Design and Test on Fuzzy Neural Network of Constant Deceleration

One of the main reasons of lag in braking control technology of hoist exchange drag is that braking torque is not adjustable. And the braking torque of speed closed-loop of fuzzy neural network controller (NNSOC) can not only ensure that deceleration in braking is constant, but also it can be regulated in the wider framework. The coil current of nozzle flapper valve of pressure closed-loop of NNSOC changes with the change of braking torque. Mze–I characteristic curve of the valve was simulated to obtain the braking torque expectation (Mze). