Introducing The Accelerating Coordination Operator Multi-Objective Co-evolutionary Algorithms And Its Application In Inverted Pendulum Control

This article mainly from three parts to write, the first part is the simulation model implementation of circular two-level handstand pendulum; second part is the study of bionic intelligence algorithm; third part is using the improved algorithm of bionic intelligence algorithm for circular two-level handstand pendulum control multi-objective optimization. The following were introduced from these three aspects.The content of this segment is the simulation model implementation of circular two-level handstand pendulum. circular two-level handstand pendulum is more complicated than one-level inverted pendulum, and it is very difficult to use Newton mechanics to realize its mathematical model, So to apply Lagrange dynamics through energy to establishing its mathematical model, getting a second-order matrix differential equation, Then through the order reduction transformed into first-order differential equations, and Then apply a fourth-order longge coulthard method to establish the computer simulation model, which is a simulation platform For the multi-objective optimization.Bionic intelligent algorithm is the main content of this section. Genetic is the basis of study in this paper, and an accelerating operator was introduced into genetic algorithm, and greatly improve the running speed of the genetic algorithm, got an improved genetic algorithm,whose calculation speed was improved at least one order of magnitude, compared with the traditional genetic algorithm. Accelerating operator's shortage is that it can only be used for the single objective optimization, but later, a more perfect accelerating collaborative operator was designed. In a single target, accelerating collaborative operator's search ability is stronger than accelerating operator. In multiobjective optimization, accelerating collaborative operator also showed a good performance, not only search speed and pass the test of DTLZ3, a high dimensional multi-objective test functions, which have 10 targets and 19 variables. On this basis, a further study of co-evolutionary algorithms, proposed a more complex layered multi-objective co-evolutionary algorithms.This section's main content is the inverted pendulum multi-objective optimization. First, select the optimum targets; Secondly, design fitness functions according to optimal targets; Finally, use multi-objective optimization algorithms for circular two-level handstand pendulum control multi-objective optimization, and achieved good effect.