Integrated control method and device for positive and negative rotation and multistage speed regulation of motor

Abstract

The invention belongs to the technical field of motor control methods, and particularly relates to an integrated control method and device for forward and reverse rotation and multistage speed regulation of a motor of handheld equipment, and the method comprises the following steps: collecting dynamic operation data of the handheld equipment in real time, the dynamic operation data comprising time sequence motion characteristics and space pressure distribution characteristics; the dynamic operation data are input into a pre-trained AI control model, a fusion control instruction is generated, and the fusion control instruction comprises a motor steering mark and a discretization speed grade; switching forward and reverse rotation driving modes of an H-bridge circuit according to the motor steering mark, and converting the discretization speed level into a corresponding PWM modulation signal; and the driving motor executes a composite action fusing the steering state and the speed regulation parameters. A fusion control instruction is intelligently generated through an AI control model to control forward and reverse rotation and multi-stage speed regulation of the motor, time sequence deviation can be eliminated, and control real-time performance is remarkably improved. In addition, the purposes of reducing power consumption and prolonging the endurance of the equipment can be achieved.

Description

Technical Field

[0001] This invention belongs to the technical field of motor control methods, and particularly relates to an integrated control method and device for forward and reverse rotation and multi-level speed regulation of a motor in a handheld device. Background Technology

[0002] Currently, motor control in handheld devices generally adopts a discrete architecture: forward and reverse switching relies on relay signal control via an H-bridge circuit, while multi-level speed regulation is achieved through duty cycle adjustment using an independent PWM generator. This discrete control mode has significant drawbacks: on the one hand, because steering and speed control commands need to be processed in a time-sharing manner by the microcontroller, a coordination delay of 10-35ms occurs, causing sudden torque changes and mechanical vibrations during motor start-up and shutdown; on the other hand, especially for miniaturized handheld devices, discrete control modules also increase circuit redundancy and power consumption by about 40%, severely limiting the device's battery life. Summary of the Invention

[0003] Based on this, the purpose of the present invention is to provide an integrated control method and device for forward and reverse rotation and multi-level speed regulation of motors in handheld devices, so as to fundamentally solve the problems of coordination delay and circuit complexity of existing discrete architecture drive motors.

[0004] The present invention is implemented as follows: It provides an integrated control method for forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, comprising the following steps:

[0005] Real-time acquisition of dynamic operation data of the handheld device, including temporal motion characteristics and spatial pressure distribution characteristics;

[0006] The dynamic operation data is input into a pre-trained AI control model to generate fused control commands, which include motor steering indicators and discretized speed levels.

[0007] The H-bridge circuit switches between forward and reverse drive modes based on the motor direction indicator, and converts the discretized speed level into a corresponding PWM modulation signal; and

[0008] The drive motor performs a combined action that integrates steering state and speed regulation parameters.

[0009] In some implementations, the AI ​​control model is constructed as follows:

[0010] The temporal motion features are processed using a temporal feature extraction unit;

[0011] The spatial pressure distribution features are processed using a spatial feature extraction unit;

[0012] The feature fusion layer outputs a control vector including a steering probability distribution and a speed level probability distribution, wherein the steering probability distribution corresponds to the electrode steering flag and the speed level probability distribution corresponds to the discretized speed level.

[0013] In some embodiments, the temporal feature extraction unit is a long short-term memory network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

[0014] In some implementations, switching the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator includes the following steps:

[0015] When the motor direction indicator is in the positive state, the first group of power switches in the H-bridge circuit is turned on.

[0016] When the motor direction indicator is in the reverse state, the second group of power switches in the H-bridge circuit is turned on.

[0017] The process of converting the discretized speed level into a corresponding PWM modulation signal includes the following steps:

[0018] Based on the speed level value, the PWM duty cycle is matched in a preset mapping table to generate a time-adjustable PWM modulation signal.

[0019] In some implementations, the training method for the AI ​​control model includes:

[0020] During the operation of the handheld device, multimodal training data is collected synchronously, including sensor data, actual steering state, and actual rotation speed value.

[0021] A supervised learning framework is constructed, with sensor data as input and steering state classification and velocity level regression as joint training objectives;

[0022] The weight parameters of the feature extraction network and the fusion layer are optimized through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked.

