Harmonic reducer performance adaptation method and device for humanoid robot

Abstract

The application discloses a harmonic reducer performance adaptation method and device for a humanoid robot, and relates to the technical field of robot joint transmission. The method comprises the following steps: constructing a multi-dimensional performance index database of harmonic reducers according to static performance parameters and dynamic performance parameters of various types of harmonic reducers; constructing a load working condition model based on a performance demand vector of a target joint of the humanoid robot in an application scenario; performing adaptation calculation based on the multi-dimensional performance index database and the load working condition model to obtain a comprehensive matching degree score of each candidate harmonic reducer type and the performance demand vector and a load prediction result; obtaining an adaptation recommendation list of a target harmonic reducer type according to the comprehensive matching degree score and the load prediction result; and performing harmonic reducer performance adaptation based on the adaptation recommendation list. In the foregoing manner, the technical problem of improper selection caused by single-dimensional adaptation in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of robot joint transmission technology, and in particular to a method and apparatus for adapting the performance of harmonic reducers for humanoid robots. Background Technology

[0002] As a precision transmission device, harmonic reducers are typically matched using traditional methods that focus only on a single or a few static performance indicators (such as rated torque and transmission accuracy), neglecting dynamic performance under multiple operating conditions (such as start-stop characteristics, torsional stiffness variations, temperature rise effects, and lifespan degradation). This can lead to performance mismatches in actual complex operating conditions. The selection process for harmonic reducers heavily relies on engineers' personal experience and historical data, lacking a systematic and standardized quantitative decision-making model. This method is highly subjective, has poor repeatability, and is too demanding for inexperienced engineers, easily resulting in inappropriate selection.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method and apparatus for performance adaptation of harmonic reducers for humanoid robots, aiming to solve the technical problem of improper selection caused by single-dimensional adaptation in the prior art.

[0005] To achieve the above objectives, this application provides a method for performance adaptation of a harmonic reducer for humanoid robots, the method comprising:

[0006] A multi-dimensional performance index database for harmonic reducers is constructed based on the static and dynamic performance parameters of various types of harmonic reducers.

[0007] Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, a load condition model is constructed.

[0008] Based on the multidimensional performance index database and the load condition model, adaptation calculations are performed to obtain the comprehensive matching score and load prediction results between each candidate harmonic reducer model and the performance requirement vector.

[0009] Based on the comprehensive matching score and the load prediction results, a recommended list of suitable models for the target harmonic reducer is obtained, and the performance of the harmonic reducer is adapted based on the recommended list.

[0010] In one embodiment, after the steps of obtaining a recommended list of suitable target harmonic reducer models based on the comprehensive matching score and the load prediction result, and performing harmonic reducer performance adaptation based on the recommended list, the method further includes:

[0011] Determine the preferred models in the adaptation recommendation list, and predict the expected operating temperature and expected total system energy consumption of the preferred models when the target joint is operating under load conditions in the application scenario.

[0012] The expected operating temperature and the expected total system energy consumption are compared with preset performance thresholds to obtain a decision result.

[0013] When the decision result does not meet the preset conditions, the performance difference is determined, and the performance index weights are optimized based on the performance difference.

[0014] Based on the performance index weights, the multidimensional performance index database, and the load condition model, an optimized adaptation recommendation list is generated.

[0015] In one embodiment, the step of constructing a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers includes:

[0016] Collect and input static performance parameters of each type of harmonic reducer, including rated torque, idle stroke, backlash and transmission error;

[0017] The dynamic performance parameters of various types of harmonic reducers were obtained through dynamic test bench experiments. The dynamic performance parameters include the torsional stiffness matrix under different axial loads, the torque fluctuation spectrum during start-stop process, and the transmission efficiency spectrum under different working conditions.

[0018] The static performance parameters and the dynamic performance parameters are cleaned to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0019] The cleaned static performance parameters and the cleaned dynamic performance parameters are linked and integrated to form the multidimensional performance index database.

[0020] In one embodiment, the step of cleaning the static performance parameters and the dynamic performance parameters to obtain cleaned static performance parameters and cleaned dynamic performance parameters includes:

[0021] Abnormal data points and distorted data caused by acquisition equipment failure are detected and removed from the static performance parameters and dynamic performance parameters to obtain the removed static performance parameters and dynamic performance parameters.

[0022] The removed dynamic performance parameters are smoothed to obtain smoothed dynamic performance parameters;

[0023] The removed static performance parameters and the smoothed dynamic performance parameters are normalized to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0024] In one embodiment, the step of constructing a load condition model based on the performance requirement vector of the humanoid robot target joint in the application scenario includes:

[0025] The dynamic characteristics of the target joint of the humanoid robot in the application scenario are determined, and the dynamic characteristics are analyzed to obtain the working cycle period;

[0026] During the work cycle, the instantaneous torque demand sequence and speed change sequence of the output end of the target joint are obtained, and the core load spectrum is obtained based on the instantaneous torque demand sequence and the speed change sequence.

[0027] Determine the constraints of the target joint in the application environment;

[0028] A performance requirement vector is constructed based on the core load spectrum and the constraints.

[0029] Priority weights are assigned based on the performance requirement vector, and a load condition model is constructed based on the performance requirement vector and the priority weights.

[0030] In one embodiment, the step of determining the dynamic characteristics of the target joint of the humanoid robot in the application scenario, and analyzing the dynamic characteristics to obtain the work cycle includes:

[0031] The dynamic characteristics of the humanoid robot when it completes a specific task in an application scenario are obtained, and the dynamic characteristics include the angle, angular velocity and angular acceleration of the target joint;

[0032] The angular acceleration and load inertia are calculated to obtain the real-time torque requirement data of the target joint;

[0033] Identify and segment typical pattern segments that recur periodically in the angle, angular velocity, and real-time torque demand data;

[0034] A complete typical pattern segment is defined as a working cycle.

[0035] In one embodiment, the step of performing adaptation calculations based on the multidimensional performance index database and the load condition model to obtain the comprehensive matching score and load prediction results between each candidate harmonic reducer model and the performance requirement vector includes:

[0036] Based on the multi-objective decision-making algorithm, the performance demand vector in the load condition model is taken as the objective, and the multi-dimensional performance index of the candidate reducers is taken as the attribute. The comprehensive matching degree score of each candidate reducer relative to the demand vector is calculated.

[0037] The comprehensive matching score is sorted, and a preset number of candidate models corresponding to the top-ranked comprehensive matching scores are determined. The core load spectrum in the load condition model is input into the simplified performance prediction model established for each model. The simplified performance prediction model outputs the efficiency loss and temperature rise trend of each candidate model under the core load spectrum.

[0038] The load prediction result is formed based on the efficiency loss and the temperature rise trend.

[0039] In one embodiment, the step of calculating the comprehensive matching score of each candidate reducer relative to the demand vector based on the multi-objective decision-making algorithm, using the performance demand vector in the load condition model as the objective and the multi-dimensional performance indicators of the candidate reducers as attributes, includes:

[0040] The weight coefficient of each performance index in the scoring is determined based on the priority of the performance requirement vector defined in the load condition model.

