In order to achieve a stable output voltage in wireless power transmission (WPT) systems and to address the complexity of tuning the adaptive PID parameters of BP neural networks, a WPT system utilizing adaptive PID control optimized by the Alpha Evolution (AE) algorithm for initial neural network weights is proposed. First, a model of the LCC-S type compensating network WPT system is established, and the Pade approximation is applied to reduce the system transfer function and design the controller. Secondly, the online learning capability of BP neural networks is utilized to optimize the control parameters in the PID controller in real-time, achieving constant voltage control for the system. Finally, a MATLAB Simulink simulation model of the LCC-S type is constructed. Simulation results show that compared to traditional PID and BPNN-PID controllers, the overshoot of the AE-BPNN-PID controller is reduced by a maximum of 18.4% during fast response, with the shortest settling time at 1.36ms, an improvement of 60.7% in speed; for slower responses, the rise time is fastest at 0.95ms, with a speed improvement of 57%. The AE-BPNN-PID controller resolves the complexity of parameter adjustment while also enhancing the system's response speed and reducing output overshoot and oscillation, thus validating the effectiveness of the AE-BPNN-PID controller in achieving stable output voltage for the LCC-S type WPT system.