Research On Information Acquisition Technology Of Tracked Vehicle Drive Wheel

Tracked vehicles are widely used in agriculture,industry and military fields because of their large grounding area,strong attachment capacity,heavy load operation and strong cross furrow ability.As the ultimate link of energy in the tracked vehicle transmission system,the active wheel provides power for the whole transmission system.It is of great significance to study the transmission efficiency of the engine and the structure optimization of the active wheel if it is able to obtain the torque and speed parameters of the driving wheel.According to the force law and the stress distribution of the active wheel,a set of torque testing method and speed test method suitable for the active wheel is set up,and the strain data of the driving wheel in the marching process of the tracked vehicle is collected in combination with the storage test technology.According to the strain data and the physical model design algorithm,the depth learning is applied to the torque information extraction,and the short-time Fu Liye and LM are applied to the speed information extraction.The main contents of the paper are the force analysis of the driving wheel,the realization of the hardware function,the design of the torque calibration scheme,and the solution of the measured data.(1)in the part of the active wheel stress analysis,the simulation is completed using NX software.According to the stress distribution,the scheme of the strain sheet distribution is determined.The framework of the depth learning algorithm is preliminarily determined according to the layout scheme.The simulation data is used as the sample data of the learning algorithm,which provides reference for the follow-up scheme verification and the circuit design.(2)in the realization part of hardware function,the characteristic of the measured signal is estimated according to the force law and the stress distribution of the driving wheel.The maximum error of the circuit is calculated according to the characteristics of the estimation and the characteristics of the full bridge circuit,and how to avoid the error distortion and how to calculate the parameters of the filtering strategy are introduced in detail,and the function of small signal extraction is realized.(3)in the design part of torque calibration,the simulation data is taken as the sample value,and the feasibility of the calibration scheme is verified by the deep learning algorithm.And the structure of calibration calibration test rig is designed.The strength check of the main body,the arm,the drive shaft and the connecting shaft of the main shaft is carried out by NX and Ansys co simulation.(4)in the calculation part of the measured data,the data of the calibration test is trained with the depth learning algorithm and the theoretical derivation is used to find a learning algorithm suitable for the sample data.Short time Fu Liye and LM algorithm are combined to deal with the actual sports car data.The algorithm of extracting the expected signal in the noise of high frequency and low frequency is studied.The speed information is extracted and calculated,and compared with the measurement value of the speed sensor,in order to verify the effectiveness of the algorithm.