[0041] For each candidate reducer model, the actual measured value of the performance index corresponding to each selected reducer model is compared with the corresponding threshold in the performance requirement vector to obtain the comparison result;

[0042] Based on the comparison results, the standardized scores of each performance indicator are determined according to the indicator type of the performance indicator. Specifically, the standardization is carried out by using the upper limit effect measurement function for benefit-type indicators and the lower limit effect measurement function for cost-type indicators, and all performance indicator scores are converted to the same dimensionless scoring interval.

[0043] The standardized scores of each performance indicator of each candidate model are weighted and summed with their corresponding weight coefficients to obtain the comprehensive matching score.

[0044] In one embodiment, the step of generating a load prediction result based on the efficiency loss and the temperature rise trend includes:

[0045] Based on the instantaneous torque and speed in the core load spectrum, the transmission efficiency mapping table of the candidate model harmonic reducer is consulted to obtain the instantaneous transmission efficiency;

[0046] Based on the instantaneous output power and the instantaneous transmission efficiency, the efficiency loss for each time step is calculated;

[0047] The efficiency loss is used as a heat source and input into the equivalent thermal network model of the candidate harmonic reducer to perform transient thermal simulation and predict the temperature rise trend during operation.

[0048] The total energy consumption during the entire working cycle is calculated, along with the highest temperature, average temperature, and temperature fluctuation range in the stated temperature rise trend.

[0049] The total energy consumption, the highest temperature, the average temperature, and the temperature fluctuation range are used as the load prediction results.

[0050] Furthermore, to achieve the above objectives, this application also proposes a harmonic reducer performance adaptation device for humanoid robots, which includes:

[0051] The database construction module is used to build a multi-dimensional performance index database of harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers.

[0052] The load condition modeling module is used to construct a load condition model based on the performance requirement vector of the target joint of the humanoid robot in the application scenario;

[0053] The adaptation calculation module is used to perform adaptation calculations based on the multi-dimensional performance index database and the load condition model to obtain the comprehensive matching degree score and load prediction result of each candidate harmonic reducer model and the performance requirement vector.

[0054] The result recommendation module is used to obtain a matching recommendation list of target harmonic reducer models based on the comprehensive matching score and the load prediction result, and to perform harmonic reducer performance matching based on the matching recommendation list.

[0055] Furthermore, to achieve the above objectives, this application also proposes a harmonic reducer performance adaptation device for humanoid robots. The harmonic reducer performance adaptation device for humanoid robots includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the harmonic reducer performance adaptation method for humanoid robots as described above.

[0056] In addition, to achieve the above objectives, the present invention also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the harmonic reducer performance adaptation method for humanoid robots as described above.

[0057] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the harmonic reducer performance adaptation method for humanoid robots as described above.

[0058] This application provides a method for performance adaptation of harmonic reducers for humanoid robots. It constructs a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various models. Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, a load condition model is constructed. Adaptation calculations are performed based on the multi-dimensional performance index database and the load condition model to obtain a comprehensive matching score and load prediction results for each candidate harmonic reducer model and its performance requirement vector. Based on the comprehensive matching score and load prediction results, a recommended list of suitable harmonic reducer models is obtained, and harmonic reducer performance is adapted based on this list. This method solves the technical problem of improper selection caused by single-dimensional adaptation in existing technologies. Attached Figure Description

[0059] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0060] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 This is a flowchart illustrating an embodiment of the harmonic reducer performance adaptation method for humanoid robots according to this application.

[0062] Figure 2 This is a temperature simulation curve of a harmonic reducer, which is an embodiment of the harmonic reducer performance adaptation method for humanoid robots according to this application.

[0063] Figure 3 This is a schematic diagram of the module structure of the harmonic reducer performance adaptation device for humanoid robots according to an embodiment of this application;

[0064] Figure 4 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the harmonic reducer performance adaptation method for humanoid robots in the embodiments of this application.

[0065] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0066] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0067] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0068] The main solution of this application embodiment is to construct a multi-dimensional performance index database of harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers.

[0069] Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, a load condition model is constructed.

[0070] Based on the multidimensional performance index database and the load condition model, adaptation calculations are performed to obtain the comprehensive matching score and load prediction results between each candidate harmonic reducer model and the performance requirement vector.

[0071] Based on the comprehensive matching score and the load prediction results, a recommended list of suitable models for the target harmonic reducer is obtained, and the performance of the harmonic reducer is adapted based on the recommended list.

[0072] Currently, as a precision transmission device, harmonic reducers are typically matched using traditional methods that focus only on a single or a few static performance indicators (such as rated torque and transmission accuracy), neglecting dynamic performance under multiple operating conditions (such as start-stop characteristics, torsional stiffness variations, temperature rise effects, and lifespan degradation). This can lead to performance mismatches in actual complex operating conditions. The selection process for harmonic reducers heavily relies on engineers' personal experience and historical data, lacking a systematic and standardized quantitative decision-making model. This method is highly subjective, has poor repeatability, and is too demanding for inexperienced engineers, easily leading to inappropriate selection.

[0073] This application provides a solution that constructs a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various models. Based on the performance requirement vector of a humanoid robot target joint in an application scenario, a load condition model is constructed. Adaptation calculations are performed based on the multi-dimensional performance index database and the load condition model to obtain a comprehensive matching score and load prediction results for each candidate harmonic reducer model and its performance requirement vector. Based on the comprehensive matching score and load prediction results, a recommended list of suitable harmonic reducer models is obtained, and harmonic reducer performance is adapted based on this list. This approach solves the technical problem of inappropriate selection caused by single-dimensional adaptation in existing technologies.

[0074] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a harmonic reducer performance adaptation device for humanoid robots. This embodiment does not specifically limit it in this regard. The following uses a harmonic reducer performance adaptation device for humanoid robots as an example to describe this embodiment and the following embodiments.

[0075] All actions involving the acquisition of signals, information, or data in this application are carried out in accordance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the owner of the relevant device.

[0076] This application provides a method for performance adaptation of a harmonic reducer for humanoid robots, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the harmonic reducer performance adaptation method for humanoid robots according to this application.

[0077] In this embodiment, the harmonic reducer performance adaptation method for humanoid robots includes steps S10~S40:

[0078] Step S10: Construct a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers;

[0079] It should be noted that static performance parameters refer to performance indices measured under stable or quasi-steady-state conditions that do not change or change slowly over time. They reflect the basic transmission capacity and accuracy characteristics of the harmonic reducer and typically include rated torque, idle distance, backlash, and transmission error. Dynamic performance parameters, on the other hand, refer to performance indices measured under varying speed and load conditions that change over time. They reveal the dynamic response and energy loss characteristics of the harmonic reducer in actual operation, such as the torsional stiffness matrix under different axial loads, the torque fluctuation spectrum during start-stop processes, and the transmission efficiency spectrum under varying operating conditions.

