A distributed penalty-based zeroing neural network for time-varying optimization with both equality and inequality constraints and its application to cooperative control of redundant robot manipulators

• DOI: 10.3389/fnbot.2025.1553623
• Publication Date: 2025-03-17T07:22:21Z
• Authors: Liu He; Hui Cheng; Yunong Zhang
• Keywords: Penalty Method

This study addresses the distributed optimization problem with time-varying objective functions and constraints in a multi-agent system (MAS). To tackle constrained (DTVCO) problem, each agent MAS communicates its neighbors while relying solely on local information, such as own function constraints, to compute optimal solution. We propose novel penalty-based zeroing neural network (PB-ZNN) solve continuous-time DTVCO (CTDTVCO) problem. The PB-ZNN model incorporates two penalty functions: first penalizes agents for deviating from states of their neighbors, driving all reach consensus, second falling outside feasible range, ensuring that solutions remain within constraints. solves CTDTVCO semi-centralized manner, where information exchange between is distributed, but computation centralized. Building model, we adopt Euler formula develop (DPB-ZNN) algorithm solving discrete-time (DTDTVCO) fully manner. present prove convergence theorems proposed DPB-ZNN algorithm. efficacy accuracy are illustrated through numerical examples, including simulation experiment applying cooperative control redundant manipulators.