Algorithm and Achieve of Self-balancing Two-wheeled Control System Based on PID Neural Network

For self-balancing two-wheeled vehicle position and angle control implementation issues, propose a control algorithm of self-balanced two-wheeled vehicle based on PID neural network. The algorithm overcomes the problem of PID controller generalization is weak, and can avoid over fitting and the BP neural network controller into a local optimum. The algorithm of the controller is constituted by a three-tier network, the first floor is the input layer, which receives two-wheel self-balancing vehicle position and Angle feedback information; a second layer of PID neuron layer, which is connected to the input layer, the state transition function respectively proportional function, integral function and the differential function; the third layer is the output layer, by weight in various proportions hidden layer output, the final composition of the control output of the controller. Control system fusion using kalman filter algorithm to get the self-balanced vehicle's attitude information. Through self-balancing two-wheeled vehicle actual experimental platform to verify the effectiveness of the control algorithm implementation.