Research On Signal Acquisition Technology Of Electric Dynamometer Based On Compressed Sensing

The internal combustion engine will still be the main power source of automobiles for a long time and the development of clean and efficient internal combustion engine is one of the most effective means to achieve the national energy conservation and emission reduction targets.The dynamometer system can simulate the load conditions of the internal combustion engine under different operating states and measure the dynamic performance,thermal efficiency,emission and fuel consumption of the internal combustion engine under various operating conditions and provide reliable data support for the improvement of internal combustion engine.However,the dynamometer system continuously collects multi-type and multi-channel signals at a high sampling rate,resulting in a huge amount of data,which will take up a lot of resources in the storage and transmission of data and seriously affects the real-time and dynamic performance of the dynamometer system.To solve these problems,according to the sparsity of the electric dynamometer signal,this paper applies the Compressed Sensing theory to the signal acquisition and studies the signal acquisition method of the electric dynamometer based on the Compressed Sensing.This method we designed avoids the limitation of the signal bandwidth,compreszes and reconstructs the signal with a low sampling rate,reduces the hardware burden of the dynamometer system and provides a new solution for improving the real-time performance and response speed of the dynamometer system.The main contents of this paper are as follows:(1)The sparsity analysis was carried out on the signal of the electric dynamometer system.We find that temperature,humidity,pressure and other signals of the electric dynamometer system are generally linear stationary signal and have simple structure in the time domain.Comparatively speaking,the torque signal and vibration signal of the engine have dynamic non-stationary characteristics and the sparsity is easily disturbed.So using the Fourier transform basis as the sparse basis,the sparsity analysis of the torque and vibration signals in the constant current mode and the constant speed mode is carried out through the simulation experiment and the feasibility of applying the Compressed Sensing theory to the signal acquisition of the electric dynamometer is verified.(2)Aiming at signal reconstruction of electric dynamometer,an orthogonal matching and subspace tracking algorithm based on Dice coefficient was proposed by introducing Dice coefficient matching criterion and integrating SP algorithm’s "backtracking idea".This algorithm is combined with OMP algorithm and SP algorithm to carry out reconstruction experiments on one-dimensional discrete signal,torque signal and vibration signal respectively and the signal reconstruction probability of signal compression ratio M/N and signal sparsity K of the three algorithms is calculated.Taking the permanent magnet synchronous motor dynamometer as the experimental object,the torque signal and vibration signal were reconstructed by DOMP-SP algorithm.The experimental results show that the DOMP-SP algorithm has high reconstruction accuracy and can accurately reconstruct the torque signal and vibration signal.(3)We design a electric dynamometer signal acquisition system based on Compressed Sensing theory,including the hardware design of signal acquisition,software design in signal compression and reconstruction.Torque signal and vibration signal are collected,compressed and reconstructed using this system.The experimental results verify that the electric dynamometer signal acquisition system we designed can collect the torque signal and vibration signal with high precision under the condition of reducing the sampling rate,which effectively improves the real-time and reliability of the system.