Brushless push rod self-learning control method and system

Abstract

The invention relates to the technical field of brushless push rod intelligent control, in particular to a brushless push rod self-learning control method and system.The method comprises the steps that by collecting user sign information and historical push rod information, collected data are divided and marked, and a self-adjusting data set is established; a self-learning control network is constructed to control the self-learning behavior of a brushless push rod, a self-learning calendar is established, time is attached to the self-learning control network, then a control decision is output and executed according to the self-learning control network and the self-learning calendar, and the brushless push rod is controlled based on the control decision. The limitation that the angle of the seat cannot be adjusted meticulously is effectively solved, quick response to user requirements is achieved, the adjustment precision of the brushless push rod is improved, and meanwhile comfortable experience is provided for a user.

Description

Technical Field

[0001] The present invention relates to the technical field of brushless push rod intelligent control, and in particular to a brushless push rod self-learning control method and system. Background Art

[0002] In the field of smart home, comfort adjustment technology, especially in sofa seat design, has become an important part of improving the quality of home life. With the advancement of technology and the increase in consumer demand, the development of smart sofa seats tends to be more personalized and automated adjustment solutions. These sofa seats usually use brushless motor technology to adjust the angle and hardness of the seat to adapt to the physical characteristics and preferences of different users. Brushless motors are widely used in smart home devices due to their high efficiency, durability and easy maintenance, especially in home environments that require silent operation.

[0003] Although the application of brushless motors in sofa seats provides basic adjustment functions, the existing adjustment systems mostly rely on manual settings by users and lack sufficient intelligence and adaptability. In the existing technology, the adjustment of sofa seats usually does not take into account the user's specific physical changes and long-term usage habits, which may lead to poor user experience and physical discomfort. In addition, the manual adjustment mechanism cannot fully realize the personalized usage experience, and users often need to repeatedly debug to find the best comfort setting.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art known to those skilled in the art. Summary of the invention

[0005] The present invention provides a brushless push rod self-learning control method and system, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: A brushless actuator self-learning control method, the method comprising: Collecting historical user vital sign information and historical putter adjustment information, and performing data annotation and division on the historical user vital sign information and historical putter adjustment information to construct a putter self-adjustment data set; Building a self-learning control network based on the push rod self-adjustment data set, wherein the self-learning control network is used to control the self-learning behavior of the brushless push rod; Establishing a self-learning calendar for the push rod self-adjusting data set, wherein the self-learning calendar attaches time to the self-learning control network; An execution control decision is output according to the self-learning control network and the self-learning calendar, and the brushless push rod is adaptively adjusted according to the execution control decision.

[0007] Further, a self-learning calendar is established for the push rod self-adjustment data set, including: Dividing the historical push rod control adjustment information according to calendar features to obtain a plurality of application time memory groups, wherein the calendar feature is the historical push rod control adjustment information under a calendar usage scenario; Acquire a target usage calendar cycle, and select the corresponding application time memory group according to the calendar characteristics of the target usage calendar cycle; The selected application time memory group is orderly reorganized using a calendar cycle according to the target to obtain a calendar self-learning control decision, and an execution control decision is obtained based on the calendar self-learning control decision.

[0008] Furthermore, the historical user vital sign information and the historical actuator control adjustment information are labeled and divided to construct an actuator self-adjustment data set, including: Performing data cleaning on the historical putter adjustment information, performing data mosaicking on the cleaned historical putter adjustment information according to the historical putter adjustment schemes corresponding to the different user vital signs information, marking control parameters based on the historical putter adjustment schemes under different users, and extracting mosaicking parameters; Based on the plurality of the mosaic parameters, integrating the mosaic parameters to obtain a self-adjusting data set; Data segmentation is performed on the self-adjusting data set, and the self-adjusting data set is segmented into a putting training set and a putting test set.

[0009] Furthermore, a self-learning control network is constructed based on the push rod self-adjustment data set, including: Setting the number of decision trees based on the self-adjusting data set, and evenly distributing the self-adjusting data set to a number of decision trees, and determining the number of features, the number of features being the number of the historical putter adjustment solutions distributed to each decision tree; Setting a target depth for the decision trees, wherein the target depth is the number of decisions made by each decision tree; The initial number of analysis samples is set based on the number of features and the depth, and decision simulation is performed on a plurality of the decision trees to obtain decision simulation results, and a judgment is made based on the decision simulation results, and the number of analysis samples is adjusted in combination with the historical putter adjustment information to determine the number of analysis samples; A decision tree is constructed based on the number of features, the target depth, and the number of analysis samples, and a plurality of the decision trees are integrated to construct a self-learning control network.

[0010] Further, outputting a control decision according to the self-learning control network and the self-learning calendar includes: Retrieving the putting training set, and evenly distributing the putting training set to a plurality of the decision trees based on the number of the decision trees; The decision trees perform deep learning based on the assigned training sets, and output a number of initial control decisions according to the comprehensive evaluation results of the self-learning control network by the self-learning calendar; A plurality of the initial control decisions are collected and aggregated, and a control decision is executed based on the Plandre law output.

[0011] Furthermore, target depths are set for the decision trees, including: Establishing a depth adjustment model based on the historical push rod control adjustment information and setting an initial depth; Randomly extract the historical putter adjustment information to form a deep training subset and a deep verification subset, use the deep training subset to train the deep adjustment model, and use the deep verification subset to perform a performance test on the deep adjustment model; Based on the initial depth, the depth is gradually increased, the performance of the depth adjustment model at each depth setting is recorded, and a target depth is established.

[0012] Further, the depth validation subset is used to perform a performance test on the depth adjustment model, including: S1: randomly dividing the putter test set into n subsets of equal size, and performing sequence identification on the n subsets; S2: selecting one of the subsets as a test subset according to the sequence identifier, combining the remaining n-1 subsets as a training subset, and recording the performance parameter of the current subset as the test subset; S3: Repeat step S2 until each of the subsets serves as the test subset once; S4: Obtain a comprehensive evaluation result of the self-learning control network by calculating each subset as the performance parameter of the test subset.

[0013] Further, adaptively adjusting the brushless push rod according to the control decision includes: Collecting the degree of freedom node information of the brushless push rod, and generating a plurality of degree of freedom spaces according to the degree of freedom interval corresponding to the degree of freedom information, wherein the plurality of degree of freedom spaces correspond to the degree of freedom node information one by one; According to the plurality of degree-of-freedom spaces and the brushless push rod control decision, setting the push rod control trajectory; Sending the push rod control trajectory to the self-learning control network, and determining the starting coordinates and the end coordinates of the brushless push rod; Analyze the push rod control trajectory, identify key points of the push rod control trajectory based on the analysis results, and output key nodes; Segmenting the push rod control trajectory based on the key nodes to obtain a front-end control trajectory and a rear-end control trajectory; Outputting the front-end trajectory control parameter and the back-end trajectory control parameter according to the starting coordinate and the end point coordinate; The brushless push rod control decision is adjusted based on the front-end trajectory control parameter and the rear-end trajectory control parameter.

[0014] Further, according to the degree of freedom space and the brushless push rod control decision, setting the push rod control trajectory includes: According to the plurality of spatial degrees of freedom, a spatial degrees of freedom model is established, and the spatial degrees of freedom model is converted into discrete point data required for the push rod control trajectory; Performing trajectory interpolation and connection on the discrete point data to generate a continuous control trajectory; A fitting algorithm is selected to smooth the continuous trajectory of the push rod, and the smoothed continuous control trajectory is converted into a control trajectory format that can be recognized by the self-learning control network.

[0015] A brushless push rod self-learning control system, the system comprising: A data set construction module collects historical user vital sign information and historical putter adjustment information, and performs data annotation and division on the historical user vital sign information and historical putter adjustment information to construct a putter self-adjustment data set; A self-learning network construction module, which constructs a self-learning control network based on the push rod self-adjustment data set, and the self-learning control network is used to control the self-learning behavior of the brushless push rod; A self-learning calendar setting module, which establishes a self-learning calendar for the push rod self-adjustment data set, and the self-learning calendar attaches time to the self-learning control network; A control decision execution module executes a control decision according to the self-learning control network and the self-learning calendar output, and adaptively adjusts the brushless push rod according to the executed control decision.

