Abstract: With the accelerated development of a new generation of information technology, how to promote the transformation and upgrading of the industrial structure, coordinate the government and market forces to effectively supply industrial policies, explore suitable policy implementation path, promote China's industry in the global value chain of high-end, build an advanced manufacturing industry with strong competitive cluster, achieve high quality development, become a key problem facing China's manufacturing industry. One of the major characteristics of China's economy is that both market and government are playing a role. The special institutional environment and the uncertainty of institutional environment brought by economic transformation affect the evolution and upgrading process of China's industrial clusters, which leads to a great difference between the development path of China's industrial clusters and that of the West. Firstly, what roles do government behavior (visible hand) and market mechanism (invisible hand) play in different stages of manufacturing cluster evolution and upgrading? What is the mechanism? Secondly, how do industry associations promote the evolution and upgrading of manufacturing clusters? What is the mechanism? Finally, how does the competitiveness of industrial clusters evolve and upgrade under the promotion of multi-stakeholder? In this regard, institutional logic theory, as a link between macro-level institutional theory and micro-level actor's behavior, is a set of formal and informal rules that shape the cognition and behavior of actors in the organizational field. It has certain historical weight variability and is conducive to explaining the process mechanism of dynamic evolution in a long period. Based on this, with the help of institutional logic theory and from the perspective of time, this paper takes the woolen weaving industry in Dalang, Dongguan as an example, and takes the government, cluster enterprises and industry associations involved in the evolution and upgrading of manufacturing clusters as the main body of interest, and discusses the multiple factors in different stages. The mechanism of institutional logic on the evolution and upgrading of manufacturing clusters opens the "black box" of the evolution and upgrading process of manufacturing clusters under multiple institutional logics, in order to provide theoretical reference for the evolution and upgrading mechanism of manufacturing clusters, and provide reference for government policy formulation . The findings are as follows: Firstly, the evolution and upgrading of manufacturing cluster is an alternating process of gradual evolution and abrupt upgrading. In the stage of gradual evolution, cluster logic is dominated by market logic. In the abrupt upgrade stage, the cluster logic is dominated by government logic. Secondly, the conflict between the original government logic and market logic is the main driving force to upgrade the manufacturing cluster. Thirdly, with the development and expansion of cluster, the professional logic with industry associations as the main body tends to replace the government logic. Fourthly, the development of manufacturing cluster is based on the joint action of efficiency mechanism (dominated by market logic) and legitimacy mechanism (dominated by government logic), and the cooperative evolution and upgrading of various actors (government, industry association and cluster enterprises, etc.). Based on the time window and the theory of institutional logic, this paper discusses the mechanism of local government, cluster enterprises and industry associations in the process of cluster evolution and upgrading, constructs the evolution framework of the dominant logic followed by different stages of cluster under the time dimension, and answers the question of how to improve the competitiveness of cluster. This paper adopts a single case study. Although the selected cases are representative to some extent and have unique advantages for theoretical construction, the universality of the research conclusion still needs to be cautious. In the future, the accuracy of the research conclusion can be tested through large sample statistics.