| This paper considers the application of multi-component force sensing mechanism under heavy load,based on the design concept of "heavy load,micro-force sensing",and based on the design idea of multi-dimensional force sensing mechanism with mixed branches.This paper innovatively divides the bearing branches into the bearing module,and improves the bearing capacity of the force sensing mechanism by adding bearing module.Aiming at the theoretical modeling of multi-dimensional force sensing mechanism,the applicability of Euler beam and Timoshenko beam to different configurations of multidimensional force sensing mechanism is studied,and the modeling accuracy of the two models is determined and compared.In this paper,configuration synthesis,global stiffness modeling and force mapping modeling of multi-dimensional force sensing mechanism,calibration experiment and data processing,decoupling algorithm analysis are carried out in turn,which have certain reference meaning for the design and calibration processing of heavy load miniaturized multi-component force sensing mechanism.First of all,based on the design concept of hybrid branch and the large range requirements of the multi-dimensional force sensing mechanism,this paper proposes to add the load module in a modular way to improve the stiffness of the whole prototype.The flexible parallel multi-dimensional force sensing mechanism based on load module is adopted as the research object of this paper,and the configuration without load module is used for comparative analysis.Secondly,based on the deformation superposition principle of the flexible series branches and the geometric compatibility condition of the parallel mechanism,the stiffness models of the two configurations are established.Starting from the flexibility matrix of the basic flexible element,the stiffness matrix of each branch is calculated,and the stiffness of the whole configuration is derived.Finally,the force mapping relationship between each branch and the moving platform is determined.The finite element analysis software is used to simulate the three-dimensional models of the two configurations.The Timoshenko beam model and Euler Bernoulli beam model are used to establish the theoretical models of the two configurations respectively,and the influence of the two beam models on the modeling accuracy of the two structures is determined.Thirdly,the calibration experiment of hybrid branch multi-dimensional force sensing mechanism is carried out.The calibration scheme is designed and the experiment is carried out.After normalizing the calibration data,the least square method,BP neural network and extreme learning machine are used to decouple the calibration data.The accuracy and decoupling time of the three decoupled algorithms are compared and analyzed.Finally,aiming at some problems of traditional BP neural network in decoupling multidimensional force sensing mechanism,swarm intelligence algorithm is used to optimize the initial weights and thresholds of BP neural network.The optimization is based on ant colony algorithm and cuckoo algorithm respectively,and the neural network decoupling training is carried out to analyze the decoupling accuracy of the two optimized neural network models. |