[0016] The technical solution of the present invention can achieve the following technical effects: It effectively solves the problem that the seat angle cannot be adjusted based on the specific needs of the user, provides automation and adaptive capabilities for the personalized control of the sofa seat, realizes efficient and precise automated operation, reduces manual intervention, and thus adapts to the user's personalized adjustment.

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a flow chart of the self-learning control method of the brushless actuator; Figure 2 Create a process diagram for the self-learning calendar; Figure 3 Schematic diagram of the process of constructing a self-regulating data set; Figure 4 This is the structure diagram of the self-learning control network. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0022] Embodiment 1 like Figure 1 As shown, the present application provides a brushless actuator self-learning control method, the method comprising: S10: collecting historical user vital sign information and historical putter adjustment information, and performing data labeling and division on the historical user vital sign information and the historical putter adjustment information to construct a putter self-adjustment data set; S20: constructing a self-learning control network based on the push rod self-adjustment data set, where the self-learning control network is used to control the self-learning behavior of the brushless push rod; S30: establishing a self-learning calendar for the push rod self-adjustment data set, and the self-learning calendar attaches time to the self-learning control network; S40: Executing a control decision according to the self-learning control network and the self-learning calendar output, and adaptively adjusting the brushless actuator according to the executed control decision.

[0023] The technical solution of the present invention effectively solves the problem that the seat angle cannot be adjusted based on the specific needs of the user, provides automation and self-adaptation capabilities for the personalized control of the sofa seat, realizes efficient and precise automated operation, reduces manual intervention, and adapts to the user's personalized adjustment.

[0024] Specifically, first, sensors and data recording systems can be deployed to collect the user's physical information such as weight, height, etc., and push rod adjustment parameters such as angle, force, etc. when using the brushless push rod to adjust the seat, and the collected information can be cleaned and pre-processed by data processing software to eliminate interference such as noise, and the processed data can be labeled to identify the push rod adjustment response related to the specific user's physical signs, which is convenient for subsequent data analysis and model training. The data can be divided based on the historical adjustment information to construct a structured push rod self-adjustment data set, which is divided into a push rod training set for model training and a push rod test set for verifying model performance. Then, a self-learning control network is established based on the self-adjustment data set, and the labeled and divided push rod adjustment training set is used to self-adjust the push rod. When training the learning control network, it should be noted that there may be overfitting problems in the training model process, that is, the model performs well on the training data, but performs poorly on unseen data or in actual applications. At the same time, the push rod test set is used to evaluate the self-learning control network to check its responsiveness and accuracy on unseen data and reduce the overfitting problem. Then, a self-learning calendar is established based on the push rod self-adjustment data set. The self-learning control network outputs appropriate control decisions based on the currently input user vital signs information and calendar information, and adjusts the angle and position of the sofa seat through the brushless push rod. The self-learning calendar is integrated with the self-learning control network so that the control network can automatically adjust its behavior according to the time information in the calendar to ensure that users get the best comfort that matches their physical signs and preferences.

[0025] The present invention effectively solves the problem that the seat angle cannot be adjusted based on the specific needs of the user. Through continuous optimization and time-sensitive adjustment, personalized and adaptive control is provided for the brushless push rod, efficient and precise automatic operation is achieved, and manual intervention is reduced, thereby meeting the personalized needs of users.

[0026] Further, if Figure 2 As shown, a self-learning calendar is established for the actuator self-adjustment data set, including: S31: dividing the historical push rod control adjustment information according to calendar characteristics to obtain a plurality of application time memory groups, where the calendar characteristic is the historical push rod control adjustment information in a calendar usage scenario; S32: obtaining a target usage calendar cycle, and selecting a corresponding application time memory group according to the calendar characteristics of the target usage calendar cycle; S33: orderly reorganizing the selected application time memory group according to the target usage calendar cycle, obtaining a calendar self-learning control decision, and obtaining an execution control decision according to the calendar self-learning control decision.

