the actual usage state of the user to optimize the sitting posture and increase comfort. The core goal of this model is to ensure that the design and configuration of the sofa meet ergonomic requirements and use the H-point to ensure the comfort of the user during the adjustment process, improving the user's seating experience and physical health.

[0033] In a possible implementation, the sofa parameter recommendation model established by the state variable and the control variable is specifically:

[0034]

[0035] Among them, represents the backrest angle of the sofa, represents the seat cushion height of the sofa, d represents the seat cushion depth of the sofa, the sofa seat cushion tilt, z represents the maximum sag of the sofa seat cushion, respectively represent the abscissa, ordinate, and vertical coordinate of the sofa H-point in the user's usage state in the coordinate system with the midpoint of the rear edge of the sofa seat cushion as the origin of coordinates. Among them, the abscissa is the coordinate of the sofa H-point in the seat cushion width direction, the ordinate is the coordinate of the sofa H-point in the seat cushion depth direction, and the vertical coordinate is the coordinate of the sofa H-point perpendicular to the seat cushion plane.

[0036] It should be noted that this model will randomly select a set of parameters from the set of sofa parameters that meet the adjustment constraint conditions calculated and output them, and obtain the user's feedback. This sofa parameter recommendation model calculates the position of the sofa H-point in three dimensions by combining the maximum sag of the sofa seat cushion in the actual usage state of the user and other relevant sofa parameters (sofa backrest angle, sofa seat cushion height, sofa seat cushion depth, sofa seat cushion tilt). The model uses these calculated coordinate values to ensure that the recommended sofa parameters not only meet ergonomic requirements but also can improve the user's comfort. By randomly selecting from the set of qualified parameters and obtaining user feedback to optimize and adjust the recommendation algorithm, this method can dynamically adjust the sofa settings to adapt to the specific needs of different users, thereby improving the user experience and satisfaction.

[0037] In a possible implementation, the adjustment constraint conditions of the sofa parameter recommendation model are specifically:

[0038]

[0039] Among them, W represents the total width of the sofa seat cushion, D represents the total depth of the sofa seat cushion, a = 3 cm represents the height adjustment coefficient based on ergonomic standards.

[0040] It should be noted that this adjustment constraint ensures that the sofa design meets specific ergonomic standards to improve user comfort and support. The constraints include that the position of the H-point of the sofa must be within the central one-tenth of the seat cushion width, the depth is between half and three-quarters of the seat cushion depth, and the vertical position of the H-point needs to be within the allowable range of the seat cushion height. In addition, the sofa backrest angle should be maintained between 95 and 105 degrees. The range selection of these parameters is designed to provide sufficient support while maintaining overall comfort and functionality.

[0041] S3: Collect the maximum sag of the sofa seat cushion with a maintenance duration greater than the preset maintenance duration and the corresponding sofa parameters.

[0042] Among them, the collected maximum sag of the sofa seat cushion and the sofa parameters form a training set.

[0043] Among them, the maintenance duration refers to the length of time that a user maintains a certain sitting posture on the sofa. The system collects data by monitoring those usage states with a maintenance duration exceeding the preset duration. Only when the sitting posture of the user on the sofa lasts longer than the set threshold, the maximum sag of the seat cushion and the corresponding sofa parameters in this state will be recorded and added to the training set. Such a data collection method helps to filter out those data that may have short sitting posture maintenance times due to discomfort, thus ensuring that the data in the training set reflects the states with higher user satisfaction. In this way, the established training set can more accurately reflect the sofa parameters with good comfort and ergonomic design, providing high-quality basic data for subsequent model training and parameter optimization. It can not only record user preferences, but also complete the recommendation of initial parameter settings based on ergonomic standards, which can make up for the problem that customers are also unable to determine which set of sofa parameters is the most suitable, reduce the health impact of abnormal sitting postures on users, and ensure the comfort of users.