[0023] Generate a first AI control model, which includes optimized weight parameters; and

[0024] The first AI control model is compressed by knowledge distillation to generate a second AI control model, which is then used as the pre-trained AI control model.

[0025] This invention also provides an integrated control device for the forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, comprising:

[0026] The data acquisition unit is used to acquire dynamic operation data of the handheld device in real time, including temporal motion characteristics and spatial pressure distribution characteristics.

[0027] The instruction generation unit is used to input the dynamic operation data into a pre-trained AI control model to generate fused control instructions, which include motor steering indicators and discretized speed levels.

[0028] A signal conditioning unit is used to switch the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator, and to convert the discrete speed level into a corresponding PWM modulation signal; and

[0029] The drive unit is used to drive the motor to perform a combined action that integrates steering state and speed regulation parameters.

[0030] In some embodiments, the instruction generation unit includes:

[0031] The first feature processing module is used to process the temporal motion features using the temporal feature extraction unit;

[0032] The second feature processing module is used to process the spatial pressure distribution features using the spatial feature extraction unit;

[0033] The fusion output module is used to output a control vector including a steering probability distribution and a speed level probability distribution through a feature fusion layer. The steering probability distribution corresponds to the electrode steering flag, and the speed level probability distribution corresponds to the discretized speed level.

[0034] In some embodiments, the temporal feature extraction unit is a long short-term memory network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

[0035] In some embodiments, the signal conditioning unit includes:

[0036] A forward conduction module is used to turn on the first group of power switches in the H-bridge circuit when the motor direction indicator is in a positive state.

[0037] A reverse conduction module is used to turn on the second set of power switches in the H-bridge circuit when the motor direction indicator is in the reverse state.

[0038] The signal modulation module is used to match the PWM duty cycle in a preset mapping table according to the speed level value and generate a time-adjustable PWM modulation signal.

[0039] In some embodiments, the integrated control device further includes a model training unit, the model training unit comprising:

[0040] The data acquisition module is used to synchronously acquire multimodal training data during the operation of the handheld device. The multimodal training data includes sensor data, actual steering state, and actual rotation speed value.

[0041] The framework building module is used to build a supervised learning framework, which takes sensor data as input and turns state classification and velocity level regression as joint training objectives.

[0042] The iterative training module is used to optimize the weight parameters of the feature extraction network and the fusion layer through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked.

[0043] The model generation module is used to generate a first AI control model, which includes optimized weight parameters; and

[0044] The model compression module is used to compress the first AI control model through knowledge distillation to generate a second AI control model, and uses the second AI control model as the pre-trained AI control model.

[0045] The integrated control method and apparatus for forward and reverse rotation and multi-level speed regulation of a handheld device provided in this invention collects dynamic operation data of the handheld device in real time, inputs the dynamic operation data into a pre-trained AI control model, generates fused control commands, switches the forward and reverse drive mode of the H-bridge circuit according to the motor direction indicator, and converts the discretized speed levels into corresponding PWM modulation signals, ultimately driving the motor to execute a composite action of fused direction state and speed regulation parameters. This integrated control method intelligently generates fused control commands through the AI ​​control model to control the forward and reverse rotation and multi-level speed regulation of the motor, eliminating the timing deviation of direction switching and speed regulation control in traditional methods, reducing the response latency to within 8 milliseconds, and significantly improving control real-time performance. In addition, through integrated control design, the circuit complexity can be greatly reduced, thereby achieving the purpose of reducing power consumption and improving device battery life. Attached Figure Description

[0046] Figure 1 This is a flowchart of the integrated control method provided in the embodiments of the present invention;

[0047] Figure 2 This is a schematic diagram of a meditation aid provided in an embodiment of the present invention;

[0048] Figure 3 This is a circuit diagram of the H-bridge circuit in the forward rotation state provided by the embodiment of the present invention;

[0049] Figure 4 This is a circuit diagram of the H-bridge circuit in the reverse state provided in the embodiment of the present invention;

[0050] Figure 5 This is a structural block diagram of the integrated control device provided in the embodiments of the present invention. Detailed Implementation

[0051] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0052] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0054] refer to Figure 1 This application provides an integrated control method for forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, comprising the following steps:

[0055] S100. Real-time acquisition of dynamic operation data of the handheld device, the dynamic operation data including temporal motion characteristics and spatial pressure distribution characteristics;

[0056] S200. Input the dynamic operation data into the pre-trained AI control model to generate a fusion control command, which includes a motor steering indicator and a discretized speed level.