[0027] As a preferred embodiment of the above, first, historical push rod control adjustment information is retrieved, and this information is marked by date and time to create clear calendar features, such as weekday mornings or weekend afternoons. Then, the system groups the data into several application time memory groups based on these calendar features, each group containing all the push rod adjustment information within the corresponding time period, thereby helping the system understand user preferences in specific scenarios. Subsequently, the system determines the target usage calendar cycle, such as seasonal changes or holidays, and selects the group that best matches the target cycle characteristics from the established memory groups. Then, the target usage calendar cycle that needs to be referenced is determined based on the actual scenario, and the target usage calendar cycle is compared with the calendar characteristics of each time memory group. The best matching memory group is selected. These selected memory groups are then reorganized in order according to the target cycle and integrated into a coherent decision-making process, taking into account the consistency of time sequence and user behavior. In this way, the reorganized data is used to generate calendar self-learning control decisions. These decisions are optimized by analyzing user needs and actuator performance in different time periods. The brushless actuator automatically adjusts its action according to the control decision to meet the user's expectations and comfort, thereby ensuring that the system can be adaptively adjusted according to the user's specific needs in different time periods. It should be noted that the selected data set should be highly relevant to the current time and situation to ensure that the control decision meets the latest user needs and environmental changes.

[0028] Furthermore, if Figure 3 As shown, historical user vital sign information and historical actuator control adjustment information are labeled and divided to construct an actuator self-adjustment data set, including: S11: performing data cleaning on the historical putter adjustment information, performing data mosaicking on the cleaned historical putter adjustment information according to the historical putter adjustment schemes corresponding to the physical sign information of different users, marking the control parameters based on the historical putter adjustment schemes under different users, and extracting the mosaicking parameters; S12: Based on a number of mosaic parameters, integrate the mosaic parameters to obtain a self-adjusting data set; S13: performing data segmentation on the self-adjusting data set, and segmenting the self-adjusting data set into a putting training set and a putting test set.

[0029] In this embodiment, by performing a preliminary analysis on the historical putter adjustment information, identifying and finding abnormal data, and using techniques such as filtering or smoothing to reduce the impact of measurement errors and occasional noise to ensure the quality and accuracy of the data, the system groups the cleaned data and analyzes the historical putter adjustment schemes for each user group, marking and extracting key control parameters therefrom. These refined parameters are called mosaic parameters, which are then aggregated and integrated, and commonalities and differences are identified through statistical analysis to construct a self-adjusting data set. The data set is further precisely divided into a putter training set and a putter test set. It should be noted that both must represent the characteristics of the entire data set and do not overlap. The training set is used to build and optimize the self-learning control network, while the test set is used to independently verify the performance and reliability of the network to ensure the performance of the model on unknown data.

[0030] Further, if Figure 4 As shown, a self-learning control network is constructed based on the push rod self-adjustment data set, including: The number of decision trees is set based on the self-adjusting data set, and the self-adjusting data set is evenly distributed to a number of decision trees, and the number of features is determined, where the number of features is the number of historical putter adjustment schemes allocated to each decision tree; Set the target depth for several decision trees, where the target depth is the number of decisions made by each decision tree; The initial number of analysis samples is set based on the number and depth of features, and decision simulation is performed on several decision trees to obtain decision simulation results. Based on the decision simulation results, judgment is made, and the number of analysis samples is adjusted in combination with historical putter adjustment information to determine the number of analysis samples; A decision tree is constructed based on the number of features, target depth, and number of analysis samples, and several decision trees are integrated to construct a self-learning control network.

[0031] Specifically, the number of decision trees to be generated is first determined according to the size of the self-adjusting data set and the complexity of the features. The self-adjusting data set is analyzed and an appropriate number of features are assigned to each decision tree. These features reflect the user's vital signs information and various parameters of the putter adjustment, such as position, angle and speed. The number of assigned features usually depends on the information gain of each feature for the predicted target, ensuring that each tree can obtain a representative feature set. The next step is to set a target depth for each decision tree. This depth is adjusted according to the expected complexity and accuracy requirements of the model. What needs to be considered is that the depth setting should consider the risk of model overfitting. In order to optimize the model performance, it is necessary to first predict the result of each decision by simulating different decisions, and adjust the number of analysis samples required based on the simulation results and historical putter adjustment information. Each decision tree is constructed based on the above parameters and the adjusted number of samples, and these decision trees are integrated into a unified self-learning control network. This method not only improves the accuracy of the decision, but also enhances the stability of the model.

