The development of Artificial Intelligence (AI) requires a significant amount of high-performance semiconductor chips. Scientifically and effectively scheduling and optimizing semiconductor wafer manufacturing systems can improve the productivity of the production system, equipment utilization, and semiconductor chip output. Therefore, scheduling optimization and decision-making for semiconductor wafer manufacturing systems have become a hot research topic for current researchers. However, bottleneck drift is an important feature of semiconductor wafer manufacturing systems, and is also a key factor that constrains scheduling and decision-making in semiconductor wafer manufacturing systems. Currently, there is a lack of research on the specific characteristics of bottleneck drift in semiconductor wafer manufacturing systems in the literature, which leads to problems such as ineffective planning and scheduling decisions by managers, and affects the performance of semiconductor wafer manufacturing systems. This paper considers the triggering conditions, occurrence time, and impact on production plan objectives of bottleneck drift, and improves two methods for bottleneck identification and bottleneck drift, based on which a new semiconductor wafer manufacturing systems scheduling start-up mechanism is proposed, which includes three parts and nine steps. Firstly, the semiconductor wafer manufacturing system scheduling process is given, including bottleneck identification, impact analysis, prediction, and planning for the impact of bottleneck drift on production plan objectives. Secondly, bottleneck identification and impact factors are predicted based on bottleneck drift, and the trigger conditions and drift time of bottleneck drift are determined. Then, based on the delivery time of the production plan objectives, the impact of bottleneck drift on production plan objectives is analyzed to determine the scheduling start-up conditions for the semiconductor wafer manufacturing system. Experimental results show that the mechanism proposed in this paper can timely and accurately identify the triggering conditions and drift time of bottleneck drift, avoid ineffective scheduling decisions before bottleneck drift, avoid unnecessary production scheduling decisions, and reduce additional performance losses. In addition, it can analyze the impact of bottleneck drift on production plan objectives according to the production plan objectives, so as to formulate scheduling decision-making schemes better, avoiding production plan delays and waste of production capacity.
Read full abstract