Motor torque dynamic adjustment method and system based on analog input

Abstract

The invention provides a motor torque dynamic adjustment method and system based on analog quantity input, and relates to the technical field of motor control, and the method comprises the steps: firstly obtaining an analog quantity input signal set received when a servo motor operates, the analog quantity input signal set comprises at least one signal representing the torque adjustment demand of external equipment; analyzing the signal set to obtain a torque adjustment parameter set containing a target torque value, a torque change rate threshold value and a torque response time window, and determining a current torque value and a torque dynamic response characteristic by combining current operation state parameters such as a motor winding current sampling value and a motor rotor rotating speed detection value; and then generating a torque dynamic adjustment strategy containing a sectional torque adjustment instruction sequence according to the parameters, and finally executing the torque dynamic adjustment strategy, so that the output torque of the servo motor is dynamically matched with a target value according to a preset window, the torque change rate threshold constraint is met, and accurate and flexible torque dynamic adjustment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and in particular to a method and system for dynamically adjusting motor torque based on analog quantity input. Background Art

[0002] In today's motor control technology, precise control of motor torque is crucial for the stable operation of many industrial production processes and complex electromechanical systems. With the continuous improvement of industrial automation and the widespread application of various complex equipment, higher and higher requirements are placed on the accuracy, flexibility and adaptability of motor torque control.

[0003] Traditional motor torque control methods mostly use relatively simple and fixed modes. Some early methods only control motor torque based on fixed parameter settings, and are unable to perceive changes in the external environment and the diverse demands of external devices for torque adjustment in real time. These methods often seem powerless when faced with frequent and complex torque adjustment demands of external devices, and it is difficult to respond in a timely and accurate manner, resulting in a large deviation between the motor output torque and actual demand, which in turn affects the operating efficiency and stability of the entire device. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a method for dynamically adjusting motor torque based on analog input, the method comprising:

[0005] Acquire a set of analog input signals received by the servo motor in a running state, wherein the set of analog input signals includes at least one analog input signal, and the analog input signal is used to characterize a torque adjustment requirement of an external device on the servo motor;

[0006] Performing signal analysis on the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal, wherein the torque adjustment parameter set includes a target torque value, a torque change rate threshold, and a torque response time window;

[0007] Determine the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor, wherein the current operating state parameters include a motor winding current sampling value and a motor rotor speed detection value;

[0008] Generate a torque dynamic adjustment strategy for the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, wherein the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence;

[0009] The segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed to make the output torque of the servo motor dynamically match the target torque value according to a preset time window while satisfying the torque change rate threshold constraint.

[0010] On the other hand, an embodiment of the present invention also provides a motor torque dynamic adjustment system based on analog input, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present application obtains a set including multiple analog input signals, and obtains a torque adjustment parameter set by deeply analyzing the analog input signal set, which not only includes the target torque value, but also introduces two key parameters, the torque change rate threshold and the torque response time window. In the torque determination link, based on the torque adjustment parameter set and the current operating state parameters of the servo motor (motor winding current sampling value and motor rotor speed detection value), the current torque value and torque dynamic response characteristics of the servo motor are determined, which fully considers the real-time operating state of the motor and can more accurately grasp the current working condition of the motor. In terms of strategy generation, according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, a torque dynamic adjustment strategy including a segmented torque adjustment instruction sequence is generated, which can flexibly adjust the torque according to the needs and motor states of different stages, and realize more refined and intelligent torque dynamic adjustment. This segmented strategy can dynamically match the target torque value for different time windows and motor operating states, and strictly meet the torque change rate threshold constraint, avoiding the impact of torque mutation on the motor and equipment, and significantly improving the stability and reliability of motor operation. Finally, in the actual execution process, the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed, so that the output torque of the servo motor can accurately and dynamically match the target torque value within the preset time window, while strictly meeting the torque change rate threshold constraint. This not only improves the accuracy and response speed of the motor torque control, but also greatly enhances the adaptability of the motor under different loads and working conditions. It can effectively improve the overall performance of the motor drive system, reduce energy consumption, and extend the service life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the execution flow of the method for dynamic adjustment of motor torque based on analog input provided by an embodiment of the present invention.

[0013] Figure 2 It is a schematic diagram of exemplary hardware and software components of a motor torque dynamic adjustment system based on analog input provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 It is a flow chart of a method for dynamic adjustment of motor torque based on analog input provided by an embodiment of the present invention. The method for dynamic adjustment of motor torque based on analog input is introduced in detail below.

[0015] Step S110, obtaining a set of analog input signals received by the servo motor in a running state, wherein the set of analog input signals includes at least one analog input signal, and the analog input signal is used to represent a torque adjustment requirement of an external device for the servo motor.

[0016] Taking the application scenario of a quadruped robot as an example, during the movement of the quadruped robot, the servo motors of each joint need to adjust the torque according to different task requirements. For example, when the quadruped robot is climbing a slope, the servo motors of the leg joints need a larger torque to support the weight of the robot and push it upward. At this time, the external control system (such as the robot's central controller) sends an analog input signal to the servo motor. These analog input signal sets may contain multiple analog input signals, each of which represents a specific requirement for the servo motor torque adjustment.

[0017] Assume that one leg of a quadruped robot is lifting. In order to achieve smooth lifting, the servo motor at the joint needs to gradually increase the torque. During this process, the central controller sends analog input signals to the servo motor based on the information obtained by the robot's attitude sensor (such as gyroscope, accelerometer, etc.). These analog input signals may include information related to the lifting angle, speed, and overall center of gravity adjustment of the robot, which together constitute the analog input signal set.

[0018] For example, when a quadruped robot needs to turn quickly, the servo motor at the steering joint needs to quickly adjust the torque to change the direction of movement. At this time, the external device (such as the robot's motion planning module) will send the corresponding analog input signal to the servo motor according to the steering angle, speed and real-time motion state of the robot. These analog input signals contain the torque adjustment information required to achieve steering, and together with the servo motor signals of other joints, they constitute the entire analog input signal set.

[0019] Step S120, performing signal analysis on the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal, wherein the torque adjustment parameter set includes a target torque value, a torque change rate threshold, and a torque response time window.

[0020] For the analog input signal received by the servo motor of the leg joint of the quadruped robot, the frame processing is first performed. For example, in a cycle of robot walking, the received continuous analog signal is divided into multiple signal frame segments according to the time sequence. Each signal frame segment contains the analog signal sampling value in a shorter continuous time window.

[0021] Taking the knee servo motor of a quadruped robot as an example, in a leg-lifting-stepping-lowering action cycle, the received analog input signal is divided into frames. The analog signal sampling value in each frame segment reflects the change in the demand for knee joint torque adjustment during this time period.

[0022] Then, feature extraction is performed on each signal frame segment. For the signal amplitude feature, it may reflect the size of the torque adjustment required within the frame segment. For example, if the signal amplitude is large, it may mean that within this time period, the knee joint needs a larger torque change to adapt to the movement of the robot leg. The signal frequency feature may be related to the rhythm of the robot's movement. For example, a faster movement rhythm may correspond to a higher signal frequency. The signal duty cycle feature may be related to the duty cycle of the torque adjustment. For example, in some intermittent motion control, the duty cycle feature can reflect the effective action time ratio of the torque in different time periods.

[0023] These signal amplitude features, signal frequency features and signal duty cycle features are input into a predefined parameter mapping model, which is obtained through a large amount of quadruped robot motion experimental data, which contains various analog signal samples and corresponding torque adjustment parameter labels. For example, in multiple robot climbing, crossing obstacles, turning and other motion experiments, the collected analog signal samples are mapped to the torque adjustment parameters (such as target torque value, torque change rate threshold and torque response time window) actually required by the corresponding knee joint, hip joint and other joint servo motors.

[0024] Through the parameter mapping model, the candidate torque adjustment parameter subset corresponding to each signal frame segment can be output. Then, the candidate torque adjustment parameter subsets corresponding to multiple signal frame segments are time-series aligned and parameter fused. For example, for the knee joint servo motor, in the entire leg-lifting-stepping-putting down action cycle, the candidate torque adjustment parameter subsets of different frame segments need to be arranged in chronological order. There may be a continuous change interval of parameters between adjacent frame segments. In this continuous change interval, the target torque value and the torque change rate threshold are weighted and smoothed to eliminate the parameter jump that may be caused by the difference between different frame segments. For the candidate torque adjustment parameter subsets of the discontinuous change interval, the preset interpolation algorithm is used to supplement the intermediate parameter values. Finally, the processed candidate torque adjustment parameter subsets are merged according to the time axis to generate a complete torque adjustment parameter set, which accurately reflects the torque adjustment requirements of the knee joint servo motor in the entire action cycle.

[0025] Step S130, determining the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor, wherein the current operating state parameters include motor winding current sampling values ​​and motor rotor speed detection values.