[0032] Further, the control decisions outputted by the self-learning control network and the self-learning calendar include: Retrieve the putting training set, and evenly distribute the putting training set to several decision trees based on the number of decision trees; Several decision trees are deep-learned based on the assigned training sets, and several initial control decisions are output according to the comprehensive evaluation results of the self-learning control network by the self-learning calendar; Several initial control decisions are collected and aggregated, and the control decision is executed based on the Plandre law output.

[0033] As a preferred embodiment of the above embodiment, a putter training set is retrieved from the overall self-adjusting data set, and the training set covers a wide range of user vital signs information and historical putter adjustment information of the putter. Based on the preset number of decision trees, the training set is evenly distributed to each decision tree to ensure that each tree can be exposed to representative data samples to comprehensively learn the operation mode under different situations to avoid a certain tree being biased towards a specific type of data. However, the model diversity can be enhanced through techniques such as self-service aggregation or random feature selection in random forests. Then, each tree is trained for deep learning, the decision rules are gradually optimized, and the parameters of the decision tree such as target depth, number of analysis samples, etc. are adjusted to optimize the prediction performance of each tree. During the training process, the parameters of the learning algorithm are adjusted according to the information of the self-learning calendar to ensure that the learning process is adapted to the time-related usage pattern. Each tree outputs an initial control decision based on its training data, and the initial control decisions of all decision trees are collected and aggregated through Pland's rule, that is, the majority voting mechanism. Pland's rule enhances the robustness of the decision and reduces the impact of possible prediction errors of a single tree on the overall system performance. The aggregated control decision is implemented in the actual brushless putter control system.

[0034] Furthermore, target depths are set for several decision trees, including: Establish a depth adjustment model based on historical push rod control adjustment information and set the initial depth; Randomly extract historical putter adjustment information to form a deep training subset and a deep verification subset, use the deep training subset to train the deep adjustment model, and use the deep verification subset to perform performance testing on the deep adjustment model; Based on the initial depth, the depth is gradually increased, the performance of the depth-adjusted model at each depth setting is recorded, and the target depth is established.

[0035] In this embodiment, first, a deep adjustment model is established based on historical putter adjustment information, and the initial depth is set. For a decision tree, this means the number of decision points of the tree. Then, data is extracted from the historical putter adjustment information, including all relevant input features and target outputs, and a part of the data is randomly selected as a deep training subset and the other part as a deep verification subset. Ensure that the two subsets are representative and diverse in terms of features and outputs. Use the deep training subset to train the deep conditioning model and adjust model parameters to optimize performance, such as learning rate, regularization parameters, etc. Use the deep validation subset to test the model to evaluate its performance on unseen data and record key performance indicators such as model accuracy. After training and validation, gradually increase the model depth based on the initial depth and repeat the training and validation process. At the same time, record and analyze the model performance under each depth setting. It is important to observe the changing trend of model performance with increasing depth, especially signs of overfitting or underfitting, compare the model performance at different depths, find the depth setting that meets the preset performance threshold, and establish the target depth. This depth is the appropriate point for the model to maintain high accuracy while avoiding overfitting and unnecessary consumption of computing resources. At the determined target depth, further adjust and optimize the model parameters.

[0036] Further, using the depth verification subset to perform performance testing on the depth adjustment model includes: S1: Randomly divide the push rod test set into n subsets of equal size, and perform sequence identification on the n subsets; S2: Select one of the subsets as the test subset according to the sequence identifier, and combine the remaining n-1 subsets as the training subset, and record the performance parameters of the current subset as the test subset; S3: Repeat step S2 until each subset serves as a test subset; S4: The comprehensive evaluation result of the self-learning control network is obtained by calculating the performance parameters of each subset as a test subset.