[0044] It should be noted that those skilled in the art can set the size of the preset maintenance duration according to actual needs, and the present invention does not limit this here.

[0045] In a possible implementation manner, S3 specifically includes:

[0046] S301: Initialize the sofa parameters.

[0047] S302: Collect the maximum sag of the sofa seat cushion.

[0048] S303: Input the maximum sag of the sofa seat cushion into the sofa parameter recommendation model, and randomly output recommended sofa parameters that meet the adjustment constraint conditions.

[0049] S304: Determine whether the maintenance duration under the recommended sofa parameters is greater than the preset maintenance duration. If so, retain the recommended sofa parameters with a maintenance duration greater than the preset maintenance duration and the corresponding maximum sag of the sofa seat cushion as a data pair. Otherwise, return to step S303.

[0050] It should be noted that those skilled in the art can set the size of the preset maintenance duration according to actual needs, and the present invention does not make any limitations here.

[0051] S305: Determine whether the number of data pairs is greater than the preset number of data pairs. If so, terminate the iteration and enter step S306. Otherwise, return to step S303.

[0052] It should be noted that those skilled in the art can set the size of the preset number of data pairs according to actual needs, and the present invention does not make any limitations here.

[0053] S306: Output the retained recommended sofa parameters and the corresponding maximum sag of the sofa seat cushion as a training set.

[0054] Specifically, first, initialize the sofa parameters. Subsequently, the system collects the maximum sag of the sofa seat cushion during use and inputs this data into the sofa parameter recommendation model, which randomly outputs sofa parameters that meet the adjustment constraint conditions. Then, the system checks whether the user's maintenance duration under these parameters exceeds the preset value. If so, this set of parameters and the corresponding sag data are saved. If not, new parameter combinations are searched again. This process is repeated until a sufficient number of data pairs are collected. Finally, these data pairs are output as a training set for further optimizing and adjusting the recommendation model. This iterative process ensures that the recommended sofa parameters can actually improve the user's comfort, and the collected data will support future model training and refinement.

[0055] S4: Combine the radial basis kernel function with an adaptive width adjustment parameter and the weight proportional to the maintenance duration, and use the training set to train the support vector regression model to obtain the optimal hyperplane that makes the prediction error less than the preset prediction error.

[0056] Among them, the prediction error is the difference between the predicted sofa parameters and the true sofa parameters.

[0057] Among them, the adaptive width adjustment parameter is a parameter in the radial basis kernel function, which is used to adjust the width of the function, thereby controlling the sensitivity of the kernel function to neighboring data points. According to different data distributions, automatically adjusting this parameter can improve the fitting effect and generalization ability of the model. The radial basis kernel function is a kernel function commonly used in support vector machines, especially suitable for feature space mapping of non-linear data. The support vector regression model is a regression analysis method based on the principle of support vector machines. It tries to find a function that minimizes the difference between the actual output value and the predicted output value while keeping the complexity of the model as low as possible. In support vector machines, the optimal hyperplane refers to the decision interface that can maximize the margin (i.e., the minimum distance from the data points to the separating plane), and this plane is as close as possible to all data points (within a certain error tolerance).

[0058] The system uses a support vector regression model with a radial basis kernel function, combined with an adaptive width adjustment parameter, to predict the sofa parameters. The goal of training this model is to find an optimal hyperplane that minimizes the prediction error between the predicted sofa parameters and the actual parameters. Through the weight proportional to the maintenance duration, the model attaches more importance to the data points with a long-term comfortable sitting posture, thereby improving the accuracy of the model in predicting the comfortable sitting posture parameters. This method can ensure that each parameter recommendation conforms to the actual usage experience of the user as much as possible, thereby enhancing the user's satisfaction and comfort experience. Moreover, during the recommendation process, this model determines the random recommendation method as an accurate recommendation method to ensure that each recommendation can maintain a state where the maintenance duration is greater than the preset maintenance duration.

