BP neural network control of single inverted pendulum

The BP neural network can combine with Levenberg-Marquardt algorithm, which effectively overcomes the disadvantages of traditional neural network such as slow convergence speed, always converging to local minimum point and poor stability. Neural network capture the data of inverted pendulum with LQR optimal control method, set up multilayer forward feedback neural network. We through matlab simulate the control system to get the each state variable and control response curve of Single inverted pendulum. Experimental results show that the improved BP neural network can successfully control the inverted pendulum and have faster speed of control response and good stability than LQR optimal control.