[0057] S300, Switch the forward / reverse drive mode of the H-bridge circuit according to the motor direction indicator, and convert the discretized speed level into a corresponding PWM modulation signal; and

[0058] S400, the drive motor performs a combined action that integrates steering state and speed regulation parameters.

[0059] First, the inertial measurement unit (IMU) collects real-time triaxial acceleration and angular velocity data from the handheld device, forming temporal motion characteristics. Simultaneously, a pressure sensor array collects pressure distribution data, forming spatial pressure distribution characteristics. These temporal motion and spatial pressure distribution characteristics are then combined to form dynamic operation data. Second, this dynamic operation data is input into a pre-trained AI control model. This model outputs a fused control command containing a motor direction indicator and discretized speed levels. The motor direction indicator shows the motor's forward or reverse rotation, while the discretized speed levels indicate the corresponding speed levels, which can be divided into 16 speed settings from 0 to 15. Next, the operating mode of the H-bridge drive circuit is switched according to the motor direction indicator: the forward path is activated when the indicator is positive, and the reverse path is activated when it is negative. Simultaneously, the speed level value is converted into a PWM modulation signal with the corresponding duty cycle based on the discretized speed levels. Finally, the drive motor executes a composite action combining the fused steering state and speed parameters, where the steering state corresponds to the forward / reverse drive mode, and the speed parameters correspond to the PWM modulation signal.

[0060] In some implementations, reference Figure 2 The handheld device is a meditation aid 1000. The user holds the meditation aid 1000, and the device assists the user's meditation through motor vibration. The aforementioned integrated control method is applied to the meditation aid 1000, ultimately driving the motor to perform a composite action that integrates directional state and speed adjustment parameters, achieving integrated control of the motor's forward and reverse rotation and multi-level speed adjustment. Specifically, during meditation training, the meditation aid 1000 is placed in the palm of the hand. The integrated control method controls the motor to perform a composite action that integrates directional state and speed adjustment parameters, thereby improving the user's concentration and preventing the user from becoming distracted or even falling asleep during meditation. This solves the problem that the uniform rotation of traditional motors is difficult to stimulate and remind the user after prolonged use.

[0061] In this application, dynamic operation data from a handheld device is collected in real time and input into a pre-trained AI control model to generate fused control commands. The forward and reverse drive modes of the H-bridge circuit are switched according to the motor's direction indication, and the discretized speed levels are converted into corresponding PWM modulation signals. Ultimately, the motor is driven to execute a composite action combining the fused steering state and speed control parameters. This integrated control method intelligently generates fused control commands through the AI ​​control model to control the motor's forward and reverse rotation and multi-level speed regulation, eliminating the timing deviations in steering switching and speed control in traditional methods. It can reduce response latency to within 8 milliseconds, significantly improving control real-time performance. Furthermore, through integrated control design, circuit complexity can be greatly reduced, thereby reducing power consumption and improving device battery life.

[0062] In some specific embodiments of this application, the AI ​​control model is constructed as follows:

[0063] The temporal motion features are processed using a temporal feature extraction unit;

[0064] The spatial pressure distribution features are processed using a spatial feature extraction unit;

[0065] The feature fusion layer outputs a control vector including a steering probability distribution and a speed level probability distribution, wherein the steering probability distribution corresponds to the electrode steering flag and the speed level probability distribution corresponds to the discretized speed level.

[0066] After real-time acquisition of dynamic operation data from the handheld device, the dynamic operation data is input into a pre-trained AI control model. The temporal feature extraction unit uses a long short-term memory network to process the continuous motion data input from the inertial measurement unit, which is also the temporal motion feature, thereby capturing the time dependence of the device's motion. The spatial feature extraction unit uses a convolutional neural network to process the two-dimensional pressure distribution data from the pressure sensor array, which is the spatial pressure distribution feature, thereby extracting the spatial topology of the user's handheld posture. The feature fusion layer concatenates the above temporal motion features and spatial pressure distribution feature vectors, and then generates a control vector through a fully connected neural network. This control vector includes a steering probability distribution and a speed level probability distribution, where the maximum probability value of the steering probability distribution corresponds to the motor steering indicator, and the maximum probability value index of the speed level probability distribution corresponds to the discretized speed level.