[0080] Understandably, the first step is to systematically acquire raw performance data for various harmonic reducers by combining data collection from datasheets with experiments on dedicated dynamic test benches. Subsequently, a rigorous data cleaning process must be performed on the raw data, including removing outliers and distorted data caused by sensor malfunctions or external interference, and smoothing and filtering the dynamic data to suppress noise. Next, the cleaned data undergoes normalization preprocessing to eliminate the influence of differences in the units and ranges of different parameters. Finally, the processed static and dynamic parameters are correlated and integrated, and a structured data storage method is used to construct a comprehensive, accurate, and easily queryable multidimensional performance index database, providing a reliable data foundation for subsequent adaptation calculations.

[0081] In one feasible implementation, the step of constructing a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers includes:

[0082] Collect and input static performance parameters of each type of harmonic reducer, including rated torque, idle stroke, backlash and transmission error;

[0083] The dynamic performance parameters of various types of harmonic reducers were obtained through dynamic test bench experiments. The dynamic performance parameters include the torsional stiffness matrix under different axial loads, the torque fluctuation spectrum during start-stop process, and the transmission efficiency spectrum under different working conditions.

[0084] The static performance parameters and the dynamic performance parameters are cleaned to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0085] The cleaned static performance parameters and the cleaned dynamic performance parameters are linked and integrated to form the multidimensional performance index database.

[0086] It should be noted that the rated torque reflects its continuous load-bearing capacity, the idle and backlash characterize the inherent clearances in the transmission chain, and the transmission error reflects the deviation in motion accuracy between the input and output shafts. The torsional stiffness matrix describes its deformation response under combined loads, the torque fluctuation spectrum reveals the smoothness of motion, and the transmission efficiency graph maps the energy conversion efficiency at different operating points.

[0087] In practical implementation, the first step is to systematically collect raw data. Static performance parameters mainly come from product datasheets provided by manufacturers or specialized tests conducted according to national standards, and the accuracy and consistency of the parameter sources must be ensured. Obtaining dynamic performance parameters is more complex, requiring a specially constructed dynamic test bench. By applying axial loads and motion curves that simulate real-world working conditions, and using high-precision sensors to record signals such as torque, speed, and temperature during start-up and shutdown processes and under varying working conditions, key dynamic characteristics such as torsional stiffness matrix, torque fluctuation spectrum, and transmission efficiency spectrum can be extracted through signal analysis techniques.

[0088] After acquiring the raw data, a rigorous data cleaning process is essential. This process aims to identify and remove obvious anomalous data points and distorted data segments caused by sensor transient failures, environmental interference, or data acquisition system malfunctions, and to perform smoothing filtering on dynamic time-series data to suppress random noise. Subsequently, the cleaned static and dynamic datasets are normalized to unify the numerical range and dimensions of each parameter. Finally, database technology is used to associate and structure the verified and preprocessed multi-dimensional parameters, with each harmonic reducer model as the core, forming a complete, reliable, and efficient multi-dimensional performance index database that supports query and analysis.

[0089] In one feasible implementation, the step of performing data cleaning on the static performance parameters and the dynamic performance parameters to obtain cleaned static performance parameters and cleaned dynamic performance parameters includes:

[0090] Abnormal data points and distorted data caused by acquisition equipment failure are detected and removed from the static performance parameters and dynamic performance parameters to obtain the removed static performance parameters and dynamic performance parameters.

[0091] The removed dynamic performance parameters are smoothed to obtain smoothed dynamic performance parameters;

[0092] The removed static performance parameters and the smoothed dynamic performance parameters are normalized to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0093] It should be noted that data cleaning is a crucial preprocessing step for improving the performance data quality of harmonic reducers. Its core task is to identify and remove unreliable data introduced by factors such as measurement errors, equipment transient failures, or external interference, and to perform necessary smoothing and standardization. Outlier data points refer to outliers that significantly deviate from the normal numerical range, while distorted data refers to data recording errors or abrupt changes caused by malfunctions in the acquisition system. Smoothing aims to filter out high-frequency random noise in dynamic performance parameters to reveal their true trend. Normalization linearly transforms parameters with different dimensions and numerical ranges to a unified dimensionless interval, laying the foundation for subsequent multi-index comprehensive evaluation.

[0094] In the specific implementation, statistical methods such as the interquartile range (ICM) or Z-score are first used to automatically detect outliers. Taking the Z-score method as an example, the standard deviation σ is calculated for each data sequence, and data points whose values ​​deviate from the mean μ by more than three times the standard deviation are identified as outliers and removed. For dynamic time-series data, a moving average filter or Savitzky-Golay convolution filter algorithm is then applied for smoothing. The moving average filter formula is: the smoothed data points are equal to the arithmetic mean of the original data within the window. Finally, minimum-maximum normalization is performed to transform each performance parameter X to the range of zero to one; the normalization formula is: After the above steps, both the static parameters and the smoothed dynamic parameters are normalized into clean and comparable datasets.

[0095] Step S20: Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, construct a load condition model;

[0096] It should be noted that the performance requirement vector is a multi-dimensional set of quantitative indicators that systematically describes the performance requirements that a specific joint of a humanoid robot must meet to complete its tasks in a predetermined application scenario. This vector typically includes key parameters such as the peak and continuous operating torque that the joint needs to withstand during movement, the desired motion accuracy and repeatability, and the speed range and acceleration characteristics of the joint. The load condition model, on the other hand, is an abstract mathematical model constructed based on this performance requirement vector. It characterizes the various load conditions, motion states, and dynamic changes faced by the target joint throughout its complete working cycle.

[0097] Understandably, constructing a load condition model first requires in-depth analysis of the typical motion sequences of the target joint in specific application scenarios, such as the swinging of the leg joint during walking or the rotation of the arm joint during grasping. By performing kinematic and dynamic decomposition on these motions, the torque, rotational speed, and inertial load borne by the joint axis at each moment can be quantitatively calculated, thus forming a time-series load spectrum. Combined with the robot's work cycle and task planning, the periodicity, impact characteristics, and thermal load boundary conditions of the load can be further determined. Finally, these discrete load data points and working condition characteristics are mathematically modeled to generate a continuous or discrete working condition model that can comprehensively reflect the actual working load of the joint, providing accurate input conditions for subsequent harmonic reducer selection.

[0098] In one feasible implementation, the step of constructing a load condition model based on the performance requirement vector of the humanoid robot target joint in the application scenario includes:

[0099] The dynamic characteristics of the target joint of the humanoid robot in the application scenario are determined, and the dynamic characteristics are analyzed to obtain the working cycle period;

[0100] During the work cycle, the instantaneous torque demand sequence and speed change sequence of the output end of the target joint are obtained, and the core load spectrum is obtained based on the instantaneous torque demand sequence and the speed change sequence.

[0101] Determine the constraints of the target joint in the application environment;

[0102] A performance requirement vector is constructed based on the core load spectrum and the constraints.

[0103] Priority weights are assigned based on the performance requirement vector, and a load condition model is constructed based on the performance requirement vector and the priority weights.

