Research Of Beam Structure Active Vibration Control Based On Fuzzy Neural Networks Algorithm

Piezoceramics have been known as low-cost, lightweight, and easy-to-implementmaterials for active control of structural vibration. This paper presented the fuzzy controlalgorithm, fuzzy neural network (FNN) algorithm and an optimal training FNN to control thepiezoceramics which bonded on the surface of a cantilever beam for the active vibration control.Firstly, the Euler-bernoulli beam theory is used to formulate the differential equation ofmotion of a piezoelectric cantilever beam. Then, the basic fuzzy control algorithm is proposedas an active controller. The fuzzy logics are widely used in control field, as the fuzzy systemshave no need for the precise mathematical model. But the fuzzy systems do not have the abilityto learn from the input signals. To promote the intelligence and adaptively of the fuzzyalgorithm we discussed neural networks control algorithm. In order to combine the advantagesof neural networks and fuzzy systems the FNN algorithm is presented.However, the FNN usually takes much time to converge, as a fact of that, we proposed anoptimal training of FNN based the learning rate to solve this problem. The optimal trainingmethod chooses the optimal learning of each training epoch to maximize the speed of errorconverge and to minimize the total error. What’s more, this new optimal training can be used inreal time control systems thanks to the simple and easy to calculate.In this paper, we presented how to test the effective of our FNN and optimal training asthe active vibration control algorithm based on the piezocremics cantilever beam model. Oursimulation was based on the MATALB and SIMULINK. According to the simulation result,our algorithm shows a good performance.