[0067] The system processes temporal motion features through a temporal feature extraction unit and spatial pressure distribution features through a spatial feature extraction unit. It employs a dual-branch feature processing structure to accurately analyze feature data. Finally, the system fuses and outputs a control vector through a feature fusion layer. The control output has high accuracy in steering probability distribution and low error in speed level probability distribution.

[0068] In some specific embodiments of this application, the temporal feature extraction unit is an LSTM network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

[0069] The temporal feature extraction unit specifically employs a two-layer LSTM network (Long Short-Term Memory network) with 128 neurons each to process triaxial acceleration and angular velocity data from the inertial measurement unit. Specifically, the LSTM network consists of two layers, each with 128 neurons. The first layer processes the original motion sequence and outputs the hidden state to the second layer, which outputs the final feature vector. Dropout regularization is used between layers to prevent overfitting. Using an LSTM network to process triaxial acceleration and angular velocity data significantly reduces processing errors.

[0070] The spatial feature extraction unit employs a convolutional neural network with 16 3x3 convolutional kernels to process the topological pressure data of the pressure sensor array. Specifically, the convolutional layers use 16 3x3 convolutional kernels with ReLU activation, and the pooling layers use 2x2 max pooling to reduce the dimensionality of the feature maps. The final flattening layer outputs a 64-dimensional spatial feature vector. Using a convolutional neural network to process topological pressure data significantly improves processing efficiency.

[0071] In some specific embodiments of this application, switching the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator includes the following steps:

[0072] When the motor direction indicator is in the positive state, the first group of power switches in the H-bridge circuit is turned on.

[0073] When the motor direction indicator is in the reverse state, the second group of power switches in the H-bridge circuit is turned on.

[0074] In some specific embodiments of this application, converting the discretized speed level into a corresponding PWM modulation signal includes the following steps:

[0075] Based on the speed level value, the PWM duty cycle is matched in a preset mapping table to generate a time-adjustable PWM modulation signal.

[0076] refer to Figure 3 and Figure 4 The H-bridge circuit contains four MOSFET power transistors, Q1, Q2, Q3, and Q4. For example, the model number of these MOSFETs could be IRF3205, where M represents the motor. (Reference) Figure 3 When the motor direction indicator is in the forward state (where forward and...) Figure 3When the forward rotation (meaning is the same as in the previous example) is in the following order: the first MOSFET Q1 and the fourth MOSFET Q4 are turned on, and the second MOSFET Q2 and the third MOSFET Q3 are turned off. The current flow is as shown by the arrows. The current flows from the positive terminal of the power supply through the first MOSFET Q1 from left to right through the motor, and then through the fourth MOSFET Q4 back to the negative terminal of the power supply, forming a forward current path. (Reference) Figure 4 When the flag is in the reverse state (where reverse and ... Figure 4 When the reverse meaning is the same as in the previous example: the second MOSFET Q2 and the third MOSFET Q3 are turned on, and the first MOSFET Q1 and the fourth MOSFET Q4 are turned off. The current flow is as shown by the arrow. The current flows from right to left through the motor, forming a reverse current path.

[0077] The discretized speed levels are converted into PWM duty cycles through a preset mapping table. Level 0 corresponds to a 6.25% duty cycle, Level 1 to a 12.5% ​​duty cycle, Level 2 to an 18.75% duty cycle, Level 3 to a 25% duty cycle, Level 4 to a 31.25% duty cycle, Level 5 to a 37.5% duty cycle, Level 6 to a 43.75% duty cycle, Level 7 to a 50% duty cycle, Level 8 to a 56.25% duty cycle, Level 9 to a 62.5% duty cycle, Level 10 to a 68.75% duty cycle, Level 11 to a 75% duty cycle, Level 12 to a 81.25% duty cycle, Level 13 to a 87.5% duty cycle, Level 14 to a 93.75% duty cycle, and Level 15 to a 100% duty cycle. The PWM carrier frequency is fixed at 20kHz.

[0078] In some specific embodiments of this application, the training method of the AI ​​control model includes:

[0079] During the operation of the handheld device, multimodal training data is collected synchronously, including sensor data, actual steering state, and actual rotation speed value.