[0026] During the operation of the hip joint servo motor of the quadruped robot, the motor winding current sampling value is first obtained. Due to environmental interference (such as vibration caused by uneven ground) and electromagnetic interference of the motor itself when the robot moves, the original motor winding current sampling value may contain noise. The sliding window mean filter is used to perform preliminary denoising on the original current sampling value. For example, during a short period of time when the robot walks, the sliding window mean calculation is performed on the current sampling value during this period to remove some random noise. Then the denoised current value is analyzed by wavelet transform to decompose the current signal into high-frequency noise components and low-frequency effective signals, and the high-frequency noise is separated and removed. According to the rated current range of the servo motor, the abnormal current value that exceeds the reasonable range is truncated and replaced. For example, if the current value exceeds the rated current range due to instantaneous overload or interference, it is replaced with a reasonable value within the rated current range. Then the processed current signal is phase compensated to eliminate the signal distortion caused by sampling delay, so that the phase of the current signal is consistent with the actual situation. Finally, the compensated current signal is synchronously calibrated with the real-time position signal of the motor rotor to obtain a stable current value.

[0027] At the same time, the motor rotor speed detection value is obtained. During the hip joint movement of the quadruped robot, the motor rotor speed is constantly changing. The motor rotor speed detection value is subjected to speed differential calculation to obtain the rotor acceleration value, which reflects the acceleration or deceleration of the hip joint movement.

[0028] According to the obtained stable current value and rotor acceleration value, a predefined torque mapping relationship table is queried. The torque mapping relationship table is established based on the physical characteristics and kinematic model of the hip joint servo motor of the quadruped robot. For example, different combinations of stable current values ​​and rotor acceleration values ​​correspond to different hip joint torque values. By querying the torque mapping relationship table, the current torque value of the servo motor is determined.

[0029] Based on the change trend of the current torque value within the preset time interval, the torque change rate and torque fluctuation amplitude are calculated. During the walking process of the quadruped robot, the torque of the hip joint will fluctuate with the movement of the legs. For example, the torque may gradually decrease when the leg is raised, and the torque may gradually increase when the leg is lowered. By analyzing the change of the torque value over a period of time, the torque change rate and torque fluctuation amplitude are calculated. According to the torque change rate and torque fluctuation amplitude, the torque dynamic response characteristics of the servo motor are determined, which include the maximum allowable adjustment rate and the minimum stabilization time threshold. For example, if the torque fluctuation amplitude is large, in order to ensure the stability of the robot's movement, the maximum allowable adjustment rate may be relatively small, and the minimum stabilization time threshold may be relatively long to avoid the robot losing balance due to excessive torque adjustment.

[0030] Step S140, generating a torque dynamic adjustment strategy for the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, wherein the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence.

[0031] Taking the ankle servo motor of a quadruped robot as an example, first calculate the difference between the target torque value and the current torque value. Assume that the robot is walking from flat ground to a slope, and the ankle joint needs to increase torque to adapt to the slope. Calculate the difference between the target torque value (the torque required to adapt to the slope) and the current torque value (the torque when walking on flat ground). Determine the torque adjustment direction based on the difference, which is the positive direction here, that is, the torque needs to be increased.

[0032] Based on the maximum allowable adjustment rate and minimum stabilization time threshold in the torque dynamic response characteristics, multiple torque adjustment stages are divided. For example, the total adjustment amount is determined based on the absolute value of the difference between the target torque value and the current torque value. If the total adjustment amount is large, the theoretical minimum adjustment time is calculated considering the torque dynamic response characteristics of the ankle joint servo motor. The number of stage divisions is determined based on the ratio of the theoretical minimum adjustment time to the torque response time window. In this process, a non-uniform time allocation strategy is adopted. For example, a shorter time window is allocated to the initial stage so that the ankle joint can quickly approach the target value to adapt to the slope change as soon as possible. A longer time window is allocated to the final stage to achieve accurate and stable torque adjustment to ensure the stability of the robot walking on the slope. A transition buffer is set inside each stage to absorb torque fluctuations caused by sudden load changes caused by obstacles such as small stones that the robot may encounter when walking on the slope.

[0033] According to the torque change rate threshold constraint, a corresponding torque adjustment slope is assigned to each adjustment stage, and the torque adjustment slope is used to control the acceleration of the torque change in this stage. For example, in the initial stage, due to the short time window, a larger torque adjustment slope can be set to enable the torque to increase rapidly; in the middle stage, according to the torque change rate threshold constraint and the smoothness requirement of the robot motion, a moderate torque adjustment slope is set; in the final stage, in order to accurately and stably reach the target torque value, a smaller torque adjustment slope is set.

[0034] The stage target torque value, adjustment time window and torque adjustment slope of each adjustment stage are combined to generate a segmented torque adjustment instruction sequence. For example, in the torque adjustment instruction sequence of the ankle joint servo motor, the instructions in the initial stage may include instructions to quickly increase to an intermediate target torque value within a short time window, the intermediate stage includes instructions to adjust to another intermediate target torque value according to a certain slope within a moderate time window, and the final stage includes instructions to slowly adjust to the target torque value that adapts to the slope within a longer time window. The segmented torque adjustment instruction sequence is time-checked to ensure that the torque adjustment slope changes of adjacent stages do not exceed the preset connection threshold to ensure the smoothness of the torque adjustment process.

[0035] Step S150, executing the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy, so that the output torque of the servo motor dynamically matches the target torque value according to a preset time window, while satisfying the torque change rate threshold constraint.

[0036] For the elbow joint servo motor of the quadruped robot, the first adjustment instruction is sent to the torque control module of the servo motor according to the first stage instruction in the segmented torque adjustment instruction sequence. For example, when the robot is performing a grasping action, the first stage target torque value may be the torque required to bend the elbow joint to an initial angle, and the corresponding adjustment time window is a shorter time.

[0037] In the first stage adjustment time window, the actual torque change curve of the servo motor is monitored in real time, and the actual torque change rate is calculated. Assume that during the elbow bending process, due to the friction of the mechanical structure or slight changes in the load, the actual torque change rate may deviate from the torque adjustment slope in the first stage instruction. When it is detected that the actual torque change rate deviates from the torque adjustment slope in the first stage instruction, the adjustment time window or target torque value of the subsequent stage is dynamically corrected according to the deviation amplitude.

[0038] For example, if the calculated deviation percentage between the actual torque change rate and the command required rate is less than the first threshold, only the torque adjustment slope of the subsequent stage is fine-tuned. If the deviation percentage is between the first threshold and the second threshold, the time window of the subsequent stage is extended proportionally. For example, the time window originally planned for a certain intermediate stage is 5 seconds, but it needs to be extended to 6 seconds due to the deviation. If the deviation percentage exceeds the second threshold, the remaining adjustment amount is recalculated and a new stage division scheme is generated. The timestamp mark of the original instruction is retained during the correction process to ensure that the corrected instruction sequence is traceable.

[0039] After the current stage adjustment is completed, the next stage adjustment instruction is triggered according to the revised segmented torque adjustment instruction sequence until all stage instructions are executed. After the final stage adjustment is completed, the output torque stability of the servo motor is continuously monitored. For example, after the robot completes the grasping action, the elbow joint needs to maintain a certain torque to stabilize the grasped object. If the torque fluctuation is detected to be beyond the allowable range, the local adjustment instruction is regenerated and inserted into the execution queue to ensure that the elbow joint can stably maintain the grasping state and ensure the accuracy and stability of the entire operation of the quadruped robot.

[0040] During the movement of the quadruped robot, the dynamic torque adjustment mechanism of the servo motor ensures that the robot joints can accurately and stably adjust the torque according to external requirements and internal conditions, thereby achieving efficient and smooth movement. Whether walking on different terrains, performing various tasks or responding to external interference, the robot's performance and reliability can be guaranteed.

[0041] Based on the above steps, the embodiment of the present application obtains a set including multiple analog input signals, and obtains a torque adjustment parameter set by deeply analyzing the analog input signal set, which not only includes the target torque value, but also introduces two key parameters, the torque change rate threshold and the torque response time window. In the torque determination link, based on the torque adjustment parameter set and the current operating state parameters of the servo motor (motor winding current sampling value and motor rotor speed detection value), the current torque value and torque dynamic response characteristics of the servo motor are determined, which fully considers the real-time operating state of the motor and can more accurately grasp the current working condition of the motor. In terms of strategy generation, according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, a torque dynamic adjustment strategy including a segmented torque adjustment instruction sequence is generated, which can flexibly adjust the torque according to the needs and motor states of different stages, and realize more refined and intelligent torque dynamic adjustment. This segmented strategy can dynamically match the target torque value for different time windows and motor operating states, and strictly meet the torque change rate threshold constraint, avoiding the impact of torque mutation on the motor and equipment, and significantly improving the stability and reliability of motor operation. Finally, in the actual execution process, the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed, so that the output torque of the servo motor can accurately and dynamically match the target torque value within the preset time window, while strictly meeting the torque change rate threshold constraint. This not only improves the accuracy and response speed of the motor torque control, but also greatly enhances the adaptability of the motor under different loads and working conditions. It can effectively improve the overall performance of the motor drive system, reduce energy consumption, and extend the service life of the equipment.

[0042] In a possible implementation, step S120 includes:

[0043] Step S121, performing frame processing on each analog input signal to obtain multiple signal frame segments, each signal frame segment contains analog signal sampling values ​​within a continuous time window.

[0044] When the quadruped robot performs an action such as crossing an obstacle, the servo motors at the leg joints receive a set of analog input signals from an external device, which are intended to adjust the torque of the servo motors to achieve precise movement of the leg joints of the quadruped robot.

