In cloud and edge computing environments, it is common to extract and combine available services to meet user needs. However, current methods struggle to cope with composition failures caused by changes in user needs or external environment. To address this challenge, this paper proposes a planning-based service composition and repair method in edge computing environments. We first combine the mobile path model and planning graph method to complete the service composition process. The construction of the graph allows for efficient evaluation and selection of service compositions that suit user demands. When the service set changes or user goals are modified, the method can generate new solutions based on the existing planning graph to meet user needs. This repair method can adapt to changes in real-time in the cloud-edge environment, enhancing system flexibility and reliability. Experimental tests have shown that the proposed repair method outperforms replanning, demonstrating its effectiveness and practicality in addressing combination failures.