Research Of Fuzzy Neural Network Control Of Semi-active Suspension Based On GA

Suspension system as an important of automobile assembly, with a buffer roadimpact and weaken the effect of body vibration that caused by uneven road, the comfortproperties of the car has a particularly significant impact. The traditional passivesuspension structure is simple, low cost, good reliability, but its parameters are fixedand can not be changed accordingly based on the actual road conditions, so the vibrationattenuation performance is limited. Active suspension due to the installation of theactuator, timely according to road conditions to adjust the suspension performance tomake the vibration attenuation performance is always in the best condition, so itsperformance remarkable; but there are also defect such as complex structure, plenty ofenergy consumption. The semi-active suspension between passive suspension and activesuspension, by adjusting the stiffness coefficient and damping coefficient method tocontrol the suspension, its vibration attenuation performance is close to the Activesuspension, but its structure is relatively simple, the energy consumption is also smaller;As a result, the semi-active suspension has become a focus in the study of automotiveengineering.In this paper, the semi-active suspension system is selected as the research object,and combined with control strategy research, to improve the semi-active suspensiondamping performance, enhance the vehicle riding comfort. Therefore, the main work ofthis paper lies in two aspects, one is the establishment of semi-active suspension systemmodel, and the other is the design of control strategy.In the aspect of the semi-active suspension system model, this paper analyzes themechanism of vehicle vibration; clearly random road roughness is the main cause ofvehicle vibration, and using the power spectral density of the road to build a model,Secondly, several commonly used suspension model modeling to derive a simplifiedconditions of suspension model; After through the simplified conditions of suspensionmodel, the Ford Granada automobile suspension model is reduced to two degrees offreedom1/4car suspension model and the semi-active suspension dynamics model isestablished in the ADAMS software. Finally, the amplitude-frequency characteristiccurve of evaluation index of vehicle ride comfort is established through simulation, andthen contrastively analyzes the relationship among the evaluation index. In the design of the control strategy, this paper firstly analyze the differentcharaceristics of the fuzzy control and neural network control method,clarify theadvantage of combining fuzzy and neural network, then establish fuzzy neural networkcontroller for semi-active suspension. In order to further improve the control effect offuzzy neural network controller, this paper further research and analysis, then aiming atthe shortcomings of the learning algorithm, this paper design a fuzzy neural networkcontrol method based on BP and GA, and applied it in semi-active suspension.On ADAMS and MATLAB co-simulation platform, the joint simulation model isestablished between the semi-active suspension dynamic model and control strategymodel, then through co-simulation verify that the design of fuzzy neural networkcontrol and fuzzy neural network control based on GA to improve the vibrationperformance of semi-active suspension, improve vehicle riding comfort is effective.