[0045] First, taking the knee servo motor of a quadruped robot as an example, during the action of crossing an obstacle, the received analog input signal is segmented in time order. Each segmented signal frame segment covers the analog signal sampling values ​​within a continuous time window. The determination of the continuous time window is based on the key action stages of the quadruped robot's movement, such as the different stages from preparing to cross the obstacle to the leg lifting, crossing, and then putting down. During the leg lifting stage, the analog signal sampling values ​​within the signal frame segment reflect the changes in the demand for knee joint torque adjustment during this specific time period.

[0046] Step S122, extracting features from each signal frame segment to obtain signal amplitude features, signal frequency features, and signal duty cycle features corresponding to each signal frame segment.

[0047] For example, according to the signal amplitude characteristics, when a quadruped robot crosses an obstacle, the knee joint needs different degrees of torque to support the body weight and complete the crossing action. If the signal amplitude is large within a certain signal frame segment, it means that the knee joint needs a larger torque adjustment during this time period. The signal frequency characteristics are related to the rhythm of the quadruped robot's movement. In the more complex action of crossing an obstacle, the robot's movement rhythm will change, and this change will be reflected in the signal frequency. For example, when the leg is raised quickly to cross an obstacle, the signal frequency may increase. The signal duty cycle characteristics are related to the duty cycle of the torque adjustment. In the process of crossing an obstacle, the knee joint may need to intermittently increase or decrease the torque, and the duty cycle characteristics can reflect the effective action time ratio of the torque in different time periods.

[0048] Step S123, inputting the signal amplitude characteristics, signal frequency characteristics and signal duty cycle characteristics into a predefined parameter mapping model, and outputting a subset of candidate torque adjustment parameters corresponding to each signal frame segment through the parameter mapping model.

[0049] In detail, the parameter mapping model is established through a large number of experimental data of quadruped robots crossing different types of obstacles. In these experiments, various analog signal samples were collected and the corresponding torque adjustment parameter labels were recorded. These torque adjustment parameter labels contain information such as target torque value, torque change rate threshold and torque response time window. The analog signal samples of the knee joint servo motor when crossing obstacles of different heights and shapes, as well as the corresponding torque adjustment parameters, are used to construct the parameter mapping model. Through the parameter mapping model, the candidate torque adjustment parameter subset corresponding to each signal frame segment is output.

[0050] Step S124, performing time alignment and parameter fusion on the candidate torque adjustment parameter subsets corresponding to the multiple signal frame segments to generate a torque adjustment parameter set corresponding to the analog input signal. The parameter mapping model is obtained by training the mapping relationship between multiple groups of analog signal samples and torque adjustment parameter labels.

[0051] In detail, during the entire process of the quadruped robot crossing the obstacle, the candidate torque adjustment parameter subsets corresponding to different signal frame segments need to be arranged in time order. Taking the knee servo motor as an example, during the entire action cycle of crossing the obstacle, from preparing to crossing to completing the crossing, the candidate torque adjustment parameter subsets of each signal frame segment must be arranged in the correct time sequence. There may be a continuous change interval of parameters between adjacent signal frame segments, and the target torque value and the torque change rate threshold are weighted and smoothed in this continuous change interval. For example, when transitioning from the leg lifting stage to the crossing stage, in order to avoid the sudden change of torque adjustment, the target torque value and the torque change rate threshold are weighted and smoothed to eliminate the parameter jump that may be caused by the difference between different frame segments. For the candidate torque adjustment parameter subsets of the discontinuous change interval, a preset interpolation algorithm is used to supplement the intermediate parameter values. For example, in the process of crossing the obstacle, if a signal frame segment is missing or some parameters are inaccurate due to interference, the interpolation algorithm is used to supplement them. The processed candidate torque adjustment parameter subsets are merged according to the time axis to finally generate a complete torque adjustment parameter set, which accurately reflects the torque adjustment requirements of the knee joint servo motor when the quadruped robot crosses obstacles.

[0052] In a possible implementation, step S130 includes:

[0053] Step S131, obtaining a motor winding current sampling value of the servo motor, and filtering the motor winding current sampling value to obtain a stable current value.

[0054] When the quadruped robot walks on rough terrain, the servo motors at the leg joints are in operation. When the servo motors at the leg joints of the quadruped robot are working, the motor winding current will fluctuate due to various factors. For example, when the leg joints of the robot walk on rough terrain, the force at the joints is constantly changing due to the uneven ground, which will cause the motor load to change, thereby changing the motor winding current. The original motor winding current sampling value contains various interference factors, such as electromagnetic interference and noise caused by the robot's own vibration. In order to obtain an accurate current value for determining the torque, the motor winding current sampling value needs to be filtered. The sliding window mean filter is used to perform preliminary denoising on the original current sampling value. For example, in a short time interval, the mean of the current sampling value in the time interval is calculated to smooth the current signal and remove some random small fluctuation noise. Then, the denoised current value is subjected to wavelet transform analysis. The current signal is decomposed into high-frequency noise components and low-frequency effective signals through wavelet transform, and the high-frequency noise is separated and removed, thereby further improving the purity of the current signal. Then, according to the rated current range of the servo motor, the abnormal current values ​​that exceed the reasonable range are truncated and replaced. For example, if the current exceeds the rated current range instantly because the robot is suddenly impacted by a large external force, this abnormal value may affect the accurate calculation of the torque, so it is replaced with a reasonable value within the rated current range. After that, the processed current signal is phase compensated. Since there may be delays in the sampling process, this will cause the phase of the current signal to be inconsistent with the actual situation. Phase compensation can eliminate this signal distortion caused by sampling delay. Finally, the compensated current signal is synchronously calibrated with the real-time position signal of the motor rotor to obtain a stable current value.

[0055] Step S132, obtaining the motor rotor speed detection value of the servo motor, and performing speed differential calculation on the motor rotor speed detection value to obtain the rotor acceleration value.

[0056] For example, when the robot's leg joint is quickly lowered from the raised state, the motor rotor speed will increase rapidly; and when the joint is slowly bent, the motor rotor speed will slow down accordingly. The motor rotor speed detection value is calculated by speed differential to obtain the rotor acceleration value, which is very important for understanding the dynamic characteristics of joint motion. For example, when the quadruped robot crosses a small earth slope, the motor rotor speed gradually increases during the process of the leg joint being lifted up, and the rotor acceleration value obtained by speed differential calculation is positive, indicating that the joint is in an accelerated motion state; when the leg joint passes the top of the earth slope and begins to lower, the motor rotor speed gradually decreases, and the rotor acceleration value is negative, indicating that the joint is in a decelerated motion state.

[0057] Step S133, according to the stable current value and the rotor acceleration value, query a predefined torque mapping relationship table to determine the current torque value of the servo motor.

[0058] The torque mapping relationship table is constructed based on the physical characteristics and kinematic model of the servo motor of the leg joint of the quadruped robot. Different torque values ​​correspond to different combinations of stable current values ​​and rotor acceleration values. For example, when the quadruped robot walks on flat ground, the stable current value and rotor acceleration value of the leg joint are in a relatively stable range, and the corresponding torque value can maintain the normal walking of the robot; when the robot is climbing a slope, the stable current value and rotor acceleration value change. By querying the torque mapping relationship table, the current torque value of the leg joint servo motor at this time can be determined, and the current torque value can meet the power requirements of the leg joint of the robot when climbing a slope.

[0059] Step S134, calculating the torque change rate and the torque fluctuation amplitude based on the change trend of the current torque value within a preset time interval.

[0060] During the walking process of a quadruped robot, the torque of the leg joints will change with the robot's motion posture and the changes in the terrain. For example, when a quadruped robot walks on uneven ground, the leg joints need to constantly adjust the torque to adapt to the ups and downs of the terrain. In a specific preset time interval, such as the time interval from when one of the robot's feet touches a raised stone to when it completely crosses the stone, analyze the change in the current torque value of the leg joint servo motor. If the torque value changes frequently and with a large amplitude within this interval, the torque change rate will be large, and the torque fluctuation amplitude will also be large. This may be caused by the complexity of the terrain causing the leg joint force to change continuously.

[0061] Step S135, determining the torque dynamic response characteristics of the servo motor according to the torque change rate and the torque fluctuation amplitude, wherein the torque dynamic response characteristics include a maximum allowable adjustment rate and a minimum stabilization time threshold.

[0062] For example, if the torque fluctuation amplitude is large during the above-mentioned stone crossing process, in order to ensure the stability and safety of the quadruped robot, the maximum allowable adjustment rate may be relatively small, because a large torque adjustment rate may cause instability in the robot joint movement. At the same time, the minimum stabilization time threshold may be relatively long to ensure that the leg joints have enough time to stabilize after the torque adjustment, avoiding shaking or imbalance of the robot caused by too fast torque adjustment.

[0063] In a possible implementation, step S140 includes:

[0064] Step S141, calculating the difference between the target torque value and the current torque value, and determining the torque adjustment direction according to the difference.

[0065] For example, when a quadruped robot needs to walk from flat ground onto a slope, the target torque value at the leg joint will be greater than the current torque value, and the difference is positive, so the torque adjustment direction is the direction of increasing the torque.

[0066] Step S142, based on the maximum allowable adjustment rate and the minimum stable time threshold in the torque dynamic response characteristics, a plurality of torque adjustment stages are divided, each stage corresponding to an adjustment time window and a stage target torque value.

