Assessment and production planning of new energy vehicles based on LCA and fuzzy multi-objective optimization

Abstract This study targets specific uncertainties in New Energy Vehicle production planning, including fluctuating material costs, variable market demand, time-varying battery degradation, and random carbon accounting parameters, to resolve the conflict between environmental and economic benefits. Traditional independent Life Cycle Assessments only support static carbon hotspot identification without linking dynamic scheduling. Meanwhile, conventional multi-objective optimization ignores fuzzy parameter randomness and suffers from low computational stability. To address the inability of either approach to simultaneously complete environmental evaluation and reliable production planning, this study develops a framework coupling Life Cycle Assessment with fuzzy multi-objective optimization. The proposed model utilizes sequential Monte Carlo simulation combined with Long Short-Term Memory Networks to quantify time-series uncertain indicators and identify carbon and energy consumption hotspots. Furthermore, it integrates a Non-dominated Sorting Genetic Algorithm II, a Radial Basis Function Neural Network surrogate, and fuzzy decision-making for optimization. This integration improves stability under volatile conditions and bridges static evaluation with dynamic production decisions. Benchmark comparisons with various state-of-the-art models quantitatively demonstrate improvements in evaluation accuracy, robustness, optimization efficiency, and scheduling performance. Regarding evaluation performance, the model reaches a 92.43% operating condition fitting accuracy, a 15.77 ms response delay, a 1.98% root mean square error for energy prediction, and a 0.948 fault tracing precision, with only a 0.01% deviation from actual data. For scheduling performance, it achieves 94.28% equipment utilization, a 96.75% scheduling success rate, and a 1.41 s global solution time, reducing resource redundancy loss to 29.02%. Under the 2025 simulation scenario, the optimized annual output reaches 969,800 vehicles, with carbon dioxide emissions rising by only 23,200 tons. The composite model overcomes the drawback of static evaluations failing to reflect dynamic production, balances low-carbon goals with economic profits, and realizes efficient scheduling, providing feasible technical support for the sustainable transformation of vehicle manufacturers.