Many engineering systems are affected by shocks from their operating environments. When the state of a component degrades to a certain threshold level, preventive maintenance is needed for the purpose of reliability improvement. However, existing studies usually ignore the impact of shocks on different components of a system and therefore on the maintenance policies. This paper proposes to model the degradation processes of components with a k-dimensional Wiener process. Under both deterministic and stochastic environmental conditions, the Eyring model is used to measure the environmental importance of the multi-dimensional degradation process. Based on different failure scenarios, different maintenance strategies are therefore proposed. A periodic inspection policy is considered for each component that may fail due to external shocks. As for multiple components, the maintenance priority is determined based on the joint importance, and optimal preventive maintenance is obtained under the constraint of limited resources. Finally, a robot system is taken as an example to illustrate the proposed methods.