Optimized Design Of Stewart Six-dimensional Force Sensor

In this paper,the Stewart six-dimensional force sensor is studied.The main contents include the relationship between the isotropy of six-dimensional force and the main structural parameters of the sensor,the design of the sensor prototype system,the finite element simulation and the optimization of decoupling algorithm.For the six-dimensional force sensor,it should have better force isotropy and moment isotropy.In order to study the relationship between the structural parameters and the isotropic performance of the sensor,the structural model of the sensor is established.For the classical 6/6 Stewart six-dimensional force sensor,the relationship between the isotropy and the five main structural parameters is analyzed by studying the condition numbers of Jacobian matrix and Jacobian matrix.In view of this relationship,a constrained optimization mathematical model is established according to the actual requirement range of sensor size,and the optimal structural parameters are obtained by genetic algorithm optimization method.The branch elastomer with cross beam structure is made of the conversion element on the elastomer by sputtering coating process.The collected signal is converted into electrical signal by Whiston bridge,and filtered,amplified and zeroed by signal conditioning circuit.The mechanical interface of the sensor is checked.The finite element analysis model of the sensor is established by using the optimal parameters and the mechanical structure such as the elastomer of the branch.The finite element simulation software ABAQUS is used to carry out the simulation analysis.Firstly,the static simulation is carried out,and the maximum force/moment is loaded separately,and the stress and stiffness are analyzed.Then,the thermodynamic simulation is carried out,and the thermal stress distribution from room temperature to ultimate high temperature and from room temperature to ultimate low temperature is analyzed respectively.The fatigue life of each direction is analyzed,which proves that the load can reach 100000 times.Finally,the natural frequency of the sensor is analyzed.Through the above simulation analysis,it is proved that the sensor under the structural parameters can meet the requirements of the indicators.According to the actual measured zero-point voltage,the relationship between the zero-point voltage and temperature of the sensor is fitted,and the zero-point drift problem of the sensor is solved from the algorithm.Two decoupling algorithms are studied,and the operation process of establishing BP neural network using MATLAB toolbox is introduced in detail.The traditional calibration method is used to collect the data after one-dimensional loading,and two decoupling algorithms are used to solve the problem respectively,and the multi-dimensional loading data is used to verify the two methods.It is found that the two methods have the problems of large coupling between dimensions and multi-dimensional loading can not be used.A new calibration method is proposed and verified by two algorithms,which proves that under the new calibration method,the new method can make the multi-dimensional loading impossible to use.The decoupling operation with BP neural network can reduce the precision of force calculation to 2.27% and moment calculation to 1.73%,which meets the actual requirements.From the initial theoretical analysis to the final practical verification,the feasibility of the sensor prototype and decoupling algorithm is proved,which provides theoretical and applied support for the production of six-dimensional force sensor.