Analytical Investigation of Intelligent Optimization Algorithms for Adaptive Neuro- Fuzzy Disturbance Observer for Spacecraft Attitude Control Simulator

In this paper, the effect of using various intelligent algorithms to optimize the adaptive neuro-fuzzy disturbance observer has been investigated .First, a model reference adaptive control is designed for the spacecraft simulator. Then, in order to reduce the disturbance effect, an adaptive neuro-fuzzy disturbance observer is used. In this paper, the fuzzy system is optimized using Intelligent Genetic Algorithm, Particle Swarm Optimization, Imperialist Competitive Algorithm, Bee Colony, Ant Colony Optimization, and especially Policy Gradient Particle Swarm Algorithm, which speeds up and optimizes the response. The Policy Gradient Particle Swarm algorithm is a combination of gradient policy reinforcement learning and particle swarming ideas and is a hybrid optimization method to control a nonlinear complex system with many applications in the real world. In this method, influenced by reinforcement idea, the policy gradient for a non-fossilized system is calculated, and in the optimization of particle swarm relations, optimization is performed in addition to the factors included in the congestion methods in the direction of the policy gradient. It is intended to optimize the fuzzy neuro system parameters and input and output data in the cost function. Finally, the neuro-fuzzy systems optimized by these algorithms are compared and it is shown that the gradient particle swarm algorithm performs better than the particle swarm algorithm.