NSGA-II-based co-optimization of energy consumption and emissions in plastic modification

Abstract

To address the energy dissipation and multimodal pollutant emission control issues arising from the high-dimensional parameter space during polymer plastic modification, this paper constructs an adaptive NSGA-II collaborative control system based on an "edge agent inference-cloud-inspired optimization" architecture. This system integrates a Gaussian truncation mutation operator and a dynamic congestion metric matrix to reconstruct the evolutionary computation process for multi-objective optimization. Experimental data show that, under the condition of ensuring rheological physical constraints, the improved architecture controls the steady-state energy consumption to 81.0 kW and reduces the logarithmic multiple of harmful VOCs emission to -1.42. Simultaneously, the core algorithm achieves Pareto convergence in 80 iterations , with the forward response latency maintained at 214.5 milliseconds. This model realizes collaborative optimization between energy efficiency constraints and pollution control indicators in the industrial plastic modification process.