A Hybrid Neural Network-Genetic Programming Intelligent Control Approach

Abstract

The proposed work aims to introduce a novel approach to Intelligent Control (IC), based on the combined use of Genetic Programming (GP) and feedforward Neural Network (NN). Both techniques have been successfully used in the literature for regression and control applications, but, while a NN creates a black box model, GP allows for a greater interpretability of the created model, which is a key feature in control applications. The main idea behind the hybrid approach proposed in this paper is to combine the speed and flexibility of a NN with the interpretability of GP. Moreover, to improve the robustness of the GP control law against unforeseen environmental changes, a new selection and crossover mechanisms, called Inclusive Tournament and Inclusive Crossover, are also introduced. The proposed IC approach is tested on the guidance control of a space transportation system and results, showing the potentialities for real applications, are shown and discussed.

Genetic Programming Guidance Control System for a Reentry Vehicle under Uncertainties

“…The work produced in [28] was later used as a foundation to design the Hybrid GP-NN controller presented in [29]. In this control system, the GP was used offline to generate a control law that was later optimized online by a NN.…”

“…In this control system, the GP was used offline to generate a control law that was later optimized online by a NN. The approach to generate offline the GP control law is the same used in this work, with the following differences: (1) A different more challenging application is considered, (2) more severe uncertainties are applied and (3) a more thorough analysis of the GP performances is conducted than the one performed in [29].…”

“…The particular GP algorithm used in this work is the Inclusive Genetic Programming (IGP), which was originally presented in [29] and later analyzed in [30]. IGP was developed in Python 3 relying on the open source library DEAP [31].…”

“…IGP was designed to promote an maintain the population's diversity throughout the evolutionary process, so to avoid losing big individuals due to bloat control operators. In fact, it was observed in [29] that bigger individuals are capable of handling the nonlinearities of the treated problem better than smaller individuals. Therefore, to maintain the population's diversity and hence also preserve big individuals, three main features were inserted in the GP algorithm so to make what is now called the IGP.…”

Inclusive Genetic Programming

“…The importance of the population diversity in an evolutionary algorithm was treated by several publications in the past: in [5] diversity is addressed as a method to control exploration and exploitation during the evolutionary process, describing also different diversity measures; in [20] a survey of methodologies for promoting diversity in evolutionary optimization is presented; while in [4] different diversity measures are analyzed. This work aims to contribute to the landscape of publications about diversity maintenance and promotion by performing a deeper analysis of a heuristic proposed in a previous work [15]: the Inclusive Genetic Programming. This heuristic is based on a different formulation of the evolutionary operations, such as crossover, mutation and selection, aimed to promote and maintain the diversity in a GP population.…”