[0059] It should be noted that those skilled in the art can set the size of the preset prediction error according to actual needs, and the present invention does not make any limitations here.

[0060] In a possible implementation manner, S4 specifically includes:

[0061] S401: Use the radial basis kernel function with an adaptive width adjustment parameter to map the maximum sag of the sofa seat cushion to a high-dimensional space for linear separation:

[0062]

[0063] Among them, represents the radial basis kernel function value representing the similarity between the maximum sag of the first sofa seat cushion and the maximum sag of the second sofa seat cushion, exp represents the exponential function, represents the width adjustment parameter that controls the radial basis kernel function, represents the square of the Euclidean distance calculation, represents the input data set including all the maximum sags of the sofa seat cushions in the training set, n represents the total number of samples in the training set, denotes the maximum sag of the i th sofa cushion in the training set, α and β represent the local adjustment parameter and the global adjustment parameter respectively, denotes the mean value of the maximum sag of the sofa cushions in the training set, represents calculating the Euclidean distance, and log represents the logarithmic function, denotes the standard deviation of the maximum sag of the sofa cushions in the training set, represents a constant to avoid a zero denominator.

[0064] Optionally, the constant is specifically .

[0065] It should be noted that in the calculation process of the width adjustment parameter, by calculating the Euclidean distance from each data point to the mean value, it is used to capture the discreteness of the data. The exponential term in the Taylor series expansion, that is, the numerator part, is used to handle the non-linear characteristics of the distance, making the influence of the near-distance data on the width parameter greater. The introduction of the logarithmic term ensures the robustness of the model to outliers and at the same time avoids overfitting. Combining the squared distance term of each point with the logarithmic function can better capture the local structure of the data and the influence of outliers on the parameters.

[0066] The system uses a radial basis kernel function with an adaptive width adjustment parameter to map the maximum sag of the sofa cushion to a high-dimensional space. This mapping allows the model to perform linear separation in the high-dimensional space, thereby effectively capturing the complex non-linear relationship between the maximum sag of the sofa cushion and the sofa adjustment parameters (such as the seat cushion tilt angle, the backrest angle, and the seat height). By mapping the data to a high-dimensional space, the model can find a hyperplane in this space, and this hyperplane can clearly express the relationship between the sag and different sofa adjustment parameters, and then achieve accurate prediction and adjustment of these relationships. The linear inseparability problem that cannot be solved in the original space is solved by increasing the dimension.

[0067] In a possible implementation manner, the local adjustment parameter and the global adjustment parameter are solved based on the maximum entropy principle. The specific solution process includes:

[0068] According to the maximum entropy principle, the selection of the local adjustment parameter and the global adjustment parameter can make the model achieve the maximum entropy while capturing global and local information, so that these two parameters can enable the model to achieve the best balance when capturing global and local features. Through the estimation of the maximum entropy principle, it is ensured that the formula has sufficient sensitivity to the global and local characteristics of the data.

[0069] Based on the maximum entropy principle, the objective function for solving the local adjustment parameter and the global adjustment parameter is determined. The objective function is specifically:

[0070]

[0071] Among them, H represents the information entropy of the training set, represents the local adjustment parameter and the global adjustment parameter when the information entropy is maximized, represents the probability density of the i th sample in the training set calculated based on kernel density estimation.

[0072] Calculate the partial derivatives of the information entropy with respect to the local adjustment parameter and the global adjustment parameter respectively to obtain the conversion objective function:

[0073]

[0074] Among them, represents taking the partial derivative.

[0075] Solve the conversion objective function by Newton's method iteration to obtain the local adjustment parameter and the global adjustment parameter.

[0076] It should be noted that partial derivative optimization can find the optimal α and β values according to the change rate of entropy, so that the model can retain the maximum amount of information in the data during the training process, which conforms to the maximum entropy principle. By maximizing the information entropy, the model can better capture the diversity and uncertainty of the data distribution, thereby improving the generalization ability and avoiding overfitting. By taking the partial derivative of them, the model can more accurately adjust the performance of the kernel function in different regions, making its response to similar data points more sensitive, so as to better capture the non-linear relationship between the input features and the target variables.

