Sliding Mode Neural Network Chaos Control, The Brushless Dc Motor

The real nature is a non-linear system, and the control ofthe nonlinear system is always the scientific research's hotspot and difficulty, in particular, the control of chaos systemis a special nonlinear system's typical representative. atpresent, The non-linear control research was not very mature yet,the traditional control theory and methods are not very ideal,in recent years ,while the intelligent control was introducedinto non-linearity control, which became one kind ofextremely promising future.This article proposed a new kind of control method based onneural networks and variable structure sliding mode in view offirst order nonlinear system, which is called the sliding modeneural networks method. The chaos's control is divided by thefollow-up control and the calm control. Now, the calm questionis studied here. Under the system controllable premise, At thebegin, the system was separated by linear part and non-linearpart, the non-linearity partially uses the RBF neural networksto approach it, adjust or compensate it's non-linear influences,then the system becomes a first order linear system after theelimination of the non-linear influences. In the end, this linear system was calmed by variable structure sliding modecontrol. The two unifies finally calmly controls this nonlinearsystem, and proposed that we can fully calm the output to theselect pot through adjust it by adding a proper amendment.Secondly, this sliding mode neural networks methodsuccessfully calm the brushless DC motor's chaos, and it'ssimulation in matlab's environment has confirmed the controlis valid and accurate. The brushless DC motor's chaos systemis difficult to control, which is a high non-linearity and acoupling strong system, therefore the application to brushlessDC motor is a typical representation of chaos control.Thirdly, brushless DC motor's chaos system worked as anapplication example, sliding mode neural networks method'srobust issue discussed from five aspects. Firstly, Thesimulation found that the control system uses differentstructure RBF neural network all be able to calm the chaos ,though the final calm point is different; Secondly, a white gaussnoise was added to disturb the system, the simulation justifiedthe control has a very good anti-jamming property; Thirdly, aperturbation was added in the chaos system which satisfied thesliding mode condition matched invariability, the simulationresult justifies the sliding mode neural networks method has a very good anti-perturbation property, Finally, a step leapsignal was added to system before control and a pulse signal wasadded after the control system is stable which confirm thesliding mode neural networks is robust.The research indicated, the sliding mode neural networksmethod proposed here is a kind of simple control method in viewof the first order nonlinear system, which is easy to realize,it has a very good anti-jamming, anti- perturbation property,in the control domain, it has much more significance particularin the control of a kind of un-definite system.This article finally proposed some question on the slidingmode neural networks method, prospected future deeply research,better development and application.