Fully investigating available parking resources based on the concepts of shared parking has emerged as a new solution, in order to alleviate the problems of traffic congestion and imbalances between parking supply and demand in mega cities caused by the excessive growth of motor vehicle ownership. Yet, the resources of shared parking cannot be efficiently utilized in real applications dur o different applicability of shared parking service methods under different modes and the strict time window constraints, exclusivity and uniqueness of shared parking. The goal of utilizing rate of idle time was maximized in this paper, in addition, the parking spaces were divded into fixed and non-fixed spaces, thence shared parking configuration optimization methods were established correspondingly. Firstly, for non-fixed parking spaces, under the conditions of constructing a time series model to dynamically predict the parking demand, the net increase of the vehicle within the maximum acceptable time of 15 minutes for the driver as the reserved berth value is used to determine the shared berth configuration model. Secondly, for fixed berths, the differences between sharing requests in random non-reservation and reservation modes are distinguished, maximizing the utilization of a single parking space’s idle time to build a shard parking configuration model is aimed for the former one, while the maximum matching rules tree theory is used for the latter one to discuss the shared parking configuration model pf maximizing the system’s idle time. Finally, numerical experiments are used to demonstrate that shared parking can effectively increase the utilization and turnover rate of parking spaces. reserved sharing mode can increase the idle utilization rate of berths by 12.9% compared with the non-reservation mode is indicated from the results.