[0104] It should be noted that the work cycle refers to a complete time series during which a joint repeatedly performs its specific functional task. The core load spectrum is a set of data characterizing the stress state of the joint core, extracted by analyzing the instantaneous torque and speed changes at the joint's output end within this cycle. Constraints define the limiting factors in the joint's working environment, such as dimensions, weight limits, or thermal management requirements. The performance requirement vector is an integrated expression of all the above key performance indicators, while priority weights are used to distinguish the importance of different performance indicators in the overall evaluation.

[0105] In practical implementation, the dynamic characteristics of the joints can be analyzed by performing inverse dynamics calculations on the robot's motion trajectory in the application scenario. Work cycle period. The duration is determined by the typical motion. Within this period, the instantaneous torque sequence τ(t) and rotational speed sequence ω(t) at the joint output are acquired at high frequency through simulation or sensor measurements. Core load spectrum This two-dimensional time-series data quantifies the actual load on the joint. For periodic movements, the load spectrum can be expressed as:

[0106]

[0107] This load spectrum provides the most basic data foundation for subsequent modeling.

[0108] The next step is to construct a performance requirement vector and assign weights. This involves identifying key statistical features from the core load spectrum, such as maximum torque. Root mean square torque Maximum speed And so on, with environmental constraints (such as the maximum allowable size) and quality Together, they form a multi-dimensional performance requirement vector V:

[0109]

[0110] Subsequently, using decision theory tools such as the Analytic Hierarchy Process (AHP), appropriate priority weights are assigned to each indicator in vector V based on the core objectives of the application scenario (e.g., if accuracy is prioritized, transmission error has a higher weight; if lifespan is prioritized, rated torque has a higher weight), forming a weight vector W. The final load condition model M is the weighted requirement description:

[0111]

[0112] This model accurately describes the performance requirements of the target joint and the relative importance of each requirement, providing a clear basis for the matching and selection of the reducer.

[0113] In one feasible implementation, the step of determining the dynamic characteristics of the target joint of the humanoid robot in the application scenario, and analyzing the dynamic characteristics to obtain the work cycle includes:

[0114] The dynamic characteristics of the humanoid robot when it completes a specific task in an application scenario are obtained, and the dynamic characteristics include the angle, angular velocity and angular acceleration of the target joint;

[0115] The angular acceleration and load inertia are calculated to obtain the real-time torque requirement data of the target joint;

[0116] Identify and segment typical pattern segments that recur periodically in the angle, angular velocity, and real-time torque demand data;

[0117] A complete typical pattern segment is defined as a working cycle.

[0118] In practical implementation, the first step relies on acquiring high-precision joint motion data. This data can be obtained through inverse dynamics calculations after kinematic modeling of the entire robot using multibody dynamics simulation software, or through actual measurements using encoders and torque sensors mounted on the joints of the physical robot. The key data acquired is the sequence of angle θt, angular velocity ωt, and angular acceleration αt as a function of time t. According to the rotational form of Newton's second law, the real-time torque requirement τt of the joint can be calculated by multiplying the moment of inertia J of the load by the measured angular acceleration αt.

[0119]

[0120] This formula establishes a direct relationship between the motion state and the driving torque, thereby obtaining the torque demand time series data.

[0121] After obtaining the complete time-series data, it is necessary to identify the periodic patterns. This process typically employs signal processing and pattern recognition algorithms. For example, the autocorrelation function of the angle sequence θt can be calculated, and its peak position indicates the underlying period. Alternatively, algorithms such as dynamic time warping can be used to match long-term time-series data with a pre-defined typical action template, thereby accurately segmenting recurring typical pattern segments. Once the start and end points of these segments are identified, an independent segment containing a complete action, such as starting from a standing posture, completing a step, and returning to a standing posture, can be defined as a standard work cycle. This period will serve as the baseline time window for all subsequent load analyses.

[0122] Step S30: Based on the multidimensional performance index database and the load condition model, perform adaptation calculations to obtain the comprehensive matching score and load prediction results of each candidate harmonic reducer model with the performance requirement vector.

[0123] It should be noted that the adaptation calculation is a multi-criteria decision-making process. Its core is to systematically compare and analyze the target joint performance requirements represented by the load condition model with the inherent technical parameters of each candidate harmonic reducer model in the multi-dimensional performance index database. The comprehensive matching score is a quantitative evaluation value that comprehensively reflects the degree to which a candidate model meets the requirements in various performance indicators. The load prediction result is a forward-looking estimate of key states such as lifespan, temperature rise, and efficiency under specified load conditions, based on the parameters of the selected reducer model.

[0124] Understandably, the first step is to establish an evaluation model. For each dimension of the performance requirement vector in the load condition model, such as rated torque, stiffness, and accuracy, a matching function is defined. This function compares the corresponding parameters of the candidate model with the required values ​​and outputs a sub-score. Next, using the preset priority weights in the load condition model, the sub-scores of each dimension are weighted and summed to calculate the comprehensive matching score of the candidate model. For load prediction, the core load spectrum in the load condition model needs to be input into simulation models such as life models and thermal models based on the physical characteristics of the reducer. After running the calculations, the performance degradation curves and reliability indicators under different application scenarios are obtained, ultimately forming a comprehensive load prediction report.

[0125] In one feasible implementation, the step of performing adaptation calculations based on the multidimensional performance index database and the load condition model to obtain the comprehensive matching score and load prediction results between each candidate harmonic reducer model and the performance requirement vector includes:

[0126] Based on the multi-objective decision-making algorithm, the performance demand vector in the load condition model is taken as the objective, and the multi-dimensional performance index of the candidate reducers is taken as the attribute. The comprehensive matching degree score of each candidate reducer relative to the demand vector is calculated.

[0127] The comprehensive matching score is sorted, and a preset number of candidate models corresponding to the top-ranked comprehensive matching scores are determined. The core load spectrum in the load condition model is input into the simplified performance prediction model established for each model. The simplified performance prediction model outputs the efficiency loss and temperature rise trend of each candidate model under the core load spectrum.

[0128] The load prediction result is formed based on the efficiency loss and the temperature rise trend.

[0129] Understandably, the adaptation calculation can be viewed as a two-stage process. The first stage employs a multi-objective decision-making algorithm, using the performance requirement vector defined in the load condition model as the evaluation criterion and the technical parameters of each candidate harmonic reducer model in the multi-dimensional performance index database as the evaluation object. A quantitative comprehensive matching score is calculated mathematically to measure the degree to which the candidate model meets the overall requirements. The second stage involves more in-depth performance simulation of the initially selected high-scoring candidate models. This is achieved by inputting the core load spectrum from the load condition model into a simplified performance prediction model specifically built for each model, simulating its dynamic response under real-world operating conditions, and outputting predicted values ​​for key performance indicators. These primarily include energy loss during transmission (efficiency loss) and temperature rise due to energy loss (temperature rise trend).

[0130] In practical implementations, the matching score is typically calculated using multi-objective decision-making algorithms such as the weighted product method or TOPSIS. First, each indicator in the performance requirement vector needs to be normalized to eliminate the influence of dimensions. Let the value of candidate model i on the j-th performance indicator be... Its corresponding target demand value is The priority weight is A common method is to calculate the matching score of the model on various indicators. After that, a comprehensive matching score was calculated. It can be obtained through a weighted geometric mean or a weighted arithmetic mean model. For example, the calculation formula using the weighted product model is:

[0131]

[0132] This model amplifies the weakest link effect, meaning that a low match in any indicator will significantly lower the overall score, which helps in selecting models with balanced performance. Calculate the scores for all candidate models. Then, you can sort them.

