Humanoid robot joint motor driving performance model training method, prediction method and equipment

Abstract

The embodiment of the invention provides a humanoid robot joint motor driving performance model training method, a humanoid robot joint motor driving performance model prediction method and humanoid robot joint motor driving performance model equipment, and belongs to the technical field of robots. The method comprises the following steps: acquiring multi-source data; the multi-source data comprises static parameter sample data, dynamic parameter sample data, driving performance sample data and joint motion mode parameter sample data of the joint motor; establishing a multi-dimensional data set associated with the joint movement mode according to the multi-source data; setting a regression model; training the regression model according to the multi-dimensional data set to obtain driving performance prediction data; and adjusting parameters of the regression model according to the driving performance sample data and the driving performance prediction data to obtain a target model. According to the embodiment of the invention, the problems of low joint motor driving performance measurement efficiency, poor dynamic adaptability and high maintenance cost of the humanoid robot can be solved, and accurate performance prediction and optimal control based on data driving are realized.

Description

Technical Field

[0001] This application relates to the field of robotics technology, and in particular to a training method, prediction method and device for a joint motor drive performance model of a humanoid robot. Background Technology

[0002] As a core driving component, the joint motors of a robot directly affect its motion accuracy, energy efficiency, and reliability. In related technologies, the driving performance of joint motors (such as losses, efficiency, and temperature rise) is usually estimated using real-time measurements or empirical formulas, but this approach has the following problems:

[0003] Measurement efficiency is low: multiple joints of the robot (such as hip joint, knee joint, shoulder joint, etc.) need to be measured independently, and the working state of the motors varies significantly under different movement modes (walking, grasping, jumping), making it time-consuming and labor-intensive to measure them one by one.

[0004] Insufficient dynamic adaptability: When switching motion modes, parameters such as motor load and speed change in real time, and traditional measurement methods cannot respond quickly, resulting in a lag in the drive strategy;

[0005] Data incompleteness: In complex motion scenarios (such as multi-joint coordinated movements), it is difficult to collect comprehensive motor operating data, which affects the system optimization effect;

[0006] High maintenance costs: The lack of accurate performance prediction methods necessitates regular disassembly and measurement, increasing maintenance workload and costs.

[0007] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0008] The main objective of this application is to propose a training method, prediction method, and device for the joint motor drive performance model of a humanoid robot, which can achieve accurate performance prediction in dynamic scenarios and provide data support for robot motion control and energy management.

[0009] To achieve the above objectives, one aspect of this application proposes a method for training a humanoid robot joint motor drive performance model, which includes the following steps:

[0010] Acquire multi-source data; the multi-source data includes static parameter sample data, dynamic parameter sample data, drive performance sample data, and joint motion mode parameter sample data of the joint motor;

[0011] A multidimensional dataset associated with joint movement patterns was established based on the aforementioned multi-source data.

[0012] Set up a regression model;

[0013] The regression model is trained based on the multidimensional dataset to obtain driving performance prediction data;

[0014] The parameters of the regression model are adjusted based on the driving performance sample data and the driving performance prediction data to obtain the target model.

[0015] In some embodiments, acquiring multi-source data includes:

[0016] Obtain sample data of static parameters of the joint motor; the sample data of static parameters includes stator resistance, rotor inductance, and back electromotive force constant;

[0017] Acquire dynamic parameter sample data; the dynamic parameter sample data includes load torque, speed, and phase current;

[0018] Acquire drive performance sample data; the drive performance sample data includes losses, efficiency, and temperature rise;

[0019] Acquire sample data of joint motion pattern parameters; the sample data of joint motion pattern parameters includes walking cadence and joint rotation range.

[0020] The static parameter sample data of the joint motor, the dynamic parameter sample data, the drive performance sample data, and the joint motion mode parameter sample data are used as multi-source data.

[0021] In some embodiments, setting the regression model includes:

[0022] Construct a neural network model, the neural network model including:

[0023] The input layer is used to receive stator resistance, rotor inductance, back EMF constant, load torque, speed, and phase current.

[0024] At least one hidden layer containing 16 neurons;

[0025] The output layer is used to output predicted values ​​for losses, efficiency, and stator temperature.

[0026] The neural network model is set as a regression model.

