| With the continuous development of aerospace science and technology,on-orbit assembly and manufacturing has gradually been the focus of future aerospace research.Space Manipulator will play a more and more important role in the aerospace industry,and the realization of automatic control and compliant operation is an indispensable function of a space manipulator.Six-axis force sensor can measure three directions of force and torque at the same time.It is commonly used at the end of the manipulator to realize the force and torque feedback of the manipulator.It is an indispensable essential sensor for the precise force control of the end mechanism of the manipulator.However,due to the unique working environment and high reliability requirements in the aerospace field.It is very difficult for ordinary commercial six-dimensional force sensor to meet the application requirements of a space manipulator.Therefore,it is necessary to develop a space fault-tolerant six-axis force sensor with high reliability,good consistency in all directions,high sensitivity,error diagnosis and redundancy,so as to meet the needs of space manipulator on-orbit operation in major scientific projects such as on-orbit assembly telescope and deep space exploration engineering in the future.Based on the in-depth analysis of the working principle and structure of the existing six-dimensional force sensor,this paper first carries out theoretical analysis and experimental verification of the classic six-axis force sensor structure,looking for a reliable method to improve the sensitivity of the sensor without loss of stiffness,and preliminaries verifies the relevant theory by processing cross beam sensor and joint torque sensor.A set of design methods to improve the strain beam structure of six-dimensional force sensor is proposed,which can effectively guide the design of fault-tolerant six-axis force sensor.Based on the theory of strain beam optimization,combined with the end application requirements of a space manipulator,a new six beam three-dimensional six-dimensional force sensor is proposed on the basis of full analysis of various elastomer structures which can realize redundant measurement.The six strain beams of the sensor are arranged symmetrically in the center,and the angle between the sensor beam and the coordinate dimensional is 15 degrees,in order to achieve the goal of sensitivity equalization in all directions.48 strain gauges are used to form 12 full bridge circuits to achieve the design requirements of one main and two standby redundancy for each channel.Aiming at the problems of many design parameters,complex relationship between design input and sensor output,and mutual influence between sensor parameters,this paper uses Isigh software to integrate UG,Patran,Nastran and MATLAB design and analysis software to optimize the 13 structural parameters which affect the sensitivity and stiffness of the sensor synchronously,avoiding the optimization of traditional six-dimensional force sensor single parameter and inconsistent optimization index have adverse effects on the final optimization model.In the aspect of optimization algorithm,the multi-objective particle swarm optimization algorithm is used to solve the Pareto optimal solution set with the sensor sensitivity,the sensitivity ratio of force and torque,the number of sensor conditions as the objectives,and the stiffness and strength of the sensor in all directions as the constraints.By comparing the options in the optimal solution set,the design parameters with the best comprehensive ability are obtained.To verify the rationality of the design of the spatial fault-tolerant six-axis force sensor,the strength,stiffness and sensitivity of the sensor was checked by using the Finite Element Method(FEM),and the optimal patch position of the strain beam of the fault-tolerant sensor was obtained.The strain gauge is pasted in strict accordance with the placement process.The principle prototype has been developed and the load platform is designed to calibrate the sensor.Measurement sensitivity of force channel and torque channel is 0.41 m V/V and 0.27 m V/V respectively.The linearity,repeatability and hysteresis error of the sensor are lower than 1.01% F.S..By comparing the decoupling accuracy of the least square method,BP neural network and RBF neural network,it can be observed that using RBF neural network to decouple the fault-tolerant six-axis force sensor can give full play to the measurement accuracy of the sensor.The decoupling accuracy is better than0.13%F.S..In order to satisfy the need for error diagnosis and fault tolerance of sensors on-orbit,an error diagnosis mechanism based on the least square method is proposed.Compared with the traditional back calculation voltage method,multi result input comparison method and basis vector space method,this method takes full advantage of the internal connection of measurement results between channels,and has the advantages of high diagnosis efficiency and high accuracy.The generalized inverse matrix operation in fault monitoring is replaced by square matrix inversion operation,and the twelve generalized inverse operations in fault diagnosis are reduced to two square matrix inversion operations,which greatly reduces the calculation amount of fault diagnosis of the fault-tolerant six-axis force sensor.In terms of model reconstruction,considering the characteristics of hot backup between channels of the fault-tolerant six-axis force sensor,the calibration matrix was refreshed by combining ground calibration with on-orbit reinjection of software.The correctness of the method of error diagnosis and model reconstruction is verified by adding artificially interference in a definite path. |
PDF Full Text Request