[0077] S402: According to the maximum depression of the sofa seat cushion after high-dimensional mapping and its corresponding sofa parameters, establish an optimization function for the hyperplane parameters with the goal of minimizing the prediction error, where the hyperplane parameters include the weight vector and the bias vector:

[0078]

[0079] Among them, W represents the weight vector, b represents the bias vector, C represents the penalty parameter, and max represents taking the maximum value, represents the i th real sofa parameter value in the j th sample in the training set, , n represents the total number of samples in the training set, , , , respectively represent the true sofa backrest angle, true sofa seat height, true sofa seat depth, and true sofa seat inclination in the i th sample, represents the i th predicted sofa parameter value corresponding to the maximum sag of the sofa seat in the j th sample, represents the tolerance error, represents taking W and b when the function takes the minimum value, represents the weight of the i th sample in the training set, represents the duration of maintenance of the i th sample in the training set, represents the maximum duration of maintenance in the T duration of maintenance of the training set samples.

[0080] It should be noted that the introduction of the weighted sum of the weights makes the model automatically bias towards samples with longer maintenance durations during training, ensuring that the model is more in line with the sitting postures of users when recommending sofa parameters and improving the comfort of the user experience. During training, the weights of different training data are assigned according to the magnitude of the maintenance duration, so that the determined hyperplane can recommend sofa parameters in the optimal recommended manner to the greatest extent and is in line with the sitting postures of users. The weight solution method of the exponential function can increase the weight difference of high-maintenance-duration samples, and the weights of high-maintenance-duration samples will be higher, meeting the recommendation requirements.

[0081] Specifically, the loss function combines the squared loss and the penalty term. The regularization term used here helps prevent the model from overfitting, that is, over-adapting to the noise of the training data rather than the underlying data pattern. This can improve the generalization ability of the model on unseen data. By introducing the penalty term, the difference between the predicted value and the true value of the model is less than or equal to When there is no penalty, this method reduces the over-punishment of the model, allows for a certain degree of error, making the model more robust, especially effective when dealing with small fluctuations in real-world data. By setting the weight of each sample as an exponential function of the ratio of its maintenance duration to the maximum maintenance duration, the model automatically biases towards samples that users can maintain for a long time. This weighting method ensures that during model training and prediction, parameter settings that are more in line with users' long-term usage habits are given higher priority, thus better conforming to users' actual usage preferences and comfort levels. The optimization function combines the adjustment of the weight vector and the bias vector, and adjusts these parameters to find the optimal hyperplane that can minimize the overall prediction error and meet the above regularization and error tolerance conditions. Such a setting helps to find a model that can not only accurately predict but also widely adapt to various user needs. It improves the model's adaptability to the time users maintain a comfortable sitting position while maintaining prediction accuracy and the generalization ability of the model.

[0082] In a possible implementation manner, after S402, it further includes:

[0083] S402A: Adjust the hyperparameters of the support vector regression model by the gradient descent method with a momentum term, where the hyperparameters include the penalty parameter and the tolerance error.

[0084] It should be noted that by introducing the gradient descent method with a momentum term to adjust the hyperparameters of the support vector regression model, specifically including the penalty parameter and the tolerance error. The addition of the momentum term helps to accelerate the convergence process in gradient descent and helps to avoid getting stuck in local minima. Such a method enables the model to find a better parameter combination faster, thereby improving the prediction performance and stability of the model, ensuring that the final model can more accurately reflect and predict users' usage habits and comfort requirements.

[0085] In a possible implementation manner, S402A specifically includes:

[0086] S402A1: Calculate the partial derivatives of the objective function value with respect to the penalty parameter and the tolerance error respectively:

[0087]

[0088] where, J represents the objective function value, represents the sign function used to judge the direction of the difference, represents the indicator function, and when the indicator function takes 1, otherwise, the indicator function takes 0.