[0027] In some embodiments, training the regression model based on the multidimensional dataset to obtain driving performance prediction data includes:

[0028] Based on the multidimensional dataset, the static parameters of the joint motor and the sample data of the joint motion mode parameters are input into the regression model to obtain drive performance prediction data.

[0029] In some embodiments, adjusting the parameters of the regression model based on the driving performance sample data and the driving performance prediction data to obtain the target model includes:

[0030] Calculate the prediction error information between the predicted driving performance data and the predicted driving performance sample data;

[0031] When the prediction error exceeds a preset threshold, parameter adjustment and optimization are performed.

[0032] When the prediction error information is less than or equal to the preset threshold, the target model is obtained.

[0033] In some embodiments, the step of performing parameter adjustment and optimization processing when the prediction error information is greater than a preset threshold includes:

[0034] Set a correlation threshold;

[0035] When the prediction error information is greater than a preset threshold, the additive interpretation model is used to calculate the parameter correlation information of each parameter type, and to determine the target parameter type whose impact on prediction accuracy is greater than the correlation threshold.

[0036] The model parameters are optimized based on the static parameter sample data of the target parameter types to obtain the optimized regression model;

[0037] Calculate the prediction error information of the optimized regression model, compare the prediction error information of the optimized regression model with the prediction error information of the regression model before optimization, and obtain the error difference.

[0038] When the error difference is greater than the preset difference threshold, the correlation threshold is updated, and the process returns to the step of determining the type of target parameter. The type of target parameter is re-determined until the error difference is less than or equal to the preset difference threshold, and the target model is obtained.

[0039] To achieve the above objectives, another aspect of this application proposes a method for predicting the joint motor drive performance of a humanoid robot, which includes the following steps:

[0040] Obtain the static parameters and current motion mode parameters of the target joint motor;

[0041] The static parameters and motion mode parameters are input into the target model trained by the method described above to obtain the predicted driving performance data of the target joint motor.

[0042] In some embodiments, the method for predicting the joint motor drive performance of a humanoid robot further includes:

[0043] A motor drive optimization strategy is generated based on the predicted drive performance data of the target joint motor; the motor drive optimization strategy includes PWM duty cycle adjustment, speed control parameter optimization, or load distribution strategy.

[0044] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0045] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0046] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the methods described above.

[0047] The embodiments of this application include at least the following beneficial effects: This application provides a training method, prediction method, and device for a humanoid robot joint motor drive performance model. This solution acquires multi-source data; the multi-source data includes static parameter sample data, dynamic parameter sample data, drive performance sample data, and joint motion mode parameter sample data of the joint motor; a multidimensional dataset associated with the joint motion mode is established based on the multi-source data; a regression model is set; the regression model is trained based on the multidimensional dataset to obtain drive performance prediction data; the parameters of the regression model are adjusted based on the drive performance sample data and the drive performance prediction data to obtain the target model. The embodiments of this application can solve the problems of low measurement efficiency, poor dynamic adaptability, and high maintenance costs of humanoid robot joint motor drive performance, achieving accurate performance prediction and optimized control based on data-driven methods. Attached Figure Description

[0048] Figure 1 This is a flowchart of the training method for the joint motor drive performance model of a humanoid robot provided in the embodiments of this application;

[0049] Figure 2 This is a flowchart of the joint motor drive performance regression model training method provided in the embodiments of this application;

[0050] Figure 3 This is a flowchart of the driving performance prediction and optimization method provided in the embodiments of this application;

[0051] Figure 4 This is a schematic diagram of the neural network model structure provided in the embodiments of this application;

[0052] Figure 5This is a parameter correlation analysis heatmap provided in the embodiments of this application;

[0053] Figure 6 This is a diagram of the architecture of the multi-joint motor drive performance prediction system provided in the embodiments of this application. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0055] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0056] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0058] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0059] 1) PWM (Pulse Width Modulation);

[0060] 2) RMSE (Root Mean Square Error);

[0061] 3) Adam (Adaptive Moment Estimation), an adaptive moment estimation optimizer;

[0062] 4) SHAP (SHapley Additive exPlanations).