[0067] Taking the quadruped robot climbing as an example, the total adjustment amount is determined according to the absolute value of the difference between the target torque value and the current torque value. If the total adjustment amount is large, the theoretical minimum adjustment time is calculated considering the torque dynamic response characteristics of the leg joint servo motor. The number of stage divisions is determined according to the ratio of the theoretical minimum adjustment time to the torque response time window. In this process, a non-uniform time allocation strategy is adopted. For example, a shorter time window is allocated for the initial stage so that the leg joints can quickly approach the target value to adapt to the slope gradient as soon as possible. In the initial stage of climbing, the leg joints need to increase the torque quickly to provide sufficient power, so the adjustment time window in the initial stage is short, and the target torque value is set to a value that allows the robot to start climbing smoothly. A longer time window is allocated for the final stage to achieve accurate and stable torque adjustment to ensure the stability of the robot walking on the slope. In the final stage of climbing, the leg joints need to accurately reach the target torque value that adapts to the slope gradient and remain stable, so the adjustment time window in the final stage is longer. A transition buffer is set inside each stage to absorb torque fluctuations caused by sudden load changes due to obstacles such as small stones that the robot may encounter when walking on a slope.

[0068] Step S143, according to the torque change rate threshold constraint, a corresponding torque adjustment slope is allocated to each adjustment stage, and the torque adjustment slope is used to control the acceleration of the torque change in this stage.

[0069] In detail, the torque adjustment slope is used to control the acceleration of the torque change in this stage. In the process of quadruped robot climbing, for the initial stage, due to the short time window, a larger torque adjustment slope can be set to enable the torque to increase rapidly; in the middle stage, according to the torque change rate threshold constraint and the robot motion stability requirements, a moderate torque adjustment slope is set; in the final stage, in order to accurately and stably reach the target torque value, a smaller torque adjustment slope is set. For example, in the middle stage of climbing, the leg joints need to gradually increase the torque without affecting the overall stability of the robot, so setting a moderate torque adjustment slope can meet this requirement.

[0070] Step S144, combining the stage target torque value, adjustment time window and torque adjustment slope of each adjustment stage to generate a segmented torque adjustment instruction sequence.

[0071] For example, in the torque adjustment instruction sequence of the servo motor of the leg joint of a quadruped robot, the instructions in the initial stage may include instructions for quickly increasing to an intermediate target torque value within a short time window, the intermediate stage includes instructions for adjusting to another intermediate target torque value according to a certain slope within a moderate time window, and the final stage includes instructions for slowly adjusting to the target torque value that adapts to the slope within a longer time window.

[0072] Step S145, performing a timing check on the segmented torque adjustment instruction sequence to ensure that the torque adjustment slope change in adjacent stages does not exceed a preset connection threshold.

[0073] For example, during the climbing process of a quadruped robot, the torque adjustment slope changes from the initial stage to the intermediate stage and from the intermediate stage to the final stage need to transition smoothly. If the torque adjustment slope changes too much in adjacent stages, it may cause the leg joints to move unsteadily, thus affecting the walking stability of the quadruped robot on the slope. Through timing verification, the stability of the entire torque adjustment process can be guaranteed, so that the leg joints can accurately adjust the torque according to the predetermined strategy to adapt to the movement requirements of the quadruped robot under different terrains.

[0074] In a possible implementation, step S150 includes:

[0075] Step S151, according to the first stage instruction in the segmented torque adjustment instruction sequence, a first adjustment instruction is sent to the torque control module of the servo motor, wherein the first adjustment instruction includes a first stage target torque value and a corresponding adjustment time window.

[0076] When the quadruped robot is ready to climb a slope, the first adjustment instruction is sent to the torque control module of the servo motor according to the first stage instruction in the segmented torque adjustment instruction sequence. This first adjustment instruction contains the first stage target torque value and the corresponding adjustment time window. For example, at the beginning of the quadruped robot climbing action, the first stage target torque value is pre-calculated based on factors such as the robot's initial posture, weight, and the starting slope of the slope. The target torque value is the key to ensure that the robot's leg joints can begin to smoothly apply sufficient power on the slope. The corresponding adjustment time window is set according to the characteristics of the quadruped robot's power system and the requirements of the initial stage of the slope. The time window is short in order to allow the servo motor of the leg joint to respond quickly so that the robot can adapt to the starting slope of the slope as soon as possible.

[0077] Step S152, within the first stage adjustment time window, real-time monitoring of the actual torque change curve of the servo motor and calculation of the actual torque change rate.

[0078] In the initial stage of the quadruped robot climbing, as the servo motor begins to adjust the torque according to the first adjustment instruction, the sensor will continue to monitor the actual operation of the motor. The actual torque change curve reflects the relationship between the torque and time in this time window. By calculating the continuously collected torque values, the actual torque change rate can be obtained. For example, at the beginning of the climbing, due to the robot leg joint switching from the flat state to the slope state, it may be affected by the sudden change of ground friction, the slight jamming of the joint mechanical structure and other factors, resulting in the actual torque change rate may not be completely consistent with the torque adjustment slope in the first stage instruction.

[0079] Step S153, when it is detected that the actual torque change rate deviates from the torque adjustment slope in the first stage instruction, the adjustment time window or target torque value of the subsequent stage is dynamically corrected according to the deviation amplitude.

[0080] Step S154, after the current stage adjustment is completed, the next stage adjustment instruction is triggered according to the revised segmented torque adjustment instruction sequence until all stage instructions are executed.

[0081] During the climbing process of the quadruped robot, when the first stage of adjustment is completed, whether it is in accordance with the original instructions or the revised instructions, the adjustment instructions for the next stage will be triggered according to the completion of this stage. For example, after the torque adjustment is completed in accordance with the adjusted instructions in the first stage, the servo motors of the robot's leg joints begin to execute the adjustment instructions of the intermediate stage. This process will continue until the instructions of all stages are executed, thereby ensuring that during the robot's climbing process, the servo motor torque of the leg joints can be gradually adjusted according to the predetermined strategy to adapt to the different slopes and the robot's power requirements.

[0082] Step S155, after the final stage adjustment is completed, the output torque stability of the servo motor is continuously monitored. If it is detected that the torque fluctuation exceeds the allowable range, a local adjustment instruction is regenerated and inserted into the execution queue.

[0083] After the quadruped robot climbs up the slope and completes the entire climbing action, the servo motor of the leg joint enters a stable operating state. At this time, it is still necessary to continuously monitor the output torque stability of the servo motor. For example, since the robot may have some slight shaking or adjustment movements on the slope, these movements may cause torque fluctuations in the servo motor. If the torque fluctuation is detected to be beyond the allowable range, this may affect the robot's stable standing on the slope or subsequent movements. At this time, it is necessary to regenerate the local adjustment instruction and insert it into the execution queue. The regenerated local adjustment instruction is calculated based on the current torque fluctuation situation, the robot's posture, and the slope situation. For example, if the torque fluctuation is caused by a slight offset of the robot's center of gravity, the local adjustment instruction may adjust the servo motor torque on one side of the leg joint to rebalance the robot's center of gravity and ensure the robot's stability on the slope. In this way, through the precise execution and dynamic correction of the torque dynamic adjustment strategy, the quadruped robot can effectively control the torque of the servo motor in various complex terrains and action situations to ensure the stable operation and normal operation of the robot.

[0084] Wherein, step S153 includes:

[0085] Step S1531, calculating the deviation percentage between the actual torque change rate and the command required rate.

[0086] During the climbing process of the quadruped robot, if the deviation percentage is less than the first threshold, only the torque adjustment slope of the subsequent stage is fine-tuned. For example, when the deviation percentage is small, it means that the actual situation is close to the expected situation, and the deviation may be caused by some small interference factors. At this time, by fine-tuning the torque adjustment slope of the subsequent stage, the subsequent torque adjustment process can gradually return to the expected trajectory. For example, in the subsequent intermediate stage, the torque adjustment slope is slightly increased or decreased to compensate for the small deviation in the first stage, ensuring the consistency and accuracy of the entire torque adjustment process.

[0087] Step S1532: when the deviation percentage is less than the first threshold, only the torque adjustment slope of the subsequent stage is fine-tuned.

[0088] Step S1533: when the deviation percentage is between the first threshold and the second threshold, the time window of the subsequent stage is extended proportionally.

[0089] For example, in the process of a quadruped robot climbing a slope, if the robot's leg joints are subject to some moderate interference in the first stage, resulting in the deviation between the actual torque change rate and the command required rate being in this range. At this time, in order to ensure that the subsequent torque adjustment can proceed smoothly and enable the robot to climb the slope stably, the time window of the subsequent stage is extended proportionally. Assuming that the original time window of the intermediate stage is 5 seconds, it is extended to 6 seconds or 7 seconds in a certain proportion according to the specific situation of the deviation. The purpose of this is to give the servo motor more time to adjust the torque to adapt to the impact of the deviation in the first stage, while avoiding the instability of the robot on the slope due to too fast adjustment.

[0090] Step S1534: when the deviation percentage exceeds the second threshold, recalculate the remaining adjustment amount and generate a new stage division scheme.