[0080] A supervised learning framework is constructed, with sensor data as input and steering state classification and velocity level regression as joint training objectives;

[0081] The weight parameters of the feature extraction network and the fusion layer are optimized through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked.

[0082] Generate a first AI control model, which includes optimized weight parameters; and

[0083] The first AI control model is compressed by knowledge distillation to generate a second AI control model, which is then used as the pre-trained AI control model.

[0084] During handheld device operation, sensor data, actual steering status, and actual rotational speed are simultaneously collected. A supervised learning framework is constructed, using sensor data as input and steering status classification and speed level regression as joint training objectives. Network weight parameters are optimized through end-to-end training until the overall accuracy of the joint training objectives consistently reaches a preset threshold, for example, 99%. The corresponding weight parameters are then locked, generating a first AI control model. Subsequently, this first AI control model is compressed using knowledge distillation. Specifically, the first AI control model serves as the teacher model, and a student model with 70% fewer neurons is constructed. Knowledge transfer is achieved through KL divergence minimization, ultimately generating a deployable second AI control model. This second AI control model serves as a pre-trained AI control model for use in integrated control methods. Using knowledge distillation for compression significantly reduces model size while maintaining control accuracy, thereby reducing the resource consumption of the handheld device.

[0085] refer to Figure 5 This invention provides an integrated control device for the forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, comprising:

[0086] The data acquisition unit 100 is used to acquire dynamic operation data of the handheld device in real time, the dynamic operation data including temporal motion characteristics and spatial pressure distribution characteristics;

[0087] The instruction generation unit 200 is used to input the dynamic operation data into a pre-trained AI control model to generate fused control instructions, which include motor steering indicators and discretized speed levels.

[0088] The signal conditioning unit 300 is used to switch the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator, and to convert the discrete speed level into a corresponding PWM modulation signal; and

[0089] The drive unit 400 is used to drive the motor to perform a combined action that integrates steering state and speed regulation parameters.

[0090] First, the inertial measurement unit (IMU) collects real-time triaxial acceleration and angular velocity data from the handheld device, forming temporal motion characteristics. Simultaneously, a pressure sensor array collects pressure distribution data, forming spatial pressure distribution characteristics. These temporal motion and spatial pressure distribution characteristics are then combined to form dynamic operation data. Second, this dynamic operation data is input into a pre-trained AI control model. This model outputs a fused control command containing a motor direction indicator and discretized speed levels. The motor direction indicator shows the motor's forward or reverse rotation, while the discretized speed levels indicate the corresponding speed levels, which can be divided into 16 speed settings from 0 to 15. Next, the operating mode of the H-bridge drive circuit is switched according to the motor direction indicator: the forward path is activated when the indicator is positive, and the reverse path is activated when it is negative. Simultaneously, the speed level value is converted into a PWM modulation signal with the corresponding duty cycle based on the discretized speed levels. Finally, the drive motor executes a composite action combining the fused steering state and speed parameters, where the steering state corresponds to the forward / reverse drive mode, and the speed parameters correspond to the PWM modulation signal.

[0091] In some implementations, reference Figure 2 The handheld device is a meditation aid 1000. The user holds the meditation aid 1000, and the device assists the user's meditation through motor vibration. The aforementioned integrated control device is applied to the meditation aid 1000, ultimately driving the motor to perform a composite action that integrates directional control and speed adjustment parameters, achieving integrated control of the motor's forward and reverse rotation and multi-level speed regulation. Specifically, during meditation training, the meditation aid 1000 is placed in the palm of the hand. The integrated control device controls the motor to perform a composite action that integrates directional control and speed adjustment parameters, thereby improving the user's concentration and preventing the user from becoming distracted or even falling asleep during meditation. This solves the problem that the uniform rotation of traditional motors is difficult to stimulate and remind the user after prolonged use.

[0092] In this application, dynamic operation data from a handheld device is collected in real time and input into a pre-trained AI control model to generate fused control commands. The forward and reverse drive modes of the H-bridge circuit are switched according to the motor's direction indication, and the discretized speed levels are converted into corresponding PWM modulation signals. Ultimately, the motor is driven to execute a composite action combining the fused steering state and speed control parameters. This integrated control method intelligently generates fused control commands through the AI ​​control model to control the motor's forward and reverse rotation and multi-level speed regulation, eliminating the timing deviations in steering switching and speed control in traditional methods. It can reduce response latency to within 8 milliseconds, significantly improving control real-time performance. Furthermore, through integrated control design, circuit complexity can be greatly reduced, thereby reducing power consumption and improving device battery life.