[0133] For the top-ranked candidate models, load prediction needs to be initiated. The key technology lies in constructing a simplified performance prediction model. The efficiency loss model is typically based on the reducer's friction characteristics and load-efficiency curve, using the instantaneous torque τ and speed ω from the core load spectrum as inputs, and the instantaneous power loss... It can be approximated as:

[0134]

[0135] in It is an efficiency mapping function fitted based on sample data. The temperature rise trend is calculated using a thermal network model, which includes power loss. As a heat source, considering heat dissipation conditions, the differential equation is solved to obtain the temperature change curve ΔTt over time. Ultimately, the efficiency loss and temperature rise trend output by the model together constitute the load prediction result for a specific model under specific operating conditions.

[0136] In one feasible implementation, the step of calculating the comprehensive matching score of each candidate reducer relative to the demand vector based on the multi-objective decision-making algorithm, using the performance demand vector in the load condition model as the objective and the multi-dimensional performance indicators of the candidate reducers as attributes, includes:

[0137] The weight coefficient of each performance index in the scoring is determined based on the priority of the performance requirement vector defined in the load condition model.

[0138] For each candidate reducer model, the actual measured value of the performance index corresponding to each selected reducer model is compared with the corresponding threshold in the performance requirement vector to obtain the comparison result;

[0139] Based on the comparison results, the standardized scores of each performance indicator are determined according to the indicator type of the performance indicator. Specifically, the standardization is carried out by using the upper limit effect measurement function for benefit-type indicators and the lower limit effect measurement function for cost-type indicators, and all performance indicator scores are converted to the same dimensionless scoring interval.

[0140] The standardized scores of each performance indicator of each candidate model are weighted and summed with their corresponding weight coefficients to obtain the comprehensive matching score.

[0141] Understandably, weighting coefficients are used to characterize the relative importance of various performance indicators in the load condition model, such as torque, accuracy, and lifespan, in the overall evaluation. The comparison results are obtained by directly comparing the actual performance parameters of candidate reducer models with the required threshold. Normalized scoring uses mathematical functions to map comparison results of different properties and dimensions to a standardized, dimensionless numerical range, making different types of indicators comparable. For benefit-type indicators, a higher index value is better, while for cost-type indicators, a lower index value is better.

[0142] In practical implementation, determining the weighting coefficients often employs the analytic hierarchy process (AHP) or direct assignment to ensure that the sum of all weights is one. Subsequent data normalization is crucial for eliminating the influence of dimensions. Let the required threshold for the j-th performance indicator be... The actual value of candidate model i on this metric is For benefit-oriented indicators, such as rated torque or efficiency, an upper limit effect measurement function is used, and its standardized score is determined. The calculation formula is:

[0143]

[0144] This formula ensures that the score is no lower than one when the actual value exceeds or reaches the threshold; the higher the value, the higher the score. For cost-related indicators, such as backlash or weight, a lower limit effect measure function is used, and its normalized scoring formula is as follows:

[0145]

[0146] This formula ensures that the score is not lower than one when the actual value is below or reaches the threshold; the smaller the value, the higher the score. Through these two functions, the scores of all indicators are converted to the range of zero to positive infinity, and the larger the value, the better the match.

[0147] After obtaining the normalized scores of all performance indicators, a comprehensive calculation can be performed. A linear weighted sum model is used to normalize the scores of each indicator for each candidate model i. Its corresponding weighting coefficient After multiplying and summing, the overall matching score for this model is obtained. The calculation formula is expressed as follows:

[0148]

[0149] The final result It is a dimensionless scalar value. A higher value indicates a better match between the candidate model's overall performance attributes and the target demand vector. All candidate models can be calculated using this method. This provides a direct basis for subsequent sorting and filtering.

[0150] In one feasible implementation, the step of generating a load prediction result based on the efficiency loss and the temperature rise trend includes:

[0151] Based on the instantaneous torque and speed in the core load spectrum, the transmission efficiency mapping table of the candidate model harmonic reducer is consulted to obtain the instantaneous transmission efficiency;

[0152] Based on the instantaneous output power and the instantaneous transmission efficiency, the efficiency loss for each time step is calculated;

[0153] The efficiency loss is used as a heat source and input into the equivalent thermal network model of the candidate harmonic reducer to perform transient thermal simulation and predict the temperature rise trend during operation.

[0154] The total energy consumption during the entire working cycle is calculated, along with the highest temperature, average temperature, and temperature fluctuation range in the stated temperature rise trend.

[0155] The total energy consumption, the highest temperature, the average temperature, and the temperature fluctuation range are used as the load prediction results.

[0156] It should be noted that instantaneous transmission efficiency refers to the energy transfer efficiency of a harmonic reducer at a specific torque and speed under dynamic operating conditions described by the core load spectrum. Efficiency loss, on the other hand, is the energy loss power calculated based on instantaneous output power and instantaneous efficiency. The equivalent thermal network model simplifies the complex physical heat dissipation process into a circuit model composed of thermal resistance and thermal capacity, used to simulate the dynamic process of heat generation and dissipation. The final form of the load prediction result is a set of quantitative indicators, including the total energy consumed throughout the entire operating cycle, the highest representative temperature during the temperature rise process, the average temperature reflecting the overall heat load, and the temperature fluctuation amplitude reflecting the severity of the thermal cycle.

[0157] In its implementation, the core load spectrum provides instantaneous torque in the form of a time series. and rotational speed First, query the transmission efficiency mapping table for this candidate model, which has been pre-calibrated through experiments. This table uses torque and speed as indexes, and the instantaneous transmission efficiency at the corresponding operating point can be obtained by interpolation. Instantaneous output power From the formula The calculation yielded the following result. Subsequently, the efficiency loss within each time step Δt, i.e., the power loss converted into heat, was calculated. It can be done through the formula or equivalent formula Accurate calculations show that this portion of energy is the main heat source causing the temperature rise.

[0158] Obtain continuous heat source power Then, this is used as input excitation and loaded into the equivalent thermal network model of the candidate model. This model is parameterized by thermal resistance and heat capacity. By solving the transient thermal equilibrium differential equation, the temperature change over time at key points inside the reducer (such as the flexspline) can be numerically simulated. . Reference Figure 2 , Figure 2 This is a temperature simulation curve for the harmonic reducer. After the simulation, a statistical analysis of the entire process is performed: total energy consumption. It is obtained by integrating the output power, i.e. From the temperature rise curve The highest temperature can be extracted from it. Average temperature and the temperature fluctuation amplitude ΔT (i.e. and (The difference). Ultimately, the vector As a quantitative load forecast output, it provides a key basis for the final decision.

[0159] Step S40: Based on the comprehensive matching score and the load prediction result, obtain the recommended list of suitable models for the target harmonic reducer, and perform harmonic reducer performance adaptation based on the recommended list.