[0089] S402A2: Update the penalty parameter and the tolerance error according to the obtained partial derivatives until the update amplitude of each hyperparameter is less than the preset update amplitude:

[0090]

[0091] where represents the learning rate, represents assignment, and both represent the momentum terms for storing gradients, represents the momentum parameter.

[0092] Among them, the update amplitude refers to the amount by which the hyperparameter is updated in each iteration during the optimization process. Specifically, this measure reflects the magnitude of the change in the penalty parameter and the tolerance error from one iteration to the next. If the update amplitude is less than the preset threshold, it means that the hyperparameter has approached its optimal value, and further adjustment will have a minimal effect, so the iteration process can be stopped.

[0093] Optionally, the momentum parameter can be set to 0.9.

[0094] It should be noted that those skilled in the art can set the size of the preset update amplitude according to actual needs, and the present invention does not limit this here.

[0095] It should be noted that the momentum term can accelerate convergence, reduce oscillations by introducing the gradient accumulation of the previous steps during the hyperparameter update, and is especially suitable for dealing with narrow optimization regions and jumping out of local optima, thereby improving the optimization efficiency and stabilizing the parameter search process. Specifically, first, the partial derivatives of the objective function with respect to the penalty parameter and the tolerance error are calculated to determine how these parameters should be adjusted to minimize the objective function. Then, the hyperparameters are updated using the gradient descent method with momentum terms based on these partial derivatives. The introduction of the momentum term can be regarded as an acceleration technique, which speeds up the convergence rate and reduces oscillations by considering the influence of the previous gradients, thus making the parameter update process more efficient and stable. This method is particularly effective in guiding to the global optimal solution, especially in complex optimization landscapes, and helps to avoid falling into local minima.

[0096] S403: Solve the optimization function and establish the optimal hyperplane of the support vector regression model based on the obtained hyperplane parameters:

[0097]

[0098] where represents the input data x i.e., the function value of the maximum sag of the sofa seat cushion.

[0099] Specifically, first, the maximum sag of the sofa seat cushion is mapped to a high-dimensional space using a radial basis kernel function, which helps to reveal the complex non-linear relationship between the sag and the sofa adjustment parameters. Then, in S402, by setting an optimization problem, the prediction error is minimized while adjusting the weight vector and the bias vector to find the optimal hyperplane. Finally, in S403, this optimization problem is solved to determine the hyperplane parameters and establish the optimal hyperplane of the support vector regression model, ensuring that the model can accurately predict and adjust the sofa parameters, thereby improving the user experience and meeting personalized needs.

[0100] S5: Collect the maximum real-time sag of the sofa seat cushion.

[0101] S6: Input the maximum real-time sag of the sofa seat cushion into the trained support vector regression model and output the predicted sofa parameters.

[0102] S7: Control the sofa according to the predicted sofa parameters.

[0103] In a possible implementation, after S7, it further includes:

[0104] Retrain the support vector regression model at preset time intervals.

[0105] It should be noted that those skilled in the art can set the size of the preset time interval according to actual needs, and the present invention does not limit it here.

[0106] It can be understood that the implementation of retraining the support vector regression model at preset time intervals is to ensure the continuous update and improvement of the model.

[0107] In the actual application process, the whole solution describes a complete intelligent sofa parameter adjustment system. From data collection to model training and then to real-time application, the whole process is closely connected to form a closed-loop feedback system. First, a sofa parameter recommendation model is established to recommend sofa parameters based on the maximum sag of the sofa seat cushion collected. Then, using these data, a support vector regression model is established, and based on ergonomics and the sofa H-point as the basis and adjustment target for parameter recommendation, ensuring compliance with the healthy sitting posture standard and comfort requirements. Finally, using the trained model, the sofa settings are dynamically adjusted according to the user's immediate usage situation, providing continuous comfort and support. It can automatically learn and optimize, adapt to individual differences, enhance the user experience, ensure comfort, and reduce health problems caused by improper sitting postures during long-term use.

