Three multi-objective memtic algorithms for observation scheduling problem of active-imaging AEOS

TL;DR

This paper introduces a new bi-objective optimization problem for agile earth observation satellites with variable image durations and proposes three enhanced multi-objective memetic algorithms to solve it effectively.

Contribution

The paper formulates the OSWVID problem and develops three novel memetic algorithms incorporating heuristic operators for improved scheduling performance.

Findings

Algorithms outperform baseline methods in simulation tests.

Parameter tuning significantly improves algorithm efficiency.

Proposed methods effectively balance image quality and energy consumption.

Abstract

Observation scheduling problem for agile earth observation satellites (OSPFAS) plays a critical role in management of agile earth observation satellites (AEOSs). Active imaging enriches the extension of OSPFAS, we call the novel problem as observation scheduling problem for AEOS with variable image duration (OSWVID). A cumulative image quality and a detailed energy consumption is proposed to build OSWVID as a bi-objective optimization model. Three multi-objective memetic algorithms, PD+NSGA-II, LA+NSGA-II and ALNS+NSGA-II, are then designed to solve OSWVID. Considering the heuristic knowledge summarized in our previous research, several operators are designed for improving these three algorithms respectively. Based on existing instances, we analyze the critical parameters optimization, operators evolution, and efficiency of these three algorithms according to extensive simulation experiments.