[0093] In some specific embodiments of this application, the instruction generation unit includes:

[0094] The first feature processing module is used to process the temporal motion features using the temporal feature extraction unit;

[0095] The second feature processing module is used to process the spatial pressure distribution features using the spatial feature extraction unit;

[0096] The fusion output module is used to output a control vector including a steering probability distribution and a speed level probability distribution through a feature fusion layer. The steering probability distribution corresponds to the electrode steering flag, and the speed level probability distribution corresponds to the discretized speed level.

[0097] After real-time acquisition of dynamic operation data from the handheld device, the dynamic operation data is input into a pre-trained AI control model. The temporal feature extraction unit uses a long short-term memory network to process the continuous motion data input from the inertial measurement unit, which is also the temporal motion feature, thereby capturing the time dependence of the device's motion. The spatial feature extraction unit uses a convolutional neural network to process the two-dimensional pressure distribution data from the pressure sensor array, which is the spatial pressure distribution feature, thereby extracting the spatial topology of the user's handheld posture. The feature fusion layer concatenates the above temporal motion features and spatial pressure distribution feature vectors, and then generates a control vector through a fully connected neural network. This control vector includes a steering probability distribution and a speed level probability distribution, where the maximum probability value of the steering probability distribution corresponds to the motor steering indicator, and the maximum probability value index of the speed level probability distribution corresponds to the discretized speed level.

[0098] The system processes temporal motion features through a temporal feature extraction unit and spatial pressure distribution features through a spatial feature extraction unit. It employs a dual-branch feature processing structure to accurately analyze feature data. Finally, the system fuses and outputs a control vector through a feature fusion layer. The control output has high accuracy in steering probability distribution and low error in speed level probability distribution.

[0099] In some specific embodiments of this application, the temporal feature extraction unit is a long short-term memory network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

[0100] The temporal feature extraction unit specifically employs a two-layer LSTM network (Long Short-Term Memory network) with 128 neurons each to process triaxial acceleration and angular velocity data from the inertial measurement unit. Specifically, the LSTM network consists of two layers, each with 128 neurons. The first layer processes the original motion sequence and outputs the hidden state to the second layer, which outputs the final feature vector. Dropout regularization is used between layers to prevent overfitting. Using an LSTM network to process triaxial acceleration and angular velocity data significantly reduces processing errors.

[0101] The spatial feature extraction unit employs a convolutional neural network with 16 3x3 convolutional kernels to process the topological pressure data of the pressure sensor array. Specifically, the convolutional layers use 16 3x3 convolutional kernels with ReLU activation, and the pooling layers use 2x2 max pooling to reduce the dimensionality of the feature maps. The final flattening layer outputs a 64-dimensional spatial feature vector. Using a convolutional neural network to process topological pressure data significantly improves processing efficiency.

[0102] In some specific embodiments of this application, the signal conditioning unit includes:

[0103] A forward conduction module is used to turn on the first group of power switches in the H-bridge circuit when the motor direction indicator is in a positive state.

[0104] A reverse conduction module is used to turn on the second set of power switches in the H-bridge circuit when the motor direction indicator is in the reverse state.

[0105] The signal modulation module is used to match the PWM duty cycle in a preset mapping table according to the speed level value and generate a time-adjustable PWM modulation signal.

[0106] refer to Figure 3 and Figure 4 The H-bridge circuit contains four MOSFET power transistors, Q1, Q2, Q3, and Q4. For example, the model number of these MOSFETs could be IRF3205, where M represents the motor. (Reference) Figure 3 When the motor direction indicator is in the forward state (where forward and...) Figure 3 When the forward rotation (meaning is the same as in the previous example) is in the following order: the first MOSFET Q1 and the fourth MOSFET Q4 are turned on, and the second MOSFET Q2 and the third MOSFET Q3 are turned off. The current flow is as shown by the arrows. The current flows from the positive terminal of the power supply through the first MOSFET Q1 from left to right through the motor, and then through the fourth MOSFET Q4 back to the negative terminal of the power supply, forming a forward current path. (Reference) Figure 4 When the flag is in the reverse state (where reverse and ... Figure 4When the reverse meaning is the same as in the previous example: the second MOSFET Q2 and the third MOSFET Q3 are turned on, and the first MOSFET Q1 and the fourth MOSFET Q4 are turned off. The current flow is as shown by the arrow. The current flows from right to left through the motor, forming a reverse current path.