[0108] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include:

[0109] In the present invention, first, the maximum downward limit of the sofa seat cushion that can reflect the user's usage state is collected. Then, combined with ergonomic standards as the constraint for the H-point of the sofa, a sofa parameter recommendation model is determined. When the maximum downward limit of the sofa seat cushion is obtained, the sofa parameter range of the H-point that not only conforms to the healthy sitting posture of the human body but also enables the user to achieve sufficient comfort can be recommended, namely the sofa backrest angle, the sofa seat cushion height, the sofa seat cushion depth, and the sofa seat cushion inclination. And a set of sofa parameters is randomly selected according to this to adjust the seat. Then, user feedback is obtained, and the sofa parameters with a duration meeting the requirements are retained to form a training set, and the support vector regression model is trained using the training set. By adaptively adjusting the width of the kernel function, the model can more flexibly adapt to the characteristics of the data, and the introduction of time weights ensures that the data used for a long time has a greater impact on model training, thus more truly reflecting the influence of usage conditions on the performance of the sofa seat cushion. This not only improves the generalization ability of the model but also makes the sofa design more in line with the actual usage requirements. It enables the model to learn the complex and implicit correlation relationship between the maximum sag of the sofa seat cushion and the sofa parameters, improves the prediction accuracy and adaptability of the model, discovers more reasonable sofa parameters that meet the requirements of health and comfort, avoids the user burden caused by the user's self-adjustment during the initial use process, and can also make reasonable recommendations considering the user's health and comfort and deeply explore more reasonable sofa parameters, enhancing the user experience and increasing the product competitiveness.

[0110] An embodiment of the present invention provides an intelligent sofa, which is applied to the intelligent sofa control method based on user habits as described in the method embodiment.

[0111] A computer-readable storage medium provided by the present invention can implement the steps and effects of the intelligent sofa control method based on user habits in the above method embodiment. To avoid repetition, the present invention will not elaborate further.

[0112] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0113] In the present invention, first, the maximum downward limit of the sofa seat cushion that can reflect the user's usage state is collected. Then, combined with ergonomic standards as the constraint for the sofa H-point, a sofa parameter recommendation model is determined. When the maximum downward limit of the sofa seat cushion is obtained, it can recommend a range of sofa parameters for the H-point that not only conforms to the healthy sitting posture of the human body but also enables the user to achieve sufficient comfort, namely the sofa backrest angle, the sofa seat height, the sofa seat depth, and the sofa seat inclination. And a set of sofa parameters is randomly selected according to this to adjust the seat. Then, user feedback is obtained, and the sofa parameters with a maintenance duration meeting the requirements are retained to form a training set, and the support vector regression model is trained using the training set. By adaptively adjusting the width of the kernel function, the model can more flexibly adapt to the characteristics of the data, and the introduction of time weights ensures that the data used for a long time has a greater impact on model training, thus more truly reflecting the influence of usage conditions on the performance of the sofa seat cushion. This not only improves the generalization ability of the model but also makes the sofa design more in line with the actual usage requirements. It enables the model to learn the complex and implicit correlation relationship between the maximum depression of the sofa seat cushion and the sofa parameters, improves the prediction accuracy and adaptability of the model, discovers more reasonable sofa parameters that meet the requirements of health and comfort, avoids the user burden caused by the user's self-adjustment during the initial use process, can also make reasonable recommendations considering the user's health and comfort and deeply explore more reasonable sofa parameters, enhances the user's usage experience, and increases the product competitiveness.

[0114] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0115] The following points need to be explained:

[0116] (1) The attached drawings of the embodiments of the present invention only relate to the structures involved in the embodiments of the present invention, and other structures can refer to the general design.

