Estimation Research Of Mechanical Parameters Of Planet Soil Based On Wheel-soil Model

The surface of the planet is covered with soft soils.Due to the lack of knowledge of the mechanics research,the wheel of planetary rovers may sink into the soil,slip and not enough traction.Improving the trafficability of planetary rovers can increase the working time so that enhancing the comprehensive power of the national deep space exploration.A rigid wheel-soil model is established in this paper and combining with single-wheel soil bin test through GA-BP and RBF algorithms to identify the shear and pressure-sinkage mechanical parameters of planet soil.It can realize the real-time prediction of soil mechanical parameters around the planetary rover.It is used for optimal route planning,risk aversion and traction control of the planetary rover.Based on the classical wheel-soil model and optimizing the stress distribution under wheel and considering the influence of wheel lugs for the wheel-soil model,a simplified wheel-soil model and a wheel-soil model considering the lugs effect are established.(1)The vertical load,driving torques and drawbar pull are predicted by operating the models through MATLAB.(2)The mechanical parameters of the simulated planet soil are tested by experiments.(3)The vertical load,driving torques,drawbar pull and slip ratio are obtained by single-wheel soil bin test.(4)By comparison,the simplified wheel-soil model can be used to predict trafficability of rigid wheels with lugs.The above four parts verify the reliability of the two models and the simplified model is more reliable.The simplified model can provide a large amount of data to train the algorithms of follow-up research.GA-BP algorithm based on simplified wheel-soil model is used to identify the shear parameters of planet soil.For GA-BP identification model,the input data are torque(T),vertical load(W)and slip ratio(s)and the output data are internal friction angle(φ)and shear deformation modulus(K).A total of 315 sets of data are used to train GA-BP algorithm.The predicted results ofφand K are 32.20°and 1.74 cm,respectively.The relative errors with the experimental values are 5.3%and 1.7%,respectively.BP network is the control group of GA-BP algorithm.The predicted results ofφand K are 36.35°and 1.71 cm,respectively.The relative errors with the experimental values are 18.87%and 3.39%,respectively.Therefore,GA-BP algorithm can accurately and effectively identify the planet soil shear parameters online.Radial Basis Function Neural Network(RBF)is used to identify the pressure-sinkage parameters of planet soils.For the RBF identification model,the input data are T,W and s and the output data are combined deformation modulus(Ks)and defomation index(n).A total of 288 sets of data are used to train the RBF algorithm.39 sets of single-wheel soil bin test data are input into the trained algorithm to identify the K_s and n of the simulated lunar soil.The experimental results show that the result of K_s is 0.7 N/cm~2 and the maximum relative error is 11.1%;the result of n is 0.916 and the maximum relative error is 8.3%.Through model equation to calculate DP shows that the predicted DP results are stable,and the deviation from the experimental values is small.The relative errors of the three vertical loads with low slip rate are 15.88%,15.05%and 13.03%,respectively.Therefore,RBF neural network can be used to predict the pressure-sinkage parameters of planet soil,estimate DP values and guide the operation of the planetary rover.