[0160] It's important to note that the various metrics in the load forecast results must first be normalized, as total energy consumption and temperature have different dimensions and physical meanings. After normalization, appropriate weights are assigned to these metrics; for example, more attention can be paid to temperature rise to prioritize reliability, or more attention can be paid to energy consumption to optimize energy efficiency. Next, the normalized load forecast metrics are combined with the overall matching score using weighted averages or constraint rules. A common strategy is to use threshold filtering, such as excluding models whose maximum temperature exceeds the material's allowable limit, and then sorting the remaining models according to the overall score. The final generated matching recommendation list clearly shows the optimal and second-best choices, directly guiding the final procurement or design decisions and achieving accurate performance matching.

[0161] In one feasible implementation, after the steps of obtaining a recommended list of suitable target harmonic reducer models based on the comprehensive matching score and the load prediction result, and performing harmonic reducer performance adaptation based on the recommended list, the method further includes:

[0162] Determine the preferred models in the adaptation recommendation list, and predict the expected operating temperature and expected total system energy consumption of the preferred models when the target joint is operating under load conditions in the application scenario.

[0163] The expected operating temperature and the expected total system energy consumption are compared with preset performance thresholds to obtain a decision result.

[0164] When the decision result does not meet the preset conditions, the performance difference is determined, and the performance index weights are optimized based on the performance difference.

[0165] Based on the performance index weights, the multidimensional performance index database, and the load condition model, an optimized adaptation recommendation list is generated.

[0166] In its implementation, the system first extracts the top-performing models from the recommended list and substitutes their load condition models into a more refined system-level simulation model to predict accurate expected operating temperatures and total system energy consumption. Then, these two key predicted values ​​are compared with preset performance thresholds. If the threshold requirements are not met, the difference is calculated, and based on the magnitude of the difference, methods such as weighted least squares or heuristic rules are used to automatically increase the weights of basic performance items strongly correlated with the unmet targets. For example, if the temperature exceeds the limit, the weights of transmission efficiency or heat dissipation performance are increased. After weight optimization, the system backtracks to the adaptation calculation module, re-evaluates all candidate models using the new weight coefficients, and finally outputs a more reliable optimized adaptation recommendation list corrected for practical objectives.

[0167] This embodiment provides a method for performance adaptation of harmonic reducers for humanoid robots. It constructs a multi-dimensional performance index database for harmonic reducers based on the static and dynamic performance parameters of various models. Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, a load condition model is constructed. Adaptation calculations are performed based on the multi-dimensional performance index database and the load condition model to obtain a comprehensive matching score and load prediction results for each candidate harmonic reducer model and its performance requirement vector. Based on the comprehensive matching score and load prediction results, a recommended list of suitable harmonic reducer models is obtained, and harmonic reducer performance is adapted based on this list. This method solves the technical problem of improper selection caused by single-dimensional adaptation in existing technologies.

[0168] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the harmonic reducer performance adaptation method for humanoid robots. Any simple modifications based on this technical concept are within the protection scope of this application.

[0169] This application also provides a harmonic reducer performance adaptation device for humanoid robots; please refer to... Figure 3 The performance adaptation device for harmonic reducers used in humanoid robots includes:

[0170] Database construction module 10 is used to construct a multi-dimensional performance index database of harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers.

[0171] The working condition modeling module 20 is used to construct a load working condition model based on the performance requirement vector of the target joint of the humanoid robot in the application scenario.

[0172] The adaptation calculation module 30 is used to perform adaptation calculations based on the multi-dimensional performance index database and the load condition model to obtain the comprehensive matching degree score and load prediction result of each candidate harmonic reducer model and the performance requirement vector.

[0173] The result recommendation module 40 is used to obtain a matching recommendation list of the target harmonic reducer model based on the comprehensive matching score and the load prediction result, and to perform harmonic reducer performance matching based on the matching recommendation list.

[0174] In one feasible implementation, the result recommendation module 40 is further configured to determine the preferred model in the adaptation recommendation list, and predict the expected operating temperature and expected total system energy consumption of the preferred model when the target joint is operating under load conditions in the application scenario.

[0175] The expected operating temperature and the expected total system energy consumption are compared with preset performance thresholds to obtain a decision result.

[0176] When the decision result does not meet the preset conditions, the performance difference is determined, and the performance index weights are optimized based on the performance difference.

[0177] Based on the performance index weights, the multidimensional performance index database, and the load condition model, an optimized adaptation recommendation list is generated.

[0178] In one feasible implementation, the database construction module 10 is also used to collect and input the static performance parameters of each type of harmonic reducer, the static performance parameters including rated torque, idle stroke, backlash and transmission error;

[0179] The dynamic performance parameters of various types of harmonic reducers were obtained through dynamic test bench experiments. The dynamic performance parameters include the torsional stiffness matrix under different axial loads, the torque fluctuation spectrum during start-stop process, and the transmission efficiency spectrum under different working conditions.

[0180] The static performance parameters and the dynamic performance parameters are cleaned to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0181] The cleaned static performance parameters and the cleaned dynamic performance parameters are linked and integrated to form the multidimensional performance index database.

[0182] In one feasible implementation, the database construction module 10 is further configured to detect and remove abnormal data points and distorted data caused by acquisition device failure from the static performance parameters and the dynamic performance parameters, so as to obtain the removed static performance parameters and the removed dynamic performance parameters.

[0183] The removed dynamic performance parameters are smoothed to obtain smoothed dynamic performance parameters;

[0184] The removed static performance parameters and the smoothed dynamic performance parameters are normalized to obtain the cleaned static performance parameters and the cleaned dynamic performance parameters.

[0185] In one feasible implementation, the working condition modeling module 20 is further used to determine the dynamic characteristics of the humanoid robot target joint in the application scenario, analyze the dynamic characteristics, and obtain the working cycle period.

[0186] During the work cycle, the instantaneous torque demand sequence and speed change sequence of the output end of the target joint are obtained, and the core load spectrum is obtained based on the instantaneous torque demand sequence and the speed change sequence.

[0187] Determine the constraints of the target joint in the application environment;

[0188] A performance requirement vector is constructed based on the core load spectrum and the constraints.

[0189] Priority weights are assigned based on the performance requirement vector, and a load condition model is constructed based on the performance requirement vector and the priority weights.

[0190] In one feasible implementation, the working condition modeling module 20 is further used to acquire the dynamic characteristics of the humanoid robot when it completes a specific task in an application scenario, the dynamic characteristics including the angle, angular velocity and angular acceleration of the target joint;

[0191] The angular acceleration and load inertia are calculated to obtain the real-time torque requirement data of the target joint;

[0192] Identify and segment typical pattern segments that recur periodically in the angle, angular velocity, and real-time torque demand data;

[0193] A complete typical pattern segment is defined as a working cycle.

[0194] In one feasible implementation, the adaptation calculation module 30 is further configured to, based on a multi-objective decision-making algorithm, take the performance demand vector in the load condition model as the objective and the multi-dimensional performance indicators of the candidate reducers as attributes, and calculate the comprehensive matching degree score of each candidate reducer relative to the demand vector.

