Research On Dynamic Modeling Method Of Seven-degree-of-freedom Heavy-duty Hydraulic Manipulato

The high speed,high precision and heavy load engineering practice of the manipulator puts forward higher requirements for its dynamics characteristics.It is necessary to establish an accurate and quickly solved dynamics model in order to master its dynamics characteristics.The nonlinear and complex coupling time-varying characteristics of multi DOF(degree of freedom),high speed,heavy load and hydraulic drive together make it very difficult to construct and solve the 7-DOF heavy-duty hydraulic manipulator dynamics model based on parameter method.This paper takes the 7-DOF hydraulic manipulator under high speed and heavy load conditions as the research object to carry out the research on its dynamics modeling method.The main research contents are as follows:(1)The mechanical system,hydraulic system and control system are designed for each joint of the 7-DOF heavy-duty hydraulic manipulator according to the working principle and technical requirements.Forward kinematics analysis is carried out on the basis of typical working conditions of heavy-duty hydraulic manipulator.And the working space of the manipulator is obtained by Monte Carlo method.The correctness of the kinematics model is verified by MATLAB and Recur Dyn simulation.(2)The rigid-flexible coupling dynamics model considering joint friction characteristics and flexibility("parametric model" for short)is proposed.It is built on the basis of Newton-Euler rigid body dynamics model,and considers the nonlinear hydraulic system,joint flexibility and friction characteristics caused by high speed,heavy load and hydraulic drive.The correctness of the modeling is verified by MATLAB and Recur Dyn simulation.The friction models of manipulator joints are established,and the friction parameters of each joint are obtained by friction identification experiment.(3)A non-parametric dynamics model construction strategy based on deep learning method is proposed aiming at the problem of accurate dynamics modeling and fast solution of multi-DOF hydraulic manipulator under high speed and heavy-duty conditions.A 7-DOF heavy-duty hydraulic manipulator non-parametric dynamics model("non-parametric model" for short)is established based on this strategy.The methods and procedures of network training are described in detail.(4)The feature engineering method is adopted to extract the parameterized features related to high speed and heavy-duty in the parameter model in order to obtain the dynamic characteristics of a manipulator that cannot be modeled through parametric method,reduce the requirements of non-parametric models for training data and further improve the modeling accuracy of heavy-duty hydraulic manipulator.Combined with non-parametric methods,the semi-parametric deep learning modeling method and the construction strategy for a semi-parametric dynamics model of heavy-duty hydraulic manipulator are proposed.A 7-DOF semi-parametric dynamics model of heavy-duty hydraulic manipulator(referred to as the "semi-parametric model")is established,and the training and parameter optimization process of the semi-parametric model is provided.(5)The 7-DOF heavy-duty hydraulic manipulator test platform is built to carry out the operation test of replacing mill liner.The non-parametric and semi-parametric models are fully trained with the test data to obtain the influence of different hyperparameters on the calculation accuracy and solving speed of the models,and the relatively optimal hyperparameter sets are finally determined.The parametric,non-parametric and semi-parametric models are tested.The results show that the calculation accuracy of the non-parametric method is improved by 50.32% and68.54% compared with the parametric method on the medium load and heavy load datasets,respectively.The calculation accuracy of the semi-parametric method is improved by 10.89% and 8.39% compared with the non-parametric method on the medium load and heavy load datasets,respectively.