[0091] When the quadruped robot is climbing, if the deviation in the first stage is very large and exceeds the second threshold, this may be because the robot encountered a major unexpected situation at the beginning of climbing, such as suddenly hitting a small stone on the slope or a large jam in the joints. At this time, the original torque adjustment scheme may not meet the needs of climbing, and the remaining adjustment amount needs to be recalculated. The recalculation of the remaining adjustment amount is based on factors such as the current actual torque value, the target torque value, and the time that has passed. Then a new stage division scheme is generated based on the recalculated results. For example, it may be necessary to increase the number of intermediate stages, or readjust the target torque value and time window of each stage. In this process, the new stage division scheme should fully consider the stability and power requirements of the quadruped robot to ensure that the robot can continue to climb smoothly.

[0092] Step S1535, retaining the timestamp of the original instruction during the correction process to ensure that the corrected instruction sequence is traceable.

[0093] During the entire torque adjustment process of the quadruped robot climbing, whether it is fine-tuning the torque adjustment slope of the subsequent stage, extending the time window or regenerating the stage division scheme, the timestamp mark of the original instruction is retained. The timestamp mark records the issuance time, execution time and correction time of each instruction. For example, when it is necessary to analyze or troubleshoot the torque adjustment during the entire climbing process, the timestamp mark can help accurately locate the situation of each stage, understand when the deviation occurred, and what kind of corrective measures were taken. It is of great significance for optimizing the torque adjustment strategy of the quadruped robot and improving the reliability and performance of the robot.

[0094] In a possible implementation, step S124 includes:

[0095] Step S1241, extracting the timestamp information of each signal frame segment, and sorting the candidate torque adjustment parameter subsets according to the time sequence of the timestamp information.

[0096] When a quadruped robot performs complex walking movements, such as walking on rugged and varied terrain (such as slopes with different slopes, potholes, and small obstacles), the analog input signals received by the servo motors of the leg joints need to be accurately processed to determine the appropriate set of torque adjustment parameters.

[0097] First, during the movement of the leg joints of the quadruped robot, the analog input signals at different times reflect different movement requirements. For example, when the robot's legs are raised to cross a small pothole, the signal frame segment corresponding to this moment and the signal frame segment in the process of the legs crossing the pothole and putting it down are in time sequence. By extracting the timestamp information of each signal frame segment, the order of each signal frame segment in the entire walking action can be clarified. Then, the candidate torque adjustment parameter subsets are sorted according to this order to ensure that these candidate torque adjustment parameter subsets are temporally coherent.

[0098] Step S1242: performing overlapping region detection on candidate torque adjustment parameter subsets of adjacent signal frame segments to determine a parameter continuous change interval.

[0099] During the walking process of the quadruped robot, there is a parameter transition between adjacent action stages. For example, when the legs transition from the normal walking state to the action of crossing an obstacle, there is a certain overlap in the candidate torque adjustment parameter subsets of the signal frame segments corresponding to the two adjacent stages. By detecting the overlapping area, it can be determined that the parameters are continuously changing within this interval. This continuous change reflects the coherence of the joint movement of the quadruped robot's legs. For example, before crossing an obstacle, the legs will gradually adjust the torque to prepare for higher load requirements, and this adjustment process is manifested as a continuous change of parameters in adjacent signal frame segments.

[0100] Step S1243, within the interval of continuous parameter change, weighted smoothing is performed on the target torque value and the torque change rate threshold to eliminate parameter jumps.

[0101] Take the process of a quadruped robot crossing obstacles of different heights as an example. When the legs cross from a lower obstacle to a higher obstacle, the target torque value and the torque change rate threshold need to be adjusted step by step. If smoothing is not performed, parameter jumps may occur, which will cause the torque adjustment of the servo motor to be unstable. Through weighted smoothing, the target torque value and the torque change rate threshold can be smoothly transitioned within the continuous change range. For example, in the process of crossing a small stone (smaller torque required) to crossing a larger stone (larger torque required), the target torque value will gradually increase, and this increase process is smooth, without sudden and large changes, thus ensuring the stability and smoothness of the leg joint movement.

[0102] Step S1244: For the candidate torque adjustment parameter subset in the discontinuous change interval, a preset interpolation algorithm is used to supplement the intermediate parameter values.

[0103] During the walking process of the quadruped robot, some special situations may occur, resulting in discontinuous change intervals between signal frame segments. For example, this situation may occur when the robot suddenly changes its walking direction or encounters sudden external force interference. Assuming that the quadruped robot was originally walking in a straight line on a flat ground, and suddenly turned to avoid a large obstacle, the turning action may cause a discontinuous change interval between the signal frame segments corresponding to the two different walking directions. At this time, a preset interpolation algorithm is used to supplement the intermediate parameter values. The interpolation algorithm reasonably supplements the missing parameter values ​​in the middle according to the parameter conditions of the two signal frame segments before and after and the time interval between them, so that the entire torque adjustment parameter set is complete on the time axis, thereby ensuring that the servo motor can accurately adjust the torque according to the complete parameters.

[0104] Step S1245 , merging the processed candidate torque adjustment parameter subsets according to the time axis to generate a complete torque adjustment parameter set.

[0105] After the quadruped robot completes the entire walking motion cycle, all candidate torque adjustment parameter subsets that have been processed as described above are merged according to the time axis. The complete set of torque adjustment parameters covers the torque adjustment requirement information of the leg joint servo motors during the entire walking process, including target torque values ​​at different stages, torque change rate thresholds, etc.

[0106] In a possible implementation, step S131 includes:

[0107] Step S1311, using a sliding window mean filter to perform preliminary denoising on the original current sampling value.

[0108] When the quadruped robot is walking, the motor winding current will be disturbed by many factors and generate noise. For example, the robot's own mechanical vibration, the electromagnetic interference of the surrounding environment, etc. will cause the original current sampling value to fluctuate. The sliding window mean filter smoothes these fluctuations by calculating the mean of the current sampling value within a window of a specific size. Assuming that the size of the sliding window is n sampling points, in a short period of time when the robot's leg joint moves, the current values ​​of these n sampling points are averaged to obtain a relatively smooth value, thereby removing some random small-amplitude noise and making the current sampling value closer to the actual current situation.

[0109] Step S1312, performing wavelet transform analysis on the denoised current value to separate high-frequency noise components from low-frequency effective signals.

[0110] During the walking process of the quadruped robot, the current value after preliminary denoising may still contain some complex noise components. Wavelet transform analysis can decompose the current signal into components of different frequencies. High-frequency noise components are usually caused by some rapidly changing interference factors, such as instantaneous electromagnetic pulses, while low-frequency effective signals reflect the basic operating status of the motor, such as load changes related to the movement of the robot's leg joints. Through wavelet transform analysis, high-frequency noise components can be accurately separated and removed, thereby further improving the purity of the current signal.

[0111] Step S1313, according to the rated current range of the servo motor, the abnormal current value exceeding the reasonable range is truncated and replaced.

[0112] When a quadruped robot is walking, it may encounter some special situations that cause abnormal current values. For example, when the leg joints of the robot are suddenly impacted by a large external force, such as stepping into a deep pothole, the motor may be overloaded, causing the current value to exceed the rated current range. This abnormal current value may affect the accurate calculation and control of the torque. At this time, according to the rated current range of the servo motor, the current value that exceeds the reasonable range is truncated and replaced with a reasonable value within the rated current range. For example, if the rated current range is [I_min, I_max], and the detected current value is greater than I_max, it is replaced with I_max to ensure that the torque value calculated based on the current value is reasonable and safe.

[0113] Step S1314, performing phase compensation on the processed current signal to eliminate signal distortion caused by sampling delay.

[0114] During the sampling of the motor winding current, there may be a sampling delay due to factors such as the response time of the sampling device. This sampling delay will cause the phase of the current signal to be inconsistent with the actual situation, thus affecting the accurate judgment of the motor's operating status. For example, when the leg joints of a quadruped robot move, the sampling delay may cause the phase of the current signal to lag behind the actual current change, resulting in a deviation between the calculated torque value and the actual demand. Through the phase compensation algorithm, the phase of the current signal is adjusted according to factors such as the known sampling delay time to make it consistent with the actual situation, thereby eliminating the signal distortion caused by the sampling delay.

[0115] Step S1315, synchronously calibrating the compensated current signal and the real-time position signal of the motor rotor to obtain a stable current value.

[0116] During the operation of the servo motor of the leg joint of the quadruped robot, the position of the motor rotor is closely related to the current signal. The real-time position signal of the motor rotor reflects the actual operating status of the motor. For example, when the leg joint is in different motion positions, the position of the motor rotor will change accordingly, which will affect the electromagnetic relationship of the motor and thus affect the magnitude and characteristics of the current. By synchronously calibrating the compensated current signal with the real-time position signal of the motor rotor, the current value can accurately reflect the actual operating status of the motor at the current rotor position, and obtain a stable current value, which provides a reliable basis for the subsequent accurate calculation of the motor torque.

[0117] In a possible implementation, step S142 includes:

[0118] Step S1421, determining the total adjustment amount according to the absolute value of the difference between the target torque value and the current torque value.

[0119] When the quadruped robot needs to change its walking posture, for example, from slow walking on flat ground to climbing a slope, the target torque value of the leg joint servo motor will change. Assuming that when walking on flat ground, the current torque value of the leg joint servo motor is T1, and the target torque value required when climbing a slope is T2, calculate the absolute value of the difference between T2 and T1, which is the total adjustment amount, which reflects the torque adjustment required from the current state to the target state.