[0037] Specifically, the test set is first randomly divided into n subsets of equal size. This division needs to ensure the consistency of data distribution in each subset to avoid deviations caused by uneven data distribution, and each subset is sequenced. In each round of testing, a subset is selected as the test set according to the sequence identifier, and the other n-1 subsets are combined as the training set. The combined training set is used to train the deep conditioning model. According to the model type, appropriate training parameters are configured, such as learning rate, number of iterations, etc. The performance of the model is evaluated with the selected test subset, and important performance indicators are recorded. Then the single verification process is repeated, and a different subset is selected as the test set each time. , until each subset is used as a test set once. This cyclic testing method comprehensively evaluates the stability and reliability of the model on different data subsets. Then the performance parameters of all iterations are summarized, and statistical methods such as calculating the mean and standard deviation are used to obtain statistical data on the overall performance of the model. The performance indicators are analyzed to evaluate the stability and reliability of the model. This testing and evaluation process not only verifies the accuracy of the deep adjustment model, but also provides key performance feedback, which helps to further optimize and adjust the model. It should be noted that the same model configuration needs to be used in each iteration to ensure that only the training / testing data set changes.

[0038] Further, the brushless actuator is adaptively adjusted based on the control decision, including: Collecting the freedom node information of the brushless actuator, generating multiple freedom spaces according to the freedom interval corresponding to the freedom information, wherein the multiple freedom spaces correspond to the freedom node information one by one; Set the actuator control trajectory based on multiple degrees of freedom space and brushless actuator control decisions; Send the actuator control trajectory to the self-learning control network and determine the starting and ending coordinates of the brushless actuator; Analyze the control trajectory of the push rod, identify key points of the control trajectory of the push rod based on the analysis results, and output key nodes; Segment the push rod control trajectory based on key nodes to obtain the front-end control trajectory and the back-end control trajectory; According to the starting coordinates and the end coordinates, output the front-end trajectory control parameters and the back-end trajectory control parameters; The brushless actuator control decision is adjusted based on the front-end trajectory control parameters and the rear-end trajectory control parameters.

[0039] As a preferred embodiment of the above, the degree of freedom information of the brushless push rod is obtained through a sensor such as a position sensor or an angle sensor, including the movement ability and restriction conditions of the push rod at different positions. Based on the collected degree of freedom information, multiple degree of freedom spaces are generated, each space defines the operating range of the push rod under specific conditions, and the operation trajectory of the push rod is set through the defined degree of freedom space and the predetermined control decision, including: path selection, speed setting, etc. The set push rod control trajectory is sent to the self-learning control network, and the start and end coordinates of the push rod are specified. Then, the push rod control trajectory sent to the network is analyzed in detail to evaluate its efficiency and feasibility, and the key points of the trajectory, such as turning points, acceleration points, etc., are identified from the analysis results. Based on the key nodes, the push rod control trajectory is divided into a front-end trajectory and a rear-end trajectory. Such segmentation helps to control each stage more meticulously. For the segmented trajectory, the corresponding control parameters are output respectively. These parameters will directly affect the action details of the push rod, such as speed adjustment, force control, etc. Based on the control parameters of the front-end and rear-end trajectories, the overall brushless push rod control decision is adjusted to ensure the continuity and smoothness of the entire trajectory.

[0040] Furthermore, according to the degree of freedom space and the brushless actuator control decision, the actuator control trajectory is set, including: According to multiple spatial degrees of freedom, a spatial degrees of freedom model is established, and the spatial degrees of freedom model is converted into discrete point data required for the control trajectory of the push rod; Perform trajectory interpolation and connection on discrete point data to generate a continuous control trajectory; A fitting algorithm is selected to smooth the continuous trajectory of the push rod, and the smoothed continuous control trajectory is converted into a control trajectory format that can be recognized by the self-learning control network.

[0041] Specifically, first, a degree of freedom model is established based on these degree of freedom definitions. The model integrates the movement capabilities of the push rod in different degree of freedom ranges and converts them into discrete point data required for the push rod control trajectory. These discrete point data represent the specific position or state reached by the push rod. It is important to ensure that these discrete point data meet the input requirements of subsequent processing and control systems. These discrete points are connected through trajectory interpolation methods, such as linear interpolation, to ensure that a smooth and continuous control trajectory is generated. A suitable fitting algorithm, such as Gaussian filtering, is selected to eliminate any sharp turns or unnatural movements. This smoothing step is necessary because the original generated trajectory may contain sharp turns or unsmooth sections, which are not conducive to the actual movement of the push rod. The smoothed trajectory is converted into a format that can be recognized and processed by the self-learning control network. It is important to test the converted trajectory before actual application to verify whether its performance in the control system meets expectations.