[0117] (2) For clarity, in the attached drawings used to describe the embodiments of the present invention, the thickness of the layer or region is enlarged or reduced, that is, these drawings are not drawn according to the actual scale. It can be understood that when an element such as a layer, a film, a region, or a substrate is referred to as being "on" or "under" another element, the element can be "directly" on or under another element or there can be an intermediate element.

[0118] (3) Without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0119] The above are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A smart sofa control method based on user habits, characterized in that: include: S1: Collect the maximum sinking amount of the sofa cushion when the user is using it; S2: Taking the ergonomic standard as the adjustment constraint condition of the sofa H point, a sofa parameter recommendation model is established with the maximum sinking amount of the sofa cushion as the state vector and the sofa parameters as the control variables, wherein the sofa parameters include the sofa backrest angle, sofa cushion height, sofa cushion depth and sofa cushion inclination; S3: collecting the maximum sinking amount of the sofa cushion whose maintenance time is longer than the preset maintenance time and the corresponding sofa parameters, wherein the collected maximum sinking amount of the sofa cushion and the sofa parameters constitute a training set; S4: combining a radial basis kernel function with an adaptive width adjustment parameter and a weight that is proportional to the maintenance time, using the training set to train a support vector regression model, and obtaining an optimal hyperplane that makes the prediction error less than a preset prediction error, wherein the prediction error is the difference between the predicted sofa parameter and the actual sofa parameter; S5: Collect the maximum sinking amount of the sofa cushion in real time; S6: inputting the real-time maximum sinking amount of the sofa cushion into the trained support vector regression model, and outputting predicted sofa parameters; S7: controlling the sofa according to the predicted sofa parameters; The sofa parameter recommendation model established by the state variables and the control variables is specifically: in, Indicates the sofa backrest angle, Indicates the height of the sofa cushion. d Indicates the depth of the sofa cushion. Sofa seat cushion tilt, z Indicates the maximum sinking amount of the sofa cushion. They respectively represent the horizontal coordinate, vertical coordinate and vertical coordinate of point H of the sofa in the user's use state in the coordinate system with the midpoint of the rear edge of the sofa cushion as the coordinate origin, wherein the horizontal coordinate is the coordinate of point H of the sofa in the width direction of the seat cushion, the vertical coordinate is the coordinate of point H of the sofa in the depth direction of the seat cushion, and the vertical coordinate is the coordinate of point H of the sofa perpendicular to the plane of the seat cushion; Among them, the adjustment constraint conditions of the sofa parameter recommendation model are specifically: in, W Indicates the total width of the sofa cushion. D Indicates the total depth of the sofa cushion. Indicates the height adjustment factor based on ergonomic standards.

2. The method for controlling a smart sofa based on user habits according to claim 1, characterized in that: The S3 specifically includes: S301: Initializing the sofa parameters; S302: collecting the maximum sinking amount of the sofa cushion; S303: inputting the maximum sinking amount of the sofa cushion into the sofa parameter recommendation model, and randomly outputting the recommended sofa parameters that meet the adjustment constraint conditions; S304: Determine whether the maintenance time under the recommended sofa parameter is greater than the preset maintenance time, if so, retain the recommended sofa parameter whose maintenance time is greater than the preset maintenance time and the corresponding maximum sinking amount of the sofa cushion as a data pair, otherwise, return to step S303; S305: Determine whether the number of data pairs is greater than a preset number of data pairs. If so, terminate the iteration and proceed to step S306. Otherwise, return to step S303. S306: Outputting the retained recommended sofa parameters and the corresponding maximum sinking amount of the sofa cushion as a training set.

