Improved beetle swarm optimization algorithm based PID neural network for decoupling control

The proportional-integral-differential neural network (PIDNN) has advantages in decoupling control of multivariable systems with strong coupling. However, the traditional PIDNN is easy to fall into local optimum due to the random initial weights. In order to promote control performance, an improved beetle swarm optimization algorithm is proposed for searching the optimal initial weights of PIDNN, which updates the inertia weight in a nonlinear manner, and a mutation factor is added to help the beetles get rid of the local optimal traps and enhance the optimization capability of algorithm. A greenhouse system is employed to demonstrate the effectiveness of the proposed method. The results suggest that it has fast response speed, high accuracy and ideal decoupling effect.