The adjustable stator vane (VSV) adjustment mechanism of an aircraft engine was taken as the research object, and a kinematic model with kinematic pair clearance was established. Considering the randomness of factors such as drive error, dimensional error and assembly clearance, a VSV adjustment mechanism motion accuracy reliability model considering the failure correlation of each stator blade motion accuracy was established. The BP neural network proxy model was introduced, and the Monte Carlo method was used to calculate the reliability of the mechanism, and the variation law of the reliability with the number of stator blades and the failure threshold was given. By comparing with the system model with independent unit failure assumptions and the fully correlated system model, the rationality of the mechanism system motion accuracy reliability model considering failure correlation was verified. Finally, a method for estimating the reliability error of the complex system caused by the proxy model error was proposed, and it was proved that the reliability calculation result of the VSV adjustment mechanism is accurate and reliable.