Abstract: Most production processes are challenged by bottlenecks that restrict the effective output of the production system. Thus, an identification method for bottleneck clusters of classes Ⅰ, Ⅱ, and Ⅲ is proposed to improve the single bottleneck identification method in identifying multiple bottlenecks in the production system simultaneously. Furthermore, to combat the uncertainty of the evaluation attribute value of the machine, the interval type is adopted to describe the evaluation attributes of the machines. Although interval description requires a substantial amount of calculation, a directed weighted network model of a production system is constructed. By analyzing the topological characteristics of the network, the candidate bottleneck machine groups were analyzed, calculated, and eliminated, thereby reducing the calculation time. Thus, the candidate bottleneck machine clusters are comprehensively evaluated and ranked regarding the following three aspects: machine utilization rate, average machine activity rate, and total energy consumption of a single machine. Finally, we use a fuzzy C-means algorithm to divide the candidate bottleneck machines into the bottleneck clusters of classes Ⅰ, Ⅱ, and Ⅲ.