[0063] This invention addresses the urgent need for an efficient method for predicting the performance of joint motor drives, reducing reliance on measurement, and achieving accurate performance prediction in dynamic scenarios to provide data support for robot motion control and energy management. This invention provides a method and device for predicting the performance of joint motor drives, involving motor control, machine learning, and robot system optimization technologies. It solves the problems of low measurement efficiency, poor dynamic adaptability, and high maintenance costs of humanoid robot joint motor drive performance, achieving accurate performance prediction and optimized control based on data-driven principles.

[0064] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0065] On one hand, embodiments of the present invention provide a method for training a humanoid robot joint motor drive performance model, referring to... Figure 1 A training method for a humanoid robot joint motor drive performance model includes the following steps:

[0066] Step S100: Obtain multi-source data; the multi-source data includes static parameter sample data, dynamic parameter sample data, drive performance sample data, and joint motion mode parameter sample data of the joint motor;

[0067] Step S200: Establish a multidimensional dataset associated with joint movement patterns based on multi-source data;

[0068] Step S300: Set up the regression model;

[0069] Step S400: Train the regression model based on the multidimensional dataset to obtain the driving performance prediction data;

[0070] Step S500: Adjust the parameters of the regression model based on the driving performance sample data and driving performance prediction data to obtain the target model.

[0071] In some embodiments of the present invention, step S100 of acquiring multi-source data includes:

[0072] Step S110: Obtain sample data of static parameters of the joint motor; the sample data of static parameters includes stator resistance, rotor inductance, and back electromotive force constant.

[0073] Step S120: Obtain dynamic parameter sample data; the dynamic parameter sample data includes load torque, speed, and phase current;

[0074] Step S130: Obtain drive performance sample data; drive performance sample data includes losses, efficiency, and temperature rise;

[0075] Step S140: Obtain sample data of joint motion mode parameters; the sample data of joint motion mode parameters includes walking cadence and joint rotation range;

[0076] Step S150: Use the static parameter sample data, dynamic parameter sample data, drive performance sample data, and joint motion mode parameter sample data of the joint motor as multi-source data.

[0077] As an optional implementation, in the data acquisition process, the present invention displays data quality indicators (such as signal-to-noise ratio > 30dB) through an AR interface; when abnormal data is detected (such as current surge), a voice prompt is triggered for the operator to confirm; invalid data intervals marked by the operator's gestures (such as data from abnormal time periods are removed) are received, and a cleaned dataset is automatically generated.

[0078] In some embodiments of the present invention, step S300 of setting a regression model includes:

[0079] Step S310: Construct a neural network model, which includes:

[0080] The input layer is used to receive stator resistance, rotor inductance, back EMF constant, load torque, speed, and phase current.

[0081] At least one hidden layer containing 16 neurons;

[0082] The output layer is used to output predicted values ​​for losses, efficiency, and stator temperature.

[0083] Step S320: Set up the neural network model as a regression model.

[0084] In some embodiments, step S400 of the present invention, which involves training a regression model based on a multidimensional dataset to obtain driving performance prediction data, includes:

[0085] Step S410: Based on the multidimensional dataset, input the sample data of the static parameters and joint motion mode parameters of the joint motor into the regression model to obtain the drive performance prediction data.

[0086] In some embodiments, step S500, as disclosed in this invention, adjusts the parameters of the regression model based on driving performance sample data and driving performance prediction data to obtain the target model, including:

[0087] Step S510: Calculate the prediction error information between the predicted driving performance data and the predicted driving performance sample data;

[0088] Step S520: When the prediction error information is greater than the preset threshold, perform parameter adjustment and optimization processing;

[0089] Step S530: When the prediction error information is less than or equal to the preset threshold, the target model is obtained.

[0090] In some embodiments, step S520 of the present invention, when the prediction error information is greater than a preset threshold, performs parameter adjustment and optimization processing, including:

[0091] Step S521: Set the correlation threshold;

[0092] Step S522: When the prediction error information is greater than the preset threshold, the additive interpretation model is used to calculate the parameter correlation information of each parameter type, and the target parameter types that have a greater impact on the prediction accuracy than the correlation threshold are determined.

[0093] Step S523: Optimize the model parameters based on the static parameter sample data of the target parameter type to obtain the optimized regression model;

[0094] Step S524: Calculate the prediction error information of the optimized regression model, compare the prediction error information of the optimized regression model with the prediction error information of the regression model before optimization, and obtain the error difference.