[0195] The comprehensive matching score is sorted, and a preset number of candidate models corresponding to the top-ranked comprehensive matching scores are determined. The core load spectrum in the load condition model is input into the simplified performance prediction model established for each model. The simplified performance prediction model outputs the efficiency loss and temperature rise trend of each candidate model under the core load spectrum.

[0196] The load prediction result is formed based on the efficiency loss and the temperature rise trend.

[0197] In one feasible implementation, the adaptation calculation module 30 is further configured to determine the weight coefficient of each performance index in the scoring based on the priority of the performance requirement vector defined in the load condition model.

[0198] For each candidate reducer model, the actual measured value of the performance index corresponding to each selected reducer model is compared with the corresponding threshold in the performance requirement vector to obtain the comparison result;

[0199] Based on the comparison results, the standardized scores of each performance indicator are determined according to the indicator type of the performance indicator. Specifically, the standardization is carried out by using the upper limit effect measurement function for benefit-type indicators and the lower limit effect measurement function for cost-type indicators, and all performance indicator scores are converted to the same dimensionless scoring interval.

[0200] The standardized scores of each performance indicator of each candidate model are weighted and summed with their corresponding weight coefficients to obtain the comprehensive matching score.

[0201] In one feasible implementation, the adaptation calculation module 30 is further configured to query the transmission efficiency mapping table of the candidate model harmonic reducer based on the instantaneous torque and speed in the core load spectrum, and obtain the instantaneous transmission efficiency.

[0202] Based on the instantaneous output power and the instantaneous transmission efficiency, the efficiency loss for each time step is calculated;

[0203] The efficiency loss is used as a heat source and input into the equivalent thermal network model of the candidate harmonic reducer to perform transient thermal simulation and predict the temperature rise trend during operation.

[0204] The total energy consumption during the entire working cycle is calculated, along with the highest temperature, average temperature, and temperature fluctuation range in the stated temperature rise trend.

[0205] The total energy consumption, the highest temperature, the average temperature, and the temperature fluctuation range are used as the load prediction results.

[0206] The harmonic reducer performance adaptation device for humanoid robots provided in this application adopts the harmonic reducer performance adaptation method for humanoid robots in the above embodiments, which can solve the technical problem of improper selection caused by single-dimensional adaptation. Compared with the prior art, the beneficial effects of the harmonic reducer performance adaptation device for humanoid robots provided in this application are the same as the beneficial effects of the harmonic reducer performance adaptation method for humanoid robots provided in the above embodiments, and other technical features in the harmonic reducer performance adaptation device for humanoid robots are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0207] This application provides a harmonic reducer performance adaptation device for humanoid robots. The harmonic reducer performance adaptation device for humanoid robots includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the harmonic reducer performance adaptation method for humanoid robots in the first embodiment described above.

[0208] The following is for reference. Figure 4 This document illustrates a structural schematic diagram of a harmonic reducer performance adaptation device suitable for implementing embodiments of this application for humanoid robots. The harmonic reducer performance adaptation device for humanoid robots in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 4 The harmonic reducer performance adaptation device for humanoid robots shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0209] like Figure 4 As shown, the harmonic reducer performance adaptation device for a humanoid robot may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the harmonic reducer performance adaptation device for the humanoid robot. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the harmonic reducer performance adapter for humanoid robots to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a harmonic reducer performance adapter for humanoid robots with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0210] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0211] The harmonic reducer performance adaptation device for humanoid robots provided in this application adopts the harmonic reducer performance adaptation method for humanoid robots in the above embodiments, and can solve the technical problem of harmonic reducer performance adaptation for humanoid robots. Compared with the prior art, the beneficial effects of the harmonic reducer performance adaptation device for humanoid robots provided in this application are the same as the beneficial effects of the harmonic reducer performance adaptation method for humanoid robots provided in the above embodiments, and other technical features in the harmonic reducer performance adaptation device for humanoid robots are the same as the features disclosed in the method of the previous embodiment, and will not be repeated here.

[0212] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0213] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0214] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the harmonic reducer performance adaptation method for humanoid robots described in the above embodiments.

[0215] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0216] The aforementioned computer-readable storage medium may be included in the harmonic reducer performance adaptation device for humanoid robots; or it may exist independently and not assembled into the harmonic reducer performance adaptation device for humanoid robots.

[0217] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the aforementioned one or more programs in the harmonic reducer performance adaptation device for humanoid robots, enable the harmonic reducer performance adaptation device for humanoid robots to: construct a multi-dimensional performance index database of harmonic reducers based on the static and dynamic performance parameters of various types of harmonic reducers.

[0218] Based on the performance requirement vector of the target joint of the humanoid robot in the application scenario, a load condition model is constructed.

[0219] Based on the multidimensional performance index database and the load condition model, adaptation calculations are performed to obtain the comprehensive matching score and load prediction results between each candidate harmonic reducer model and the performance requirement vector.

[0220] Based on the comprehensive matching score and the load prediction results, a recommended list of suitable models for the target harmonic reducer is obtained, and the performance of the harmonic reducer is adapted based on the recommended list.

[0221] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0222] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0223] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0224] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described harmonic reducer performance adaptation method for humanoid robots, thereby solving the technical problem of harmonic reducer performance adaptation for humanoid robots. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the harmonic reducer performance adaptation method for humanoid robots provided in the above embodiments, and will not be repeated here.

[0225] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described harmonic reducer performance adaptation method for humanoid robots.

[0226] The computer program product provided in this application can solve the technical problem of performance adaptation of harmonic reducers for humanoid robots. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the harmonic reducer performance adaptation method for humanoid robots provided in the above embodiments, and will not be repeated here.

[0227] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A harmonic reducer performance adaptation method for a humanoid robot, characterized by, The harmonic reducer performance adaptation method for the humanoid robot comprises: According to the static performance parameters and dynamic performance parameters of the harmonic reducers of multiple types, a multi-dimensional performance index database of the harmonic reducers is constructed; Based on the performance demand vector of the target joint of the humanoid robot in the application scenario, a load working condition model is constructed; Based on the multi-dimensional performance index database and the load working condition model, adaptation calculation is performed to obtain a comprehensive matching degree score of each candidate harmonic reducer type and the performance demand vector and a load prediction result; According to the comprehensive matching degree score and the load prediction result, an adaptation recommendation list of the target harmonic reducer type is obtained, and harmonic reducer performance adaptation is performed based on the adaptation recommendation list; After the step of obtaining the adaptation recommendation list of the target harmonic reducer type according to the comprehensive matching degree score and the load prediction result and performing harmonic reducer performance adaptation based on the adaptation recommendation list, the following steps are further included: A preferred type in the adaptation recommendation list is determined, and the expected operating temperature and the expected system total energy consumption of the preferred type when the target joint operates in the load working condition in the application scenario are respectively predicted; The expected operating temperature and the expected system total energy consumption are respectively compared with a preset performance threshold to obtain a decision result; When the decision result does not satisfy a preset condition, a performance difference value is determined, and the performance index weight is optimized according to the performance difference value; According to the performance index weight, the multi-dimensional performance index database and the load working condition model, adaptation calculation is performed to generate an optimized adaptation recommendation list.