3. The method for controlling a smart sofa based on user habits according to claim 1, characterized in that: The S4 specifically includes: S401: Mapping the maximum sinking amount of the sofa cushion to a high-dimensional space using a radial basis kernel function with an adaptive width adjustment parameter for linear separation: in, represents the radial basis kernel function value representing the similarity between the maximum sinking amount of the first sofa cushion and the maximum sinking amount of the second sofa cushion, exp represents the exponential function, represents the width adjustment parameter that controls the radial basis kernel function, Indicates the calculation of the square of the Euclidean distance, represents the input data set including the maximum sinkage of all sofa cushions in the training set, n represents the total number of samples in the training set, Indicates the training set i Maximum sinking amount of a sofa seat cushion, α and β Represent local adjustment parameters and global adjustment parameters respectively, represents the mean maximum sinking amount of the sofa cushion in the training set, Indicates the calculation of Euclidean distance, log indicates the logarithmic function, represents the standard deviation of the maximum sinkage of the sofa cushion in the training set, represents a constant that avoids zero denominator; S402: According to the maximum sinking amount of the sofa cushion after high-dimensional mapping and its corresponding sofa parameters, an optimization function for hyperplane parameters is established with the goal of minimizing the prediction error, wherein the hyperplane parameters include a weight vector and a bias vector: in, W represents the weight vector, b represents the bias vector, C represents the penalty parameter, max represents the maximum value, Indicates the training set i The first j The actual sofa parameter values, , n represents the total number of samples in the training set, , , , Respectively represent i The real sofa back angle, real sofa cushion height, real sofa cushion depth and real sofa cushion inclination in the samples. Indicates i The maximum sinking amount of the sofa cushion in the sample corresponds to the j Predicted sofa parameter values, represents the tolerance error, It means taking the function when it takes the minimum value W and b , Indicates the training set i The weight of the samples, Indicates the training set i The duration of the sample, Indicates the maintenance time of the training set samples T The maximum duration of maintenance in S403: Solving the optimization function, and establishing the optimal hyperplane of the support vector regression model based on the hyperplane parameters obtained by the solution: in, Represents input data x That is, the function value of the maximum sinking amount of the sofa cushion.

4. The intelligent sofa control method based on user habits according to claim 3 is characterized in that: The local adjustment parameter and the global adjustment parameter are solved based on the maximum entropy principle, and the solving process specifically includes: The objective function for solving the local adjustment parameter and the global adjustment parameter is determined based on the maximum entropy principle. The objective function is specifically: in, H represents the information entropy of the training set, It represents the local adjustment parameters and global adjustment parameters that maximize the information entropy. Represents the first i The probability density of samples; The partial derivatives of the information entropy with respect to the local adjustment parameter and the global adjustment parameter are calculated respectively to obtain the conversion objective function: in, It means to find partial derivative; The conversion objective function is iteratively solved by Newton's method to obtain the local adjustment parameter and the global adjustment parameter.

5. The method for controlling a smart sofa based on user habits according to claim 3, characterized in that: After S402, the method further includes: S402A: Adjusting the hyperparameters of the support vector regression model by using a gradient descent method with a momentum term, wherein the hyperparameters include the penalty parameter and the tolerance error.

6. The method for controlling a smart sofa based on user habits according to claim 5, characterized in that: The S402A specifically includes: S402A1: Calculate the partial derivatives of the objective function value with respect to the penalty parameter and the tolerance error respectively: in, J represents the objective function value, Indicates the use for judgment The sign function of the difference direction, represents the indicator function, when When the indicator function takes 1, otherwise, the indicator function takes 0; S402A2: The penalty parameter and the tolerance error are updated according to the obtained partial derivatives until the update amplitude of each hyperparameter is less than the preset update amplitude: in, represents the learning rate, Indicates assignment, and Both represent the momentum term of the stored gradient, represents the momentum parameter.

7. The method for controlling a smart sofa based on user habits according to claim 1, characterized in that: After S7, the method further includes: The support vector regression model is retrained at preset time intervals.

8. A smart sofa, characterized in that: Applicable to the smart sofa control method based on user habits as described in any one of claims 1 to 7.

Images