The Study Of The Hydraulic Hoist Based On AMESIM And Neural Network

The study object of this paper is JKY2/1.5B type explosion protection hydraulichoist used in mine. Currently, the hydraulic hoist used in mine is mainly manual, thatis, the lift speed and the docking position are controlled by the driver who coordinatesthe driving system and braking system in the boot process, which leads to inaccuratecontrol. Especially when parking, multiple adjustments are often required in order toensure the accuracy of the parking position. The disadvantages brought by manuallycontrolling hydraulic hoist include lowly accurate layer parking, poorly comfortableriding and poorly coordination between driving and braking system.In this paper the mathematical model of JKY2/1.5B hydraulic hoist’s drivingsystems and braking systems is established. On the basis of this mathematical modeland by using amesim software the simulation model of hydraulic hoist including itsdriving system and braking system is built. Using amesim simulation model, theinfluence of adjusting the time delay of the braking system startup, the orifice size andthe high-pressure chamber volume of drive system on the coordination betweendriving system and braking system in start-up process is analyzed. To improve therapidity, accuracy and stability of the response of the hydraulic hoist, the single outputPID neural network controller (SPIDNN), whose learning efficiency and smoothingfactor are optimized by using genetic algorithm, is designed and simulated andanalyzed by compiling the corresponding program in the matlab software. Due to theexistence of strong nonlinear of electro-hydraulic servo system, the use of linearmathematical model can not meet the requirements for some higher-requirementsystems. Therefore, the paper provides a method of building a nonlinear model withfuzzy neural network, that is, we use the interface between amesim and matlab toobtain input and output data of the simulation model of hydraulic hoist, determine thecluster centers with subtractive clustering, and consider these cluster centers as initialcenter of membership function of adaptive neural network fuzzy inference model(ANFIS). Then ANFIS is trained with error back propagation algorithm and leastsquares algorithm, and finally we built the nonlinear model of hydraulic hoist drivesystem to prove the effectiveness of this method for establishing a nonlinear model.