2. The method of claim 1, wherein, The step of constructing the multi-dimensional performance index database of the harmonic reducers according to the static performance parameters and dynamic performance parameters of the harmonic reducers of multiple types comprises: Static performance parameters of each type of harmonic reducer are collected and input, and the static performance parameters include rated torque, free travel, backlash and transmission error; Dynamic performance parameters of each type of harmonic reducer obtained through dynamic test bench experiments are acquired, and the dynamic performance parameters include torsional stiffness matrix under different axial loads, torque fluctuation frequency spectrum in start-stop process and transmission efficiency spectrum under variable working conditions; Data cleaning is performed on the static performance parameters and the dynamic performance parameters to obtain cleaned static performance parameters and cleaned dynamic performance parameters; The cleaned static performance parameters and the cleaned dynamic performance parameters are associated and integrated to form the multi-dimensional performance index database.

3. The method of claim 2, wherein, The step of performing data cleaning on the static performance parameters and the dynamic performance parameters to obtain cleaned static performance parameters and cleaned dynamic performance parameters comprises: Abnormal data points and distorted data caused by faults of collection equipment in the static performance parameters and the dynamic performance parameters are detected and removed to obtain removed static performance parameters and removed dynamic performance parameters; The removed dynamic performance parameters are smoothed to obtain smoothed dynamic performance parameters; The removed static performance parameters and the smoothed dynamic performance parameters are normalized to obtain cleaned static performance parameters and cleaned dynamic performance parameters.

4. The method of claim 1, wherein, The performance demand vector of the target joint of the humanoid robot in the application scene, and the step of constructing the load working condition model comprises: determining the dynamics characteristics of the target joint of the humanoid robot in the application scene, and analyzing the dynamics characteristics to obtain a working cycle period; in the working cycle period, obtaining a transient torque demand sequence and a rotational speed change sequence of the output end of the target joint, and obtaining a core load spectrum based on the transient torque demand sequence and the rotational speed change sequence; determining the constraint conditions of the target joint in the application environment; constructing a performance demand vector based on the core load spectrum and the constraint conditions; based on the performance demand vector, assigning a priority weight, and constructing a load working condition model according to the performance demand vector and the priority weight.

5. The method of claim 4, wherein, The step of determining the dynamics characteristics of the target joint of the humanoid robot in the application scene, and analyzing the dynamics characteristics to obtain a working cycle period comprises: obtaining the dynamics characteristics of the humanoid robot when completing a specific task in the application scene, the dynamics characteristics including the angle, angular velocity and angular acceleration of the target joint; calculating the angular acceleration and the load inertia to obtain real-time torque demand data of the target joint; identifying and segmenting a typical mode segment that periodically appears in the angle, angular velocity and real-time torque demand data; defining a complete typical mode segment as a working cycle period.

6. The method of claim 1, wherein, The step of performing adaptive calculation based on the multi-dimensional performance index database and the load working condition model to obtain a comprehensive matching degree score of each candidate harmonic reducer model and the performance demand vector and a load prediction result comprises: based on a multi-objective decision algorithm, taking the performance demand vector in the load working condition model as a target and taking the multi-dimensional performance index of the candidate reducer as an attribute, calculating the comprehensive matching degree score of each candidate reducer relative to the demand vector; sorting the comprehensive matching degree scores to determine the candidate models corresponding to the top pre-set number of comprehensive matching degree scores, inputting the core load spectrum in the load working condition model into a simplified performance prediction model established for each model, and the simplified performance prediction model outputs the efficiency loss and temperature rise trend of each candidate model under the core load spectrum; forming a load prediction result according to the efficiency loss and the temperature rise trend.

7. The method of claim 6, wherein, The step of performing adaptive calculation based on the multi-dimensional performance index database and the load working condition model to obtain a comprehensive matching degree score of each candidate harmonic reducer model and the performance demand vector and a load prediction result comprises: determining the weight coefficients of each performance index in the score according to the priority of the performance demand vector in the load working condition model; for each candidate reducer model, comparing the actual measured value of the performance index corresponding to each candidate reducer model with the corresponding threshold value in the performance demand vector to obtain a comparison result; According to the comparison result, the normalized scores of each performance index are determined based on the index type of the performance index, specifically including: for benefit type index, using upper limit effect measure function for normalization, for cost type index, using lower limit effect measure function for normalization, and converting all performance index scores to the same dimensionless score interval; The normalized scores of each performance index of each candidate model are weighted and summed with the corresponding weight coefficients to obtain a comprehensive matching degree score.

8. The method of claim 6, wherein, The step of forming a load prediction result according to the efficiency loss and the temperature rise trend includes: According to the instantaneous torque and speed in the core load spectrum, the transmission efficiency mapping table of the candidate model harmonic reducer is queried to obtain the instantaneous transmission efficiency; Based on the instantaneous output power and the instantaneous transmission efficiency, the efficiency loss of each time step is calculated; The efficiency loss is input into the equivalent thermal network model of the candidate model harmonic reducer as a heat source to perform transient thermal simulation and predict the temperature rise trend during operation; The total energy consumption in the entire working cycle period, the maximum temperature, the average temperature and the temperature fluctuation amplitude in the temperature rise trend are counted; The total energy consumption, the maximum temperature, the average temperature and the temperature fluctuation amplitude are taken as the load prediction result.

9. A harmonic reducer performance adaptation device for a humanoid robot, characterized by, The harmonic reducer performance adaptation device for a humanoid robot includes: A database construction module is configured to construct a multi-dimensional performance index database of harmonic reducers based on static performance parameters and dynamic performance parameters of a plurality of models of harmonic reducers; A working condition modeling module is configured to construct a load working condition model based on a performance demand vector of a target joint of a humanoid robot in an application scenario; An adaptation calculation module is configured to perform adaptation calculation based on the multi-dimensional performance index database and the load working condition model to obtain a comprehensive matching degree score of each candidate harmonic reducer model and the performance demand vector and a load prediction result; A result recommendation module is configured to obtain an adaptation recommendation list of a target harmonic reducer model based on the comprehensive matching degree score and the load prediction result, and perform harmonic reducer performance adaptation based on the adaptation recommendation list. After the step of obtaining an adaptation recommendation list of a target harmonic reducer model based on the comprehensive matching degree score and the load prediction result, and performing harmonic reducer performance adaptation based on the adaptation recommendation list, the following steps are further included: Preferred models in the adaptation recommendation list are determined, and expected operating temperatures and expected system total energy consumptions of the preferred models when the target joint operates in the load working condition in the application scenario are predicted respectively; The expected operating temperatures and the expected system total energy consumptions are compared with preset performance thresholds respectively to obtain a decision result; When the decision result does not satisfy a preset condition, a performance difference value is determined, and the performance index weight is optimized according to the performance difference value; Adaptation calculation is performed based on the performance index weight, the multi-dimensional performance index database and the load working condition model to generate an optimized adaptation recommendation list.

Images