Diffusion Model-Based Multiobjective Optimization for Gasoline Blending Scheduling

Wenxuan Fang, Wei Du, Renchu He, Yang Tang, Yaochu Jin, Gary G. Yen

2024-02-04

Abstract:Gasoline blending scheduling uses resource allocation and operation sequencing to meet a refinery's production requirements. The presence of nonlinearity, integer constraints, and a large number of decision variables adds complexity to this problem, posing challenges for traditional and evolutionary algorithms. This paper introduces a novel multiobjective optimization approach driven by a diffusion model (named DMO), which is designed specifically for gasoline blending scheduling. To address integer constraints and generate feasible schedules, the diffusion model creates multiple intermediate distributions between Gaussian noise and the feasible domain. Through iterative processes, the solutions transition from Gaussian noise to feasible schedules while optimizing the objectives using the gradient descent method. DMO achieves simultaneous objective optimization and constraint adherence. Comparative tests are conducted to evaluate DMO's performance across various scales. The experimental results demonstrate that DMO surpasses state-of-the-art multiobjective evolutionary algorithms in terms of efficiency when solving gasoline blending scheduling problems.

Artificial Intelligence

What problem does this paper attempt to address?

This paper proposes a new method to solve the gasoline blending and scheduling problem, based on the Diffusion Model Optimization (DMO) for multi-objective optimization. Gasoline blending and scheduling involves resource allocation and operation sequencing to meet the production requirements of oil refineries. However, due to the presence of nonlinearity, integer constraints, and a large number of decision variables, this problem becomes complex and poses challenges to traditional and evolutionary algorithms. DMO introduces a new optimization strategy by creating multiple intermediate distributions between Gaussian noise and the feasible region using the diffusion model, gradually transitioning from noise to feasible scheduling solutions during the iteration process while optimizing the objective function. This method is able to simultaneously achieve objective optimization and constraint satisfaction. Experimental results demonstrate that DMO is more efficient than existing multi-objective evolutionary algorithms for solving the gasoline blending and scheduling problem. The main contributions of this paper include the application of the diffusion model to practical scheduling optimization problems, the proposal of a new optimization path, and the superiority in efficiency and stability compared to existing algorithms.