[0095] Step S525: When the error difference is greater than the preset difference threshold, update the correlation threshold, return to the step of determining the target parameter type, redetermine the target parameter type, until the error difference is less than or equal to the preset difference threshold, and obtain the target model.

[0096] As an optional implementation, the relevance threshold of this embodiment of the invention is a quantitative standard for screening key input parameters. The relevance threshold is a critical value of the SHAP value. When the contribution of a parameter is greater than the relevance threshold, it is determined to be a target parameter type.

[0097] As an optional implementation, the feature importance ranking generated by SHAP analysis in this embodiment of the invention can be presented to the operator through a visual interface; it can also receive manual adjustment instructions (such as forcibly retaining the load torque parameter), retrain the model, and compare the changes.

[0098] On the other hand, embodiments of the present invention also provide a method for predicting the joint motor drive performance of a humanoid robot, the method comprising the following steps:

[0099] Step S600: Obtain the static parameters and current motion mode parameters of the target joint motor;

[0100] Step S700: Input the static parameters and motion mode parameters into the target model trained by the method described above to obtain the predicted data of the drive performance of the target joint motor.

[0101] In some embodiments, the method for predicting the joint motor drive performance of a humanoid robot disclosed in this invention further includes:

[0102] Step S800: Generate motor drive optimization strategy based on the predicted drive performance data of the target joint motor; the motor drive optimization strategy includes PWM duty cycle adjustment, speed control parameter optimization or load distribution strategy.

[0103] As an optional implementation, the humanoid robot joint motor drive performance prediction method of this embodiment of the invention also includes human-computer interaction technology, which can generate voice summaries (such as "Efficiency is expected to decrease by 5%, it is recommended to check the load"), provide a touch slider, and allow manual fine-tuning of prediction thresholds (such as temperature rise alarm limits). By integrating human-computer interaction technology, the accuracy of response to abnormal operating conditions is improved, and maintenance decision-making time is shortened.

[0104] As an optional implementation, the training method for the joint motor drive performance model of a humanoid robot according to embodiments of the present invention includes:

[0105] Obtain static parameter sample data and corresponding drive performance sample data of the joint motor. The static parameter sample data includes at least two parameter types and is associated with the joint motion mode.

[0106] The static parameter sample data is input into the preset regression model to obtain the drive performance prediction data corresponding to the static parameter sample data. The regression model is established based on the regression relationship between the motor static parameters and drive performance.

[0107] The parameters of the regression model are adjusted based on the driving performance sample data and prediction data to obtain the target regression model.

[0108] Optionally, the static parameter sample data includes motor stator resistance, rotor inductance, back electromotive force constant, and dynamic load parameters related to joint motion modes, while the drive performance sample data includes motor losses, efficiency, and temperature rise data.

[0109] Optionally, the model parameters can be adjusted based on the sample data and the prediction data, including:

[0110] Calculate the prediction error information between the driver performance prediction data and the sample data;

[0111] The parameters of the regression model are adjusted based on the prediction error information until the error information is less than or equal to a preset threshold.

[0112] Optionally, the model parameters are adjusted based on the prediction error information, including:

[0113] An additivity interpretation model is used to calculate the parameter correlation information for each parameter type, and to determine the target parameter types whose impact on prediction accuracy is greater than the correlation threshold.

[0114] The model parameters are optimized based on the static parameter sample data of the target parameter type to obtain the optimized regression model.

[0115] Optionally, optimizing model parameters based on static parameter sample data of the target parameter type also includes:

[0116] Calculate the error difference between the model before and after optimization. When the error difference is greater than the difference threshold, update the correlation threshold and redetermine the types of target parameters until the error difference is less than or equal to the difference threshold. Then, determine the final optimized model as the target regression model.

[0117] As an optional implementation, the method for predicting the joint motor drive performance of a humanoid robot according to embodiments of the present invention includes:

[0118] Obtain the static parameters and current motion mode parameters of the target joint motor;

[0119] The static parameters and motion mode parameters are input into the target regression model trained by the previous method to predict the driving performance data of the motor.

[0120] Motor drive optimization strategies are generated based on drive performance prediction data. These strategies include PWM duty cycle adjustment, speed control parameter optimization, or load distribution strategies.