[0107] The discretized speed levels are converted into PWM duty cycles through a preset mapping table. Level 0 corresponds to a 6.25% duty cycle, Level 1 to a 12.5% ​​duty cycle, Level 2 to an 18.75% duty cycle, Level 3 to a 25% duty cycle, Level 4 to a 31.25% duty cycle, Level 5 to a 37.5% duty cycle, Level 6 to a 43.75% duty cycle, Level 7 to a 50% duty cycle, Level 8 to a 56.25% duty cycle, Level 9 to a 62.5% duty cycle, Level 10 to a 68.75% duty cycle, Level 11 to a 75% duty cycle, Level 12 to a 81.25% duty cycle, Level 13 to a 87.5% duty cycle, Level 14 to a 93.75% duty cycle, and Level 15 to a 100% duty cycle. The PWM carrier frequency is fixed at 20kHz.

[0108] In some specific embodiments of this application, the integrated control device further includes a model training unit, which comprises:

[0109] The data acquisition module is used to synchronously acquire multimodal training data during the operation of the handheld device. The multimodal training data includes sensor data, actual steering state, and actual rotation speed value.

[0110] The framework building module is used to build a supervised learning framework, which takes sensor data as input and turns state classification and velocity level regression as joint training objectives.

[0111] The iterative training module is used to optimize the weight parameters of the feature extraction network and the fusion layer through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked.

[0112] The model generation module is used to generate a first AI control model, which includes optimized weight parameters; and

[0113] The model compression module is used to compress the first AI control model through knowledge distillation to generate a second AI control model, and uses the second AI control model as the pre-trained AI control model.

[0114] During handheld device operation, sensor data, actual steering status, and actual rotational speed are simultaneously collected. A supervised learning framework is constructed, using sensor data as input and steering status classification and speed level regression as joint training objectives. Network weight parameters are optimized through end-to-end training until the overall accuracy of the joint training objectives consistently reaches a preset threshold, for example, 99%. The corresponding weight parameters are then locked, generating a first AI control model. Subsequently, this first AI control model is compressed using knowledge distillation. Specifically, the first AI control model serves as the teacher model, and a student model with 70% fewer neurons is constructed. Knowledge transfer is achieved through KL divergence minimization, ultimately generating a deployable second AI control model. This second AI control model serves as a pre-trained AI control model for retrieval and use in the integrated control device. Using knowledge distillation for compression significantly reduces the model size while maintaining control accuracy, thereby reducing the resource consumption of the handheld device.

[0115] The present invention provides a computer device, the computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the integrated control method described above.

[0116] The present invention also provides a storage device storing a computer program (instructions) thereon, which, when executed by a processor, implements the steps of the integrated control method described above.

[0117] For example, a computer program can be divided into one or more modules, one or more of which are stored in memory and executed by a processor to perform the present invention. One or more modules can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a computer device. For example, the computer program can be divided into the steps of the integrated control method provided in the above-described method embodiments.

[0118] Those skilled in the art will understand that the above description of the computer device is merely an example and does not constitute a limitation on the computer device. It may include more or fewer components than described above, or combine certain components, or different components, such as input / output devices, network access devices, buses, etc.

[0119] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting various parts of the computer device via various interfaces and lines.

[0120] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as interface display function, interface interaction function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as map interface, selection interface, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0121] If the modules / units integrated into the computer device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, an electrical signal, and a software distribution medium, etc.

[0122] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0123] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. An integrated control method for forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, characterized in that, Includes the following steps: Real-time acquisition of dynamic operation data of the handheld device, including temporal motion characteristics and spatial pressure distribution characteristics; The dynamic operation data is input into a pre-trained AI control model to generate fused control commands, which include motor steering indicators and discretized speed levels. The forward and reverse drive modes of the H-bridge circuit are switched according to the motor direction indicator, and the discrete speed level is converted into the corresponding PWM modulation signal. as well as The drive motor performs a combined action that integrates steering state and speed regulation parameters.

