Adaptive Human Simulated Intelligent Controller And Its Application In Inverted Pendulum System With Uniform Rods

How to make the robots run fast as human being is continuously a hot and difficult issue in science and engineering circle.Taking the people's running as an example,it covers the body balance with single foot,the body balance with feet,the body posture control and the switch between balance status.The foot-joint,knee-joint and articulation coxae are equal to the joints in inverted pendulum.The body's bones between joints are equal to the rods in inverted pendulum.So muli-inverted pendulum is the prototype of human body.They both have the common essence in balance problem.The control mechanism of multi-inverted-pendulum is also similar to rocket-launching,mechanical arms and self-banlance vehicles.What's more,inverted pendulum with uniform rods has stronger nonlinearity,coupleness and unstabibility comparing the inverted pendulum with short inner rod and long outer rod.This makes the research more challenging and interesting.Based on the inverted pendulum with uniform rods,its target is to uncover the essence in human being's running and provide a new method,frame and technology for designing the complex sytems,such as robots,mechanical arms etc.The exsting controllers of multi-inverted-pendulum all have narrow initial conditions,which show the ability of anti-disturbance and robust should be improved.So aiming at realizing the stabilization of multi-inverted-pendulum with wide initial conditions by a controller with strong anti-disturbance and robust,the three key steps are formed as following:(1)Widen the narrow initial conditions in multi-inverted-pendulum with uniform rods;(2)Filter the noise in multi-inverted-pendulum;(3)Divide the wide-nonlinear-area,finally form the solution.According the above three key steps,details are following:First,differential equations of double-inverted-pendulum and triple-invertedpendulum are built by Lagrange-theorem before parameter-identification.Through linearizing differential eqations at up-right position,the linear state-space is abtained to make the solid foundation for further research.Then aiming at narrow initial conditions of inverted pendulum,the whole phase-plane is divided into linear area and nonlinear area.Adaptive human simulated intelligent controller(AHSIC)with loop-structure and fuzzy parametres is proposed.The simulation and real-time experiments verified AHSIC was able to widen the initial conditions efficiently.Next,aiming at noise-sensibility of inverted pendulum,an improved Kalman Filter is proposed,which combines the stable-Kalman-Filter with decaying-memory-skill to form a method to design the Gain-Matrix of Kalman Filter.It is able to meet the real-time and overcome the model errors.The simulation and real-time experiments verified the method could filter the noise efficiently and relieve the motor's vibration.Finally aiming at wide-nonlinear-area in stabilizing triple inverted pendulum,extending the borders between linear area and nonlinear area to divide the whole phase-plane into nine areas,where different PD controllers are used,so nine-point PD controller is formed.Combining loop-structure with the nine-point PD controller,a new novel controller: loop-nine-point-AHSIC is proposed,which can be associated with improved Kalman Filter.Through above steps,the comprehensive control plan is formed.It realizes the stabilization of triple inverted pendulum with uniform rods and widens the initial conditions to 10 degrees.It also realized stabilization of triple inverted pendulum with different outer-rods' length under the same control parametres at the first time.The experiments verified the comprehensive control plan had excellent performances of anti-disturbances,robust and applicability.