[0120] Step S1422: Calculate the theoretical minimum adjustment time based on the ratio of the total adjustment amount to the maximum allowable adjustment rate.

[0121] During the climbing process of the quadruped robot, due to the physical characteristics of the servo motor and the overall stability requirements of the quadruped robot, there is a maximum allowable adjustment rate, which limits the speed of torque adjustment to avoid robot instability due to excessively fast torque adjustment. For example, if the total adjustment amount is ΔT and the maximum allowable adjustment rate is v, then the theoretical minimum adjustment time t=ΔT / v, which is the shortest time required for torque adjustment at the maximum allowable adjustment rate under ideal conditions.

[0122] Step S1423, determining the number of stage divisions according to the ratio of the theoretical minimum adjustment time to the torque response time window.

[0123] In the torque control strategy of the quadruped robot, there is a torque response time window, which specifies the total time range of the entire torque adjustment process. For example, if the theoretical minimum adjustment time is t and the torque response time window is T, then the number of stage divisions n=T / t (where n is an integer value). The number of stage divisions determined in this way can reasonably divide the entire torque adjustment process into multiple stages according to the actual torque adjustment requirements and time constraints.

[0124] Step S1424, adopting a non-uniform time allocation strategy, allocating a shorter time window to the initial stage to quickly approach the target value, and allocating a longer time window to the final stage to achieve precise stability.

[0125] In the process of quadruped robot climbing, the leg joints need to quickly obtain enough torque to adapt to the slope in the initial stage. For example, at the moment of starting to climb, a short time window is allocated for the initial stage, within which the servo motor can adjust the torque at a relatively fast speed so that the torque value quickly approaches an intermediate range of the target value. In the final stage, in order to ensure the precise stability of the leg joints during the climbing process, a longer time window needs to be allocated. In this stage, the torque adjustment speed will be relatively slow to accurately reach the target torque value, and it can remain stable after reaching the target torque value to avoid torque fluctuations affecting the stability of the robot on the slope.

[0126] Step S1425, a transition buffer is set inside each stage to absorb torque fluctuations caused by sudden load changes.

[0127] During the climbing process of the quadruped robot, it may encounter some small obstacles or uneven ground, which will cause the load of the leg joints to change suddenly, thus causing torque fluctuations. For example, when the robot's leg steps on a small stone, the load on the leg joints will increase instantly, causing torque fluctuations. The transition buffer set inside each torque adjustment stage can absorb the torque fluctuations caused by the sudden change in load. The transition buffer can be implemented through some special control algorithms or hardware circuits. When torque fluctuations are detected, they can be adjusted within the buffer so that the torque fluctuations will not affect the stability and accuracy of the entire torque adjustment process, ensuring that the quadruped robot can climb smoothly.

[0128] In a possible implementation, the method further includes:

[0129] Step S210, after the servo motor output torque reaches the target torque value, starting the torque maintenance monitoring mode.

[0130] When a quadruped robot walks on complex terrain (such as those with different slopes, potholes, obstacles, etc.) or performs specific tasks, the servo motors of the leg joints operate according to the established torque adjustment strategy.

[0131] In this embodiment, when the leg joint servo motor of the quadruped robot completes the torque adjustment required for switching from the flat ground walking state to the climbing state and reaches the target torque value when climbing, the torque maintenance monitoring mode is turned on. In this torque maintenance monitoring mode, it is necessary to ensure the stable output of torque in real time to ensure the stability and normal movement of the quadruped robot during the climbing process.

[0132] Step S220, collecting vibration spectrum data and temperature rise data of the servo motor in real time, and analyzing the change of the mechanical load state.

[0133] During the climbing process of the quadruped robot, the vibration spectrum data of the servo motor reflects the operating state of the motor's internal mechanical structure and the influence of the external load. For example, as the robot's climbing angle increases, the load on the leg joints gradually increases, which may cause the vibration spectrum of the motor to change. At the same time, the motor generates heat during the continuous output of torque, and the temperature rise data can reflect the thermal state of the motor. These data are collected by temperature sensors and vibration sensors respectively. If the amplitude of certain frequency components in the vibration spectrum suddenly increases, or the temperature rise rate accelerates, this may indicate that the mechanical load state has changed. For example, the robot's leg joints may encounter a steeper slope or greater friction during the climbing process, causing the motor to bear a larger load.

[0134] Step S230: when a sudden load change feature is detected, the current adjustment strategy is immediately suspended and an emergency braking instruction is triggered.

[0135] Assume that when a quadruped robot is climbing a slope, its legs are suddenly stuck in a stone hidden in the slope, which will cause the motor load to increase instantly, resulting in a load mutation. When this load mutation feature is detected through the analysis of vibration spectrum data and temperature rise data, in order to avoid damage to the motor and other parts of the robot, the currently executed torque adjustment strategy needs to be suspended immediately. At the same time, the emergency braking command is triggered to stop the motor or reduce the speed. The emergency braking command can be sent to the servo motor driver through the control circuit. The driver quickly adjusts the motor current or voltage according to the emergency braking command to achieve braking.

[0136] Step S240: after the braking is completed, reacquire the analog input signal based on the latest load state and generate an adaptive adjustment strategy.

[0137] In the above example, after the servo motor of the leg joint completes braking, the current load state needs to be re-evaluated. For example, factors such as the degree to which the leg is stuck by the stone, the change in the center of gravity of the robot as a whole, and the remaining climbing task are determined. Then, based on these latest load state information, the analog input signal is re-acquired. These analog input signals may come from the central control system of the quadruped robot, which recalculates and sends analog input signals to the servo motor based on the robot's posture sensors (such as gyroscopes, accelerometers, etc.), task planning modules (such as continuing to climb or trying to bypass stones), etc. Based on these new analog input signals, the torque adjustment strategy adapted to the current load state is generated again according to the steps described above. For example, parameters such as the target torque value, the torque change rate threshold, and the torque response time window may need to be adjusted to ensure that the leg joints can continue to work normally under the new load conditions, so that the quadruped robot can safely complete the climbing or take other appropriate actions.

[0138] Step S250, recording the load mutation event and the response strategy into a history database for optimizing the subsequent training process of the parameter mapping model.

[0139] In a possible implementation, step S250 includes:

[0140] Step S251, extracting load mutation event records within a preset time range from the historical database, each load mutation event record including an event triggering timestamp, load mutation feature data, response strategy execution parameters and adjusted operating status data.

[0141] For example, when a quadruped robot walks on various terrains over the past period of time (such as the last month), each time a load mutation event occurs, this information is recorded. The event trigger timestamp accurately records the moment when the load mutation occurs, which helps to analyze the temporal relationship of the event. The load mutation feature data contains detailed information such as torque fluctuation amplitude, vibration spectrum peak, and temperature rise rate of change. For example, when the robot crosses a large pothole at a certain moment, the torque fluctuation amplitude may suddenly increase, the vibration spectrum peak may appear at a certain frequency, and the temperature rise rate of change will also change accordingly. All of this data is recorded. The response strategy execution parameters record the emergency braking instructions taken when the load mutation is detected, and the various parameters in the readjusted torque adjustment strategy (such as target torque value, torque change rate threshold, etc.). The adjusted operating status data reflects the overall operating status of the servo motor and the quadruped robot after the response strategy is adopted, such as whether normal movement is successfully restored or whether new abnormal conditions occur.

[0142] Step S252, classifying the load mutation event records into event types, and dividing them into stable load mutation events, transient impact events and continuous overload events according to the torque fluctuation amplitude, vibration spectrum peak value and temperature rise change rate in the load mutation feature data.

[0143] Taking the operation of a quadruped robot in different scenarios as an example, if the torque fluctuation amplitude is small, the vibration spectrum peak is low, and the temperature rise rate of change is relatively stable, this situation may be classified as a stable load mutation event. For example, when the robot walks on a relatively flat but slightly undulating ground, it occasionally encounters small bumps or depressions, and the resulting load mutation belongs to this category. When the torque fluctuation amplitude is large, the vibration spectrum peak suddenly increases, and the temperature rise rate of change increases rapidly in a short period of time, it may be a transient impact event. For example, the robot's leg suddenly hits a hard obstacle, generating a large impact force instantly. If the torque fluctuation amplitude continues to be at a high level, the vibration spectrum peak remains at a high position for a long time, and the temperature rise rate of change continues to increase, this is a continuous overload event, which may be caused by the robot's continuous excessive load when performing a certain task (such as climbing a slope for a long time or dragging a heavy object).

[0144] Step S253, extracting a feature data set associated with the parameter mapping model from the classified load mutation event record, wherein the feature data set includes an analog input signal segment before the event occurs, a torque adjustment parameter set, current operating state parameters, and dynamic adjustment parameters in the response strategy.

[0145] In the scenario of a quadruped robot, the analog input signal fragment before the event reflects the signal sent by the central control system to the servo motor for torque adjustment before the load mutation. For example, before the robot is about to encounter an obstacle, the analog input signal may be adjusted according to the normal walking mode. The torque adjustment parameter set includes parameters such as the target torque value, the torque change rate threshold, and the torque response time window. These parameters are set according to normal operating requirements before the load mutation. Current operating status parameters such as the motor winding current sampling value and the motor rotor speed detection value reflect the actual operating status of the servo motor before the load mutation. The dynamic adjustment parameters in the response strategy are the relevant parameters in the response measures taken after the load mutation is detected, such as the braking speed during emergency braking and the readjusted torque adjustment slope.