[0121] As an optional implementation method, refer to Figure 2 The training method (regression model training method) for the joint motor drive performance model of a humanoid robot according to embodiments of the present invention includes:

[0122] 1. Multi-source data acquisition: Collect static parameters (stator resistance, rotor inductance, back EMF constant) and dynamic correlation parameters (load torque, speed, current) of the joint motor under different motion modes, and simultaneously acquire drive performance data (loss, efficiency, temperature rise) to establish a multidimensional dataset associated with joint motion modes (such as walking step frequency, joint rotation range).

[0123] 2. Model Prediction and Optimization: Input static parameters and motion mode parameters into a preset regression model (such as a neural network model), and generate prediction data based on the regression relationship between motor parameters and drive performance; adjust model parameters by calculating prediction errors (such as root mean square error and average relative error), analyze parameter correlation by combining SHAP model, and select key parameters (such as back electromotive force constant and load torque) to optimize the model structure until the prediction error meets the threshold requirements.

[0124] 3. Dynamic parameter adaptation: To address the time-varying characteristics of robot joint motion, motion mode feature parameters (such as joint acceleration and motion period) are introduced to construct a dynamic parameter mapping mechanism, enabling the model to adapt to the performance prediction requirements under different working conditions.

[0125] As an optional implementation method, refer to Figure 3 The method for predicting the joint motor drive performance of a humanoid robot according to an embodiment of the present invention (drive performance prediction method) includes:

[0126] The system acquires real-time static parameters (such as stator resistance and back EMF constant) and current motion mode parameters (such as joint angular velocity and load torque) of the target joint motor, inputs them into the trained target regression model, and predicts motor performance indicators such as loss, efficiency, and temperature rise in real time. Based on the prediction results, it generates drive optimization strategies, such as adjusting the PWM duty cycle to reduce loss, optimizing speed control parameters to improve efficiency, or providing early warning of abnormal temperature rise to trigger maintenance strategies.

[0127] As an optional implementation, embodiments of the present invention also include a method for training a hip joint motor drive performance model:

[0128] Taking the robot's left hip joint servo motor (rated power 200W, rated speed 3000rpm) as an example, the steps include:

[0129] 1. Data Acquisition:

[0130] In three modes—walking (step frequency 1.5Hz), squatting (joint rotation angle 0-120°), and gripping (load 0-5kg)—high-precision sensors are used to measure the motor's static parameters: stator resistance R = 0.8Ω, rotor inductance L = 15mH, and back electromotive force constant Ke = 0.12V·s / rad.

[0131] Synchronously collect dynamic parameters: load torque T (0-30 N·m), speed n (0-2500 rpm), phase current I (0-8 A), and drive performance data: copper loss Pcu, iron loss Pfe, efficiency η, stator temperature Tstator, and construct a dataset containing 300 samples (150 walking samples, 100 squatting samples, and 50 grabbing samples).

[0132] 2. Model building and training:

[0133] refer to Figure 4 A four-layer neural network model was constructed (input layer with 6 nodes: R, L, Ke, T, n, I; hidden layer with 16-8 nodes; output layer with 3 nodes: Pcu, η, Tstator);

[0134] Divide the training set, validation set, and test set into an 8:1:1 ratio, input the training set data, and calculate the RMSE between the predicted value and the actual value (e.g., the initial RMSE for copper loss is 12%).

[0135] The Adam optimization algorithm is used to iteratively adjust the hidden layer weights and bias parameters, and training is stopped when RMSE ≤ 5%.

[0136] refer to Figure 5 The correlation of parameters was analyzed using the SHAP model: back EMF constant (32%), load torque (28%), phase current (22%), stator resistance (10%), speed (6%), and rotor inductance (2%). The correlation threshold was set to 15%, and the target parameters were determined to be back EMF constant, load torque, and phase current.

[0137] 3. Model optimization and validation:

[0138] Based on the target parameters, the model was reconstructed and trained to obtain an optimized model, which reduced the copper loss RMSE to 3.2%, the efficiency prediction error to ≤2%, and the temperature prediction error to ≤3℃.

[0139] In the test set, the predicted and measured values ​​of copper loss in walking mode were 2.8%, the predicted efficiency in squatting mode was 1.5%, and the predicted temperature in grasping mode was 2.3%, all of which meet the requirements for industrial applications.