[0042] Embodiment 2 Based on the same inventive concept as the brushless push rod self-learning control method in the aforementioned embodiment, the present invention also provides a brushless push rod self-learning control system, the system comprising: A data set construction module collects historical user vital signs information and historical putter adjustment information, and performs data annotation and division on the historical user vital signs information and historical putter adjustment information to construct a putter self-adjustment data set; A self-learning network building module, which builds a self-learning control network based on the actuator self-adjustment data set, and the self-learning control network is used to control the self-learning behavior of the brushless actuator; A self-learning calendar setting module is used to establish a self-learning calendar for the actuator self-adjustment data set, and the self-learning calendar is used to attach time to the self-learning control network; The control decision execution module executes the control decision according to the self-learning control network and the self-learning calendar output, and adaptively adjusts the brushless actuator according to the executed control decision.

[0043] The above-mentioned adjustment system in the present invention can effectively realize the brushless push rod self-learning control method, and the technical effects that can be achieved are as described in the above-mentioned embodiments, which will not be repeated here.

[0044] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined therein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A brushless actuator self-learning control method, characterized in that: The method comprises: Collecting historical user vital sign information and historical putter adjustment information, and performing data annotation and division on the historical user vital sign information and historical putter adjustment information to construct a putter self-adjustment data set; Building a self-learning control network based on the push rod self-adjustment data set, wherein the self-learning control network is used to control the self-learning behavior of the brushless push rod; Establishing a self-learning calendar for the push rod self-adjusting data set, wherein the self-learning calendar attaches time to the self-learning control network; An execution control decision is outputted according to the self-learning control network and the self-learning calendar, and the brushless push rod is adaptively adjusted according to the execution control decision.

2. The brushless actuator self-learning control method according to claim 1, characterized in that: Establishing a self-learning calendar for the actuator self-adjustment data set, including: Dividing the historical push rod control adjustment information according to calendar features to obtain a plurality of application time memory groups, wherein the calendar feature is the historical push rod control adjustment information under a calendar usage scenario; Acquire a target usage calendar cycle, and select the corresponding application time memory group according to the calendar characteristics of the target usage calendar cycle; The selected application time memory group is orderly reorganized using a calendar cycle according to the target to obtain a calendar self-learning control decision, and an execution control decision is obtained based on the calendar self-learning control decision.

3. The brushless actuator self-learning control method according to claim 1, characterized in that: The historical user vital sign information and the historical actuator control adjustment information are labeled and divided to construct an actuator self-adjustment data set, including: Performing data cleaning on the historical putter adjustment information, performing data mosaicking on the cleaned historical putter adjustment information according to the historical putter adjustment schemes corresponding to the different user vital signs information, marking control parameters based on the historical putter adjustment schemes under different users, and extracting mosaicking parameters; Based on the plurality of the mosaic parameters, integrating the mosaic parameters to obtain a self-adjusting data set; Data segmentation is performed on the self-adjusting data set, and the self-adjusting data set is segmented into a putting training set and a putting test set.

4. The brushless actuator self-learning control method according to claim 3, characterized in that: A self-learning control network is constructed based on the push rod self-adjustment data set, including: Setting the number of decision trees based on the self-adjusting data set, and evenly distributing the self-adjusting data set to a number of decision trees, and determining the number of features, the number of features being the number of the historical putter adjustment solutions distributed to each decision tree; Setting a target depth for the decision trees, wherein the target depth is the number of decisions made by each decision tree; The initial number of analysis samples is set based on the number of features and the depth, and decision simulation is performed on a plurality of the decision trees to obtain decision simulation results, and a judgment is made based on the decision simulation results, and the number of analysis samples is adjusted in combination with the historical putter adjustment information to determine the number of analysis samples; A decision tree is constructed based on the number of features, the target depth, and the number of analysis samples, and a plurality of the decision trees are integrated to construct a self-learning control network.

