The Giant-circle Movement Control Of A Circle-rail Double Inverted Pendulum Using Human Simulated Intelligent Control

The Inverted pendulum system is an ideal platform to validate the correctness, validity and application problems in real time control of all kinds of control theory, algorithms and strategy, because it is inherently unstable, non-liner, under-actuated, multi-variable and strong coupled. Now, taking the double inverted pendulum as an example, the control aims can be classified three levels in terms of control difficulty, that is: 1) handstand control near the upright position; 2) swing-up and handstand control that the rods movement from the hanging to the upright position and keeps balance, 3) the arbitrary switch control, in which the pendulum can be transferred arbitrarily between the four balance positions of the two rods. This paper used human simulated intelligent control (HSIC) to achieve the giant-circle movement control of circle-rail double inverted pendulum (CRDP).The giant-circle movement control of a CRDP is defined for two initial states: Up-Up and Up-Down. That is, the inner and outer rods of the CRDP rotate around the rotating arm in an approximately straight line (for Up-Up) or in the approximate single pendulum way (for Up-Down) from an initial upright position (Up-Up, Up-Down). This is enabled by controlling the horizontal back and forth movement of the rotating arm and then returning it to the initial position and maintaining its balance.In the first, the model of CRDP is deduced and its physical parameters is identified by improve genetic algorithm (IGA). Then, in the simulation platform, by imitating the human control behavior and analyzing the control process, the giant-circle movement control is divided into four stages, for Up-Up giant-circle movent control, fall down attitude adjustment phase, consumed energy compensation phase, swing up attitude adjustment phase and balance control phase are included, and the Up-Down giant-circle movement control includes the initial movement phase, the attitude adjustment phase, the energy consumed phase and the balance control phase. Using Human Simulated Intelligent Control theory, the control laws (Sensory-Motor Intelligent Schemas) is designed respectively, including the sensor schema (characteristic model) and the corresponding motor schema (control mode) of each of the four phases. The effectiveness of these control laws is proved by successful control both in a computer simulation and in real-time control of a physical experimental device.