[0140] As an optional implementation, embodiments of the present invention also include a real-time drive performance prediction and optimization method:

[0141] When the robot performs a climbing task:

[0142] 1. Parameter Acquisition: Real-time acquisition of static parameters of the left hip joint motor: R = 0.82Ω (after temperature compensation), Ke = 0.125V·s / rad; dynamic parameters: load torque T = 25N·m, speed n = 1800rpm, phase current I = 6.5A; motion mode parameters: joint acceleration a = 15° / s². 2 Climbing angle θ = 30°;

[0143] 2. Performance Prediction: Input the target regression model and output the prediction results: copper loss Pcu = 18.5W, efficiency η = 87.2%, stator temperature Tstator = 68℃;

[0144] 3. Strategy optimization:

[0145] Based on predicted losses, the PWM duty cycle was adjusted from 70% to 65%, reducing copper losses to 16.8W and increasing efficiency to 88.5%.

[0146] The predicted temperature triggers the pre-start of the cooling system to stabilize the stator temperature below 65°C and prevent overheating protection shutdown.

[0147] refer to Figure 6 The system architecture diagram of the multi-joint motor drive performance prediction system illustrates the three-layer architecture design of the joint motor drive performance prediction and optimization system, which adopts a modular approach to achieve closed-loop management of the entire process from data acquisition to control optimization.

[0148] 1. Data Acquisition Layer

[0149] Motor operation data is collected in real time using a sensor array;

[0150] After interference suppression by signal conditioning circuitry, a standardized data stream is formed.

[0151] 2. Model layer

[0152] The regression model library integrates algorithms such as neural networks and random forests, and supports parallel training of multiple models.

[0153] The parameter optimization engine dynamically adjusts hyperparameters and combines them with the SHAP analysis module to identify key features (such as the back electromotive force constant Ke).

[0154] 3. Application Layer

[0155] The performance prediction module outputs real-time estimates of losses, efficiency, and temperature rise.

[0156] The optimization strategy generation module automatically adjusts control parameters such as PWM frequency;

[0157] The drive control interface interacts with the robot joint controller in real time, and the status monitoring interface provides visualized data analysis.

[0158] The beneficial effects brought about by the embodiments of the present invention include:

[0159] 1. Improved measurement efficiency: The model prediction replaces a large amount of traditional measurement work, which is especially suitable for multi-joint and multi-condition scenarios, and the measurement time is significantly reduced;

[0160] 2. Dynamic response optimization: Real-time response to changes in motion patterns, prediction latency of less than 50ms, significantly improved speed of drive strategy adjustment, adapting to the agile motion requirements of robots;

[0161] 3. Enhanced energy efficiency and reliability: Drive parameters are optimized based on loss prediction, resulting in improved motor energy efficiency, reduced temperature rise, and extended lifespan of key components;

[0162] 4. Reduced maintenance costs: By predicting and warning of potential faults through temperature rise, unplanned downtime is significantly reduced, resulting in a marked decrease in maintenance costs.

[0163] The robots in the embodiments of this invention may include humanoid robots (humanoid robots), quadrupedal robots, etc.

[0164] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0165] This electronic device can be any intelligent terminal, including motor drivers, human-computer interaction terminals, etc.

[0166] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0167] As an optional implementation, the hardware structure of the electronic device in this embodiment of the invention includes:

[0168] The processor can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to achieve the technical solutions provided in the embodiments of this application.

[0169] The memory can be implemented in the form of read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory and called by the processor to execute the methods described in the embodiments of this application.

[0170] Input / output interfaces are used to implement information input and output;

[0171] The communication interface is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0172] A bus is used to transfer information between various components of a device, such as processors, memory, input / output interfaces, and communication interfaces.

[0173] The processor, memory, input / output interfaces, and communication interfaces communicate with each other within the device via a bus.

[0174] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0175] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0176] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0177] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0178] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0179] The training method, prediction method, electronic device, storage medium, and program product for the joint motor drive performance model of a humanoid robot provided in this application embodiment acquire static parameter sample data and corresponding drive performance sample data of the joint motor, input them into a preset regression model to obtain prediction data, and adjust the model parameters based on the sample data and prediction data to construct a target regression model. Using this model, drive performance indicators, including motor losses, efficiency, and temperature rise, can be accurately predicted based on the motor's static parameters and motion mode parameters, reducing the workload of real-time measurement and improving prediction efficiency. Simultaneously, key influencing parameters are screened through parameter correlation analysis, and the model is optimized in conjunction with joint motion characteristics to achieve dynamic performance prediction and optimized control of the robot's multi-joint motor drive system.

