关键词: “双碳”目标, 分布式多项目调度问题, 汽车制造行业, 多目标优化

Abstract: The green and low-carbon production mode has a significant impact on promoting the “dual carbon” goal. The automobile manufacturing industry is a pillar industry in the development of the economy. Its production management process has a large total amount and high intensity of carbon emissions, with the characteristics of multi-project parallel management and distributed resources competition. However, the existing research on distributed resources-constrained multi-project scheduling problem (DRCMPSP) lacks effective integration with the automotive manufacturing industry. Most research focuses only on economic benefit objectives such as project duration, resources cost and project quality, neglecting environment benefit objectives, which poses challenges for achieving the “dual carbon” goals. Given the project duration is an important economic benefit objective and the carbon emission is an important environment benefit objective, we study the optimization and balance of multi-project duration and carbon emission according to the characteristics of DRCMPSP in automobile manufacturing industry.
It is believed in the management practice that reducing project duration or resources cost investment can reduce the resources usage time and quantity, achieving emission reduction effects. However, the unit costs of different resources types are not consistent with the unit carbon emissions. Moreover, the occupation time of resources with higher carbon emissions can be shortened further by optimizing the activity sequence under the same multi-project duration. Therefore, neither objective can completely replace the carbon emission objective, and it is necessary to set a direct carbon emission objective function. The automotive manufacturing process involves a variety of resources types, and the resources management process has both centralized and distributed characteristics, which poses challenges for setting scientific carbon emission objective function. At the same time, solving the project duration of DRCPSP is a strong NP-hard problem. On this basis, the addition of carbon emission objective with double objective optimization increases the difficulty of problem modeling and solving.
In view of the above shortcomings, we have conducted the research into the construction of carbon emission objective function and double objectives optimization. The main work of this paper is as follows: Firstly, an objective function is constructed from the resource usages which account for more than 90% carbon emission source during the automobile manufacturing process. Combined with the characteristics of local autonomy within sub-projects and resources competition among sub-projects, a dual objective optimization model of multi-project duration and carbon emission is constructed based on the multi-agent system. Secondly, an adaptive time step conflict activities selection strategy is designed according to the actual situation of conflict activities when global resources conflict, and a catastrophic strategy elite genetic algorithm or enumeration algorithm is selected to coordinate global resources conflicts according to the scale of conflict activities. Finally, 60 modified distributed multi-project scheduling testcases are used to test the model and algorithm, and explore the influence factors and influence rules on multi-project duration and carbon emission objectives.
The results show that: (1)The model and algorithm designed in this study can effectively solve the DRCMPSP in the automotive manufacturing industry. Moreover, the global resource coordination strategy designed in this study can obtain better solutions than sequential game coordination strategy and stochastic coordination strategy. (2)The carbon emissions of different types of resources can be reduced by increasing the carbon emission objective weight ratio or reducing the equipment energy consumption power level while ensuring the stability of multi-project duration, with local resources being more susceptible to emission reduction than global resources. (3)When faced with the expansion of different dimensions of problem scale, the multi-project duration objective can be controlled by alleviating the growth of problem scale within sub-projects, and the carbon emission objective can be controlled by alleviating the growth of problem scale among sub-projects. Then, the two objectives can be further optimized and balanced.

Key words: “dual carbon” goals, DRCMPSP, automobile manufacturing industry, multi-objective optimization