Study On The Control Of The Wheel Loader With Limited Slip Differential On Neural Network

As poor working conditions of the wheel loader, wheels always slip. That causes thetraction and through capacity declining sharply. So that reduces the work efficiency of the loader.In our country, the loaders generally use the Open Differential and Passive Limited SlipDifferential (PLSD). Active Limited Slip Differential (ALSD) has been used in sedan widely.However, has not been used in Wheel Loader.Compared with the development of ALSD at home and abroad, analyzed the structure andthe weakness of the ALSD. ZL50Loader is taken as the object completed the improvementbased on comparison of the structure and working principal of Open Differential and PLSD.Proposed one new type electromagnetic ALSD. The paper completed the type selection,structural design and transmission analysis of electromagnetic ALSD.With ZL50Loader’s complex road condition, the control input and output ofelectromagnetic ALSD are nonlinear. Proposed the BP neural network model, confirmed neuronnumbers of input, layer and output. Controlled the magnetic valve based on BP neural networkand PID controller which on the electromagnetic ALSD. Confirmed the initial weight andthreshold value, and calculated output of the BP neural network which used for controlparameters of the PID controller. The difference of the system output and sample sent back toadjust the weight. Calculated repeatedly, so that got the three parameters of the controller.Analyzed and modeled the ZL50Wheel Loader drive system, and modeled the composedunits. The engine, hydraulic converter have been modeled, and the phase-matching characteristicof them have been investigated. It also built the transmission, the drive axle and the tire models.The Loader in straight drive and steering drive are both established.Separated the differential from the drive system, controlled the press in hydro-cylinderwhich impacted by the push rod of electromagnetic ALSD by the PID control algorithm based onthe BP neural network. ZL50loader is used as the simulation vehicle, three μ-slipt roads aretaken to build the simulation system of the electromagnetic ALSD utilizing neural networkcontroller. In addition, simulated the Loader in straight drive and steering drive, compared theresult of simulation with Open Differential, PLSD and ALSD.The simulations results verify the fact that the electromagnetic ALSD by the PID controlalgorithm based on the BP neural network can overcome the shortcoming of distributing the torque averagely. It can adapt the μ-slipt roads well. The bigger adhesion coefficient differencesof the two sides, the better the limited slip effect is. It can achieve the limit slip well, and controlthe proper slip ratio during the poor adhesion road. It can improve the drive force and the lateralforce efficiently. In summarize, the structure parameters and the control strategy of the new typeelectromagnetic ALSD are proved reasonable and suitable.