[0180] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0181] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0182] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0183] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0184] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0185] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0186] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0187] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0188] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0189] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0190] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for training a performance model of a humanoid robot's joint motor drive, characterized in that, The training method for the joint motor drive performance model of the humanoid robot includes the following steps: Acquire multi-source data; the multi-source data includes static parameter sample data, dynamic parameter sample data, drive performance sample data, and joint motion mode parameter sample data of the joint motor; A multidimensional dataset associated with joint movement patterns was established based on the aforementioned multi-source data. Set up a regression model; The regression model is trained based on the multidimensional dataset to obtain driving performance prediction data; The parameters of the regression model are adjusted based on the driving performance sample data and the driving performance prediction data to obtain the target model.

2. The method according to claim 1, characterized in that, The acquisition of multi-source data includes: Obtain sample data of static parameters of the joint motor; the sample data of static parameters includes stator resistance, rotor inductance, and back electromotive force constant; Acquire dynamic parameter sample data; the dynamic parameter sample data includes load torque, speed, and phase current; Acquire drive performance sample data; the drive performance sample data includes losses, efficiency, and temperature rise; Acquire sample data of joint motion pattern parameters; the sample data of joint motion pattern parameters includes walking cadence and joint rotation range. The static parameter sample data of the joint motor, the dynamic parameter sample data, the drive performance sample data, and the joint motion mode parameter sample data are used as multi-source data.

3. The method according to claim 1, characterized in that, The setting of the regression model includes: Construct a neural network model, the neural network model including: The input layer is used to receive stator resistance, rotor inductance, back EMF constant, load torque, speed, and phase current. At least one hidden layer containing 16 neurons; The output layer is used to output predicted values ​​for losses, efficiency, and stator temperature. The neural network model is set as a regression model.

4. The method according to claim 1, characterized in that, The step of training the regression model based on the multidimensional dataset to obtain driving performance prediction data includes: Based on the multidimensional dataset, the static parameters of the joint motor and the sample data of the joint motion mode parameters are input into the regression model to obtain drive performance prediction data.

5. The method according to claim 1, characterized in that, The step of adjusting the parameters of the regression model based on the driving performance sample data and the driving performance prediction data to obtain the target model includes: Calculate the prediction error information between the predicted driving performance data and the predicted driving performance sample data; When the prediction error exceeds a preset threshold, parameter adjustment and optimization are performed. When the prediction error information is less than or equal to the preset threshold, the target model is obtained.

6. The method according to claim 5, characterized in that, When the prediction error information is greater than a preset threshold, parameter adjustment and optimization processing is performed, including: Set a correlation threshold; When the prediction error information is greater than a preset threshold, the additive interpretation model is used to calculate the parameter correlation information of each parameter type, and to determine the target parameter type whose impact on prediction accuracy is greater than the correlation threshold. The model parameters are optimized based on the static parameter sample data of the target parameter types to obtain the optimized regression model; Calculate the prediction error information of the optimized regression model, compare the prediction error information of the optimized regression model with the prediction error information of the regression model before optimization, and obtain the error difference. When the error difference is greater than the preset difference threshold, the correlation threshold is updated, and the process returns to the step of determining the type of target parameter. The type of target parameter is re-determined until the error difference is less than or equal to the preset difference threshold, and the target model is obtained.

7. A method for predicting the joint motor drive performance of a humanoid robot, characterized in that, The method for predicting the joint motor drive performance of a humanoid robot includes the following steps: Obtain the static parameters and current motion mode parameters of the target joint motor; The static parameters and motion mode parameters are input into the target model trained by the method described in any one of claims 1-6 to obtain the predicted driving performance data of the target joint motor.

8. The method according to claim 7, characterized in that, The method for predicting the joint motor drive performance of the humanoid robot also includes: A motor drive optimization strategy is generated based on the predicted drive performance data of the target joint motor; the motor drive optimization strategy includes PWM duty cycle adjustment, speed control parameter optimization, or load distribution strategy.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 8.

Images