2. The integrated control method according to claim 1, characterized in that, The AI ​​control model is constructed as follows: The temporal motion features are processed using a temporal feature extraction unit; The spatial pressure distribution features are processed using a spatial feature extraction unit; The feature fusion layer outputs a control vector including a steering probability distribution and a speed level probability distribution, wherein the steering probability distribution corresponds to the electrode steering flag and the speed level probability distribution corresponds to the discretized speed level.

3. The integrated control method according to claim 2, characterized in that, The temporal feature extraction unit is a long short-term memory network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

4. The integrated control method according to claim 1, characterized in that, The step of switching the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator includes the following steps: When the motor direction indicator is in the positive state, the first group of power switches in the H-bridge circuit is turned on. When the motor direction indicator is in the reverse state, the second group of power switches in the H-bridge circuit is turned on. The process of converting the discretized speed level into a corresponding PWM modulation signal includes the following steps: Based on the speed level value, the PWM duty cycle is matched in a preset mapping table to generate a time-adjustable PWM modulation signal.

5. The integrated control method according to any one of claims 1-4, characterized in that, The training method for the AI ​​control model includes: During the operation of the handheld device, multimodal training data is collected synchronously, including sensor data, actual steering state, and actual rotation speed value. A supervised learning framework is constructed, with sensor data as input and steering state classification and velocity level regression as joint training objectives; The weight parameters of the feature extraction network and the fusion layer are optimized through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked. Generate a first AI control model, which includes optimized weight parameters; and The first AI control model is compressed by knowledge distillation to generate a second AI control model, which is then used as the pre-trained AI control model.

6. An integrated control device for forward and reverse rotation and multi-level speed regulation of a motor in a handheld device, characterized in that, include: The data acquisition unit is used to acquire dynamic operation data of the handheld device in real time, including temporal motion characteristics and spatial pressure distribution characteristics. The instruction generation unit is used to input the dynamic operation data into a pre-trained AI control model to generate fused control instructions, which include motor steering indicators and discretized speed levels. The signal conditioning unit is used to switch the forward and reverse drive modes of the H-bridge circuit according to the motor direction indicator, and to convert the discrete speed level into the corresponding PWM modulation signal. as well as The drive unit is used to drive the motor to perform a combined action that integrates steering state and speed regulation parameters.

7. The integrated control device according to claim 6, characterized in that, The instruction generation unit includes: The first feature processing module is used to process the temporal motion features using the temporal feature extraction unit; The second feature processing module is used to process the spatial pressure distribution features using the spatial feature extraction unit; The fusion output module is used to output a control vector including a steering probability distribution and a speed level probability distribution through a feature fusion layer. The steering probability distribution corresponds to the electrode steering flag, and the speed level probability distribution corresponds to the discretized speed level.

8. The integrated control device according to claim 7, characterized in that, The temporal feature extraction unit is a long short-term memory network that processes triaxial acceleration and angular velocity data from the inertial measurement unit; the spatial feature extraction unit is a convolutional neural network that processes topological pressure data from the pressure sensor array.

9. The integrated control device according to claim 6, characterized in that, The signal conditioning unit includes: A forward conduction module is used to turn on the first group of power switches in the H-bridge circuit when the motor direction indicator is in a positive state. A reverse conduction module is used to turn on the second set of power switches in the H-bridge circuit when the motor direction indicator is in the reverse state. The signal modulation module is used to match the PWM duty cycle in a preset mapping table according to the speed level value and generate a time-adjustable PWM modulation signal.

10. The integrated control device according to any one of claims 6-9, characterized in that, The integrated control device further includes a model training unit, which includes: The data acquisition module is used to synchronously acquire multimodal training data during the operation of the handheld device. The multimodal training data includes sensor data, actual steering state, and actual rotation speed value. The framework building module is used to build a supervised learning framework, which takes sensor data as input and turns state classification and velocity level regression as joint training objectives. The iterative training module is used to optimize the weight parameters of the feature extraction network and the fusion layer through end-to-end training until the overall accuracy of the joint training objective continuously reaches a preset threshold, at which point the corresponding weight parameters are locked. The model generation module is used to generate a first AI control model, which includes optimized weight parameters; and The model compression module is used to compress the first AI control model through knowledge distillation to generate a second AI control model, and uses the second AI control model as the pre-trained AI control model.

Images