[0146] Step S254, performing time alignment processing on the feature data set, matching the analog input signal segments, torque adjustment parameter sets and dynamic adjustment parameters corresponding to the same load mutation event according to the time axis, and generating an original training data set with time series labels.

[0147] For example, for a transient impact event in which a quadruped robot's leg hits an obstacle, the analog input signal fragments in a short period of time before the event are accurately matched with the torque adjustment parameter set, motor winding current sampling value, motor rotor speed detection value, and dynamic adjustment parameters in the response strategy (such as the start time of emergency braking, the time point of re-adjusting torque, etc.) in chronological order, and each data element is annotated with an accurate time label to generate the original training data set. The purpose of this is to accurately reflect the temporal relationship between data in subsequent model training, so as to better learn the parameter mapping relationship under load mutation conditions.

[0148] Step S255, reconstructing the analog input signal segments in the original training data group, removing the noise interference introduced in the signal acquisition process, and standardizing the signal amplitude range to generate a denoised standard signal sample.

[0149] In the actual operation of the quadruped robot, due to environmental interference (such as electromagnetic interference, mechanical vibration, etc.) and the limitations of the acquisition equipment, the analog input signal fragments may contain noise. The high-frequency noise is removed by signal processing algorithms, such as filtering algorithms, and then standardized according to the known signal amplitude range. For example, if the amplitude range of the original analog input signal is [-10V to 10V], it is standardized to the range of [0 to 1]. The generated denoised standard signal sample is purer, which is conducive to the parameter mapping model to accurately learn the relationship between the signal and the torque adjustment parameter.

[0150] Step S256, performing correlation analysis on the torque adjustment parameter set and the dynamic adjustment parameters in the original training data group, identifying conflicting items in the parameter mapping relationship, and marking weight coefficients of the conflicting items based on the actual torque response effects in the adjusted operating status data.

[0151] In the process of dealing with sudden load changes of quadruped robots, there may be conflicts between some parameters in the torque adjustment parameter set and the dynamic adjustment parameters in the response strategy. For example, the adjustment of the target torque value may not match the torque change rate threshold in some cases, resulting in unsatisfactory actual torque response effect. By analyzing the actual torque response effect of the servo motor after the load mutation (such as whether smooth torque adjustment is achieved, whether abnormal operation of the motor is avoided, etc.), if a certain parameter (such as the torque change rate threshold is set too high, resulting in too fast torque adjustment, causing excessive motor vibration) is found to have a negative impact on the actual torque response effect, it is marked as a conflict item and the weight coefficient is annotated according to the severity of the impact. For example, if a conflict item causes obvious unstable operation of the motor, it may be given a higher weight coefficient, indicating that the parameter relationship needs to be focused on and adjusted in subsequent model training.

[0152] Step S257, combining the denoised standard signal sample, the torque adjustment parameter set with the weight coefficients marked, and the dynamic adjustment parameters to generate an enhanced training data set, and adding the enhanced training data set to the training data set of the parameter mapping model.

[0153] In the scenario of a quadruped robot, after the previous processing, the denoised standard signal samples, the torque adjustment parameter set containing weight coefficient annotations (such as the target torque value and its corresponding weight coefficient), and the dynamic adjustment parameters (such as the relevant parameters and their weight coefficients during emergency braking) are combined in a certain format to form an enhanced training data set. This enhanced training data set is then added to the training data set of the parameter mapping model to provide more targeted data for the training of the model, so that the model can better learn how to accurately perform parameter mapping under load mutation conditions and improve the model's adaptability to different load conditions.

[0154] Step S258, performing redundant data cleaning on the training data set of the parameter mapping model, removing the historical data set that does not match the current servo motor model and the redundant data set with repeated mapping relationships, and generating an optimized training data set.

[0155] In the entire operation history of the quadruped robot, there may be some data that is not related to the servo motor model currently used, such as data of other models of servo motors used before or invalid data generated due to data acquisition errors. These data will interfere with the training of the model and need to be removed from the training data set. At the same time, redundant data groups with repeated mapping relationships also need to be cleaned up. For example, if there are multiple data groups with exactly the same mapping relationship between signal samples, torque adjustment parameter sets, and dynamic adjustment parameters, only one of the data groups can be retained. After such redundant data cleaning, an optimized training data set is obtained, which is more streamlined and accurate, and is conducive to improving the training efficiency and accuracy of the parameter mapping model.

[0156] Step S259, using the optimized training data set to perform incremental training on the parameter mapping model, adjusting the signal feature weight matrix and the parameter mapping layer connection coefficients in the parameter mapping model, and generating an updated parameter mapping model.

[0157] In the application scenario of the quadruped robot, the parameter mapping model is trained using the optimized training data set. During the training process, the signal feature weight matrix in the model is gradually adjusted according to the sample data in the data set. For example, if a certain signal feature (such as a frequency component in the analog input signal) is proven to have a strong correlation with the torque adjustment parameter in more sample data, the weight value of the signal feature in the weight matrix will be increased. At the same time, the parameter mapping layer connection coefficient is adjusted to optimize the model's mapping relationship from the input signal to the torque adjustment parameter. Through such an incremental training process, an updated parameter mapping model is generated, which can better adapt to the torque adjustment needs of the quadruped robot under different load mutation conditions.

[0158] Step S2510, deploying the updated parameter mapping model to the control unit of the servo system, and verifying the matching degree between the torque adjustment parameter set output by the model and the actual load demand in the next adjustment cycle.

[0159] In the actual operation of the quadruped robot, the updated parameter mapping model is deployed to the control unit of the leg joint servo system. In the subsequent torque adjustment process (for example, the next time the robot encounters a similar load mutation or normal torque adjustment demand), observe whether the torque adjustment parameter set output by the model (such as the target torque value, torque change rate threshold, etc.) matches the actual load demand. For example, when the robot climbs a slope or crosses an obstacle again, according to the actual load conditions (such as the slope of the slope, the size of the obstacle, etc.), check whether the torque adjustment parameters output by the model can enable the servo motor to output the appropriate torque to ensure the normal movement of the robot.

[0160] Step S2511, when it is detected that the matching degree is lower than the preset threshold, a model rollback instruction is triggered and the significance weight event records in the historical database are re-extracted for local training until the matching degree is restored to within the allowable range.

[0161] If it is found during the verification process that the matching degree between the torque adjustment parameter set output by the model and the actual load demand is lower than the preset threshold, this indicates that there may be problems with the updated model. For example, some data may be overfitted during the incremental training process or the key parameter mapping relationship may not be fully learned. At this time, the model rollback instruction is triggered to restore the parameter mapping model to a previous version or state. Then, event records with significant weights are extracted from the historical database again. These event records are records that are considered to have an important impact on model training in previous analysis (such as some special load mutation events or events that have a greater impact on the actual torque response effect). Local training is performed on these significant weight event records, focusing on adjusting the model parameters related to these events, and repeatedly verifying and training until the matching degree between the torque adjustment parameter set output by the model and the actual load demand is restored to the allowable range, thereby ensuring that the parameter mapping model always maintains high accuracy and reliability during the torque adjustment process of the quadruped robot.

[0162] Figure 2 A schematic diagram of exemplary hardware and software components of a motor torque dynamic adjustment system 100 based on analog input that can implement the concept of the present application is shown in some embodiments of the present application. For example, the processor 120 can be used in the motor torque dynamic adjustment system 100 based on analog input and used to perform the functions in the present application.

[0163] The motor torque dynamic adjustment system 100 based on analog input can be a general server or a special-purpose server, both of which can be used to implement the motor torque dynamic adjustment method based on analog input of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0164] For example, the motor torque dynamic adjustment system 100 based on analog input may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the motor torque dynamic adjustment system 100 based on analog input may also include program instructions stored in ROM, RAM, or other types of non-temporary storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The motor torque dynamic adjustment system 100 based on analog input also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0165] For ease of explanation, only one processor is described in the motor torque dynamic adjustment system 100 based on analog input. However, it should be noted that the motor torque dynamic adjustment system 100 based on analog input in the present application may also include multiple processors, so the steps performed by one processor described in the present application may also be performed jointly or individually by multiple processors. For example, if the processor of the motor torque dynamic adjustment system 100 based on analog input executes step A and step B, it should be understood that step A and step B may also be performed jointly by two different processors or individually in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor execute steps A and B together.

[0166] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above method for dynamically adjusting the motor torque based on analog input is implemented.

[0167] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A method for dynamic adjustment of motor torque based on analog input, characterized in that: The method comprises: Acquire a set of analog input signals received by the servo motor in a running state, wherein the set of analog input signals includes at least one analog input signal, and the analog input signal is used to characterize a torque adjustment requirement of an external device on the servo motor; Performing signal analysis on the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal, wherein the torque adjustment parameter set includes a target torque value, a torque change rate threshold, and a torque response time window; Determine the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor, wherein the current operating state parameters include a motor winding current sampling value and a motor rotor speed detection value; Generate a torque dynamic adjustment strategy for the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, wherein the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence; The segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed to make the output torque of the servo motor dynamically match the target torque value according to a preset time window while satisfying the torque change rate threshold constraint.

2. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The signal analysis of the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal includes: Performing frame processing on each analog input signal to obtain multiple signal frame segments, each signal frame segment contains analog signal sampling values ​​within a continuous time window; Extract features from each signal frame segment to obtain the signal amplitude feature, signal frequency feature and signal duty cycle feature corresponding to each signal frame segment; Inputting the signal amplitude feature, signal frequency feature and signal duty cycle feature into a predefined parameter mapping model, and outputting a candidate torque adjustment parameter subset corresponding to each signal frame segment through the parameter mapping model; The candidate torque adjustment parameter subsets corresponding to the multiple signal frame segments are time-aligned and parameter-fused to generate a torque adjustment parameter set corresponding to the analog input signal; wherein the parameter mapping model is obtained by training the mapping relationship between multiple groups of analog signal samples and torque adjustment parameter labels.

3. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The step of determining the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor includes: Obtaining a motor winding current sampling value of a servo motor, and filtering the motor winding current sampling value to obtain a stable current value; Obtaining a motor rotor speed detection value of the servo motor, and performing speed differential calculation on the motor rotor speed detection value to obtain a rotor acceleration value; According to the stable current value and the rotor acceleration value, a predefined torque mapping relationship table is queried to determine the current torque value of the servo motor; Calculating the torque change rate and the torque fluctuation amplitude based on the change trend of the current torque value within a preset time interval; According to the torque change rate and the torque fluctuation amplitude, the torque dynamic response characteristics of the servo motor are determined, and the torque dynamic response characteristics include a maximum allowable adjustment rate and a minimum stabilization time threshold.

4. The method for dynamic adjustment of motor torque based on analog input according to claim 3, characterized in that: The step of generating a torque dynamic adjustment strategy of the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics comprises: Calculating the difference between the target torque value and the current torque value, and determining the torque adjustment direction according to the difference; Based on the maximum allowable adjustment rate and the minimum stable time threshold in the torque dynamic response characteristics, a plurality of torque adjustment stages are divided, each stage corresponding to an adjustment time window and a stage target torque value; According to the torque change rate threshold constraint, a corresponding torque adjustment slope is allocated to each adjustment stage, and the torque adjustment slope is used to control the acceleration of the torque change in the stage; The stage target torque value, adjustment time window and torque adjustment slope of each adjustment stage are combined to generate a segmented torque adjustment instruction sequence; The segmented torque adjustment instruction sequence is time-checked to ensure that the torque adjustment slope change in adjacent stages does not exceed a preset connection threshold.

5. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The step of executing the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy includes: According to the first stage instruction in the segmented torque adjustment instruction sequence, a first adjustment instruction is sent to the torque control module of the servo motor, wherein the first adjustment instruction includes a first stage target torque value and a corresponding adjustment time window; In the first stage adjustment time window, the actual torque change curve of the servo motor is monitored in real time, and the actual torque change rate is calculated; When it is detected that the actual torque change rate deviates from the torque adjustment slope in the first stage command, the adjustment time window or target torque value of the subsequent stage is dynamically corrected according to the deviation amplitude; After the current stage adjustment is completed, the next stage adjustment instruction is triggered according to the revised segmented torque adjustment instruction sequence until all stage instructions are executed; After the final stage of adjustment is completed, the output torque stability of the servo motor is continuously monitored. If the torque fluctuation is detected to be beyond the allowable range, the local adjustment instruction is regenerated and inserted into the execution queue; The dynamically correcting the adjustment time window or target torque value of the subsequent stage according to the deviation amplitude includes: Calculate the deviation percentage between the actual torque change rate and the command required rate; When the deviation percentage is less than the first threshold, only the torque adjustment slope in the subsequent stage is fine-tuned; When the deviation percentage is between the first threshold and the second threshold, the time window of the subsequent stage is extended proportionally; When the deviation percentage exceeds a second threshold, recalculate the remaining adjustment amount and generate a new stage division scheme; The timestamp of the original instruction is retained during the correction process to ensure that the corrected instruction sequence is traceable.

6. The method for dynamic adjustment of motor torque based on analog input according to claim 2, characterized in that: The step of performing timing alignment and parameter fusion on the candidate torque adjustment parameter subsets corresponding to the plurality of signal frame segments to generate a torque adjustment parameter set corresponding to the analog input signal includes: Extracting timestamp information of each signal frame segment, and sorting the candidate torque adjustment parameter subsets according to the time sequence of the timestamp information; Performing overlapping region detection on candidate torque adjustment parameter subsets of adjacent signal frame segments to determine a parameter continuous change interval; In the interval of continuous parameter change, the target torque value and torque change rate threshold are weighted and smoothed to eliminate parameter jumps; For a subset of candidate torque adjustment parameters in a discontinuous variation interval, a preset interpolation algorithm is used to supplement intermediate parameter values; The processed candidate torque adjustment parameter subsets are merged according to the time axis to generate a complete torque adjustment parameter set.

7. The method for dynamic adjustment of motor torque based on analog input according to claim 3, characterized in that: The filtering process of the motor winding current sampling value to obtain a stable current value includes: A sliding window mean filter is used to perform preliminary denoising on the original current sampling values; Perform wavelet transform analysis on the denoised current value to separate high-frequency noise components from low-frequency effective signals; According to the rated current range of the servo motor, the abnormal current value that exceeds the reasonable range is truncated and replaced; Perform phase compensation on the processed current signal to eliminate signal distortion caused by sampling delay; The compensated current signal is synchronously calibrated with the real-time position signal of the motor rotor to obtain a stable current value.

8. The method for dynamic adjustment of motor torque based on analog input according to claim 4, characterized in that: The step of dividing the torque adjustment stages into multiple stages based on the maximum allowable adjustment rate and the minimum stabilization time threshold in the torque dynamic response characteristics includes: Determine the total adjustment amount according to the absolute value of the difference between the target torque value and the current torque value; Based on the ratio of the total adjustment amount to the maximum allowable adjustment rate, the theoretical minimum adjustment time is calculated; The number of stage divisions is determined according to the ratio of the theoretical minimum adjustment time to the torque response time window; A non-uniform time allocation strategy is adopted, which allocates a shorter time window to the initial stage to quickly approach the target value, and a longer time window to the final stage to achieve precise stability; A transition buffer is set inside each stage to absorb torque fluctuations caused by sudden load changes.

9. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The method further comprises: After the servo motor output torque reaches the target torque value, the torque maintenance monitoring mode is started; Collect the vibration spectrum data and temperature rise data of the servo motor in real time and analyze the changes in mechanical load status; When a sudden load change is detected, the current adjustment strategy is immediately suspended and an emergency braking command is triggered; After braking is completed, the analog input signal is re-acquired based on the latest load status and an adaptive adjustment strategy is generated; Record load mutation events and response strategies in a historical database to optimize the training process of subsequent parameter mapping models; The process of recording the load mutation event and the response strategy in the history database for optimizing the subsequent training process of the parameter mapping model includes: Extracting load mutation event records within a preset time range from the historical database, each load mutation event record including an event triggering timestamp, load mutation feature data, response strategy execution parameters, and adjusted operating status data; Classifying the load mutation event records into event types, and dividing them into stable load mutation events, transient impact events, and continuous overload events according to the torque fluctuation amplitude, vibration spectrum peak value, and temperature rise change rate in the load mutation feature data; Extracting a feature data set associated with the parameter mapping model from the classified load mutation event records, the feature data set including an analog input signal segment before the event occurs, a torque adjustment parameter set, current operating state parameters, and dynamic adjustment parameters in a response strategy; Performing time alignment processing on the characteristic data set, matching the analog input signal segments, torque adjustment parameter sets and dynamic adjustment parameters corresponding to the same load mutation event according to the time axis, and generating an original training data set with time series labels; Reconstructing the analog input signal segments in the original training data group, removing the noise interference introduced in the signal acquisition process, and standardizing the signal amplitude range to generate a denoised standard signal sample; Performing correlation analysis on the torque adjustment parameter set and the dynamic adjustment parameter in the original training data group, identifying conflicting items in the parameter mapping relationship, and marking weight coefficients of the conflicting items based on actual torque response effects in the adjusted operating state data; The denoised standard signal sample, the torque adjustment parameter set with the weight coefficients marked, and the dynamic adjustment parameters are combined to generate an enhanced training data set, and the enhanced training data set is added to the training data set of the parameter mapping model; Clean the redundant data of the training data set of the parameter mapping model, remove the historical data sets that do not match the current servo motor model and the redundant data sets with repeated mapping relationships, and generate an optimized training data set; Performing incremental training on the parameter mapping model using the optimized training data set, adjusting the signal feature weight matrix and the parameter mapping layer connection coefficient in the parameter mapping model, and generating an updated parameter mapping model; Deploy the updated parameter mapping model to the control unit of the servo system, and verify the matching degree between the torque adjustment parameter set output by the model and the actual load demand in the next adjustment cycle; When it is detected that the matching degree is lower than the preset threshold, the model rollback instruction is triggered and the significant weight event records in the historical database are re-extracted for local training until the matching degree is restored to the allowable range.

10. A motor torque dynamic adjustment system based on analog input, characterized in that: The motor torque dynamic adjustment system based on analog input includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the motor torque dynamic adjustment method based on analog input as described in any one of claims 1 to 9.

Images