5. The brushless actuator self-learning control method according to claim 4, characterized in that: Outputting a control decision according to the self-learning control network and the self-learning calendar includes: Retrieving the putting training set, and evenly distributing the putting training set to a plurality of the decision trees based on the number of the decision trees; The decision trees perform deep learning based on the assigned training sets, and output a number of initial control decisions according to the comprehensive evaluation results of the self-learning control network by the self-learning calendar; A plurality of the initial control decisions are collected and aggregated, and a control decision is executed based on the Plandre law output.

6. The brushless actuator self-learning control method according to claim 4, characterized in that: Setting target depths for several of the decision trees includes: Establishing a depth adjustment model based on the historical push rod control adjustment information and setting an initial depth; Randomly extract the historical putter adjustment information to form a deep training subset and a deep verification subset, use the deep training subset to train the deep adjustment model, and use the deep verification subset to perform a performance test on the deep adjustment model; Based on the initial depth, the depth is gradually increased, the performance of the depth adjustment model at each depth setting is recorded, and a target depth is established.

7. The brushless actuator self-learning control method according to claim 6, characterized in that: The deep validation subset is used to perform a performance test on the deep adjustment model, including: S1: randomly dividing the putter test set into n subsets of equal size, and performing sequence identification on the n subsets; S2: selecting one of the subsets as a test subset according to the sequence identifier, combining the remaining n-1 subsets as a training subset, and recording the performance parameter of the current subset as the test subset; S3: Repeat step S2 until each of the subsets serves as the test subset once; S4: Obtain a comprehensive evaluation result of the self-learning control network by calculating each subset as the performance parameter of the test subset.

8. The brushless actuator self-learning control method according to claim 1, characterized in that: Adaptively adjusting the brushless push rod according to the control decision includes: Collecting the degree of freedom node information of the brushless push rod, and generating a plurality of degree of freedom spaces according to the degree of freedom interval corresponding to the degree of freedom information, wherein the plurality of degree of freedom spaces correspond to the degree of freedom node information one by one; According to the plurality of degree-of-freedom spaces and the brushless push rod control decision, setting the push rod control trajectory; Sending the push rod control trajectory to the self-learning control network, and determining the starting coordinates and the end coordinates of the brushless push rod; Analyze the push rod control trajectory, identify key points of the push rod control trajectory based on the analysis results, and output key nodes; Segmenting the push rod control trajectory based on the key nodes to obtain a front-end control trajectory and a rear-end control trajectory; Outputting the front-end trajectory control parameter and the back-end trajectory control parameter according to the starting coordinate and the end point coordinate; The brushless push rod control decision is adjusted based on the front-end trajectory control parameter and the rear-end trajectory control parameter.

9. The brushless actuator self-learning control method according to claim 8, characterized in that: According to the degree of freedom space and the brushless push rod control decision, the push rod control trajectory is set, including: According to the plurality of spatial degrees of freedom, a spatial degrees of freedom model is established, and the spatial degrees of freedom model is converted into discrete point data required for the push rod control trajectory; Performing trajectory interpolation and connection on the discrete point data to generate a continuous control trajectory; A fitting algorithm is selected to smooth the continuous trajectory of the push rod, and the smoothed continuous control trajectory is converted into a control trajectory format that can be recognized by the self-learning control network.

10. Brushless actuator self-learning control system, characterized in that: The system comprises: A data set construction module collects historical user vital sign information and historical putter adjustment information, and performs data annotation and division on the historical user vital sign information and historical putter adjustment information to construct a putter self-adjustment data set; A self-learning network construction module, which constructs a self-learning control network based on the push rod self-adjustment data set, and the self-learning control network is used to control the self-learning behavior of the brushless push rod; A self-learning calendar setting module, which establishes a self-learning calendar for the push rod self-adjustment data set, and the self-learning calendar attaches time to the self-learning control network; A control decision execution module executes a control decision according to the self-learning control network and the self-learning calendar output, and adaptively adjusts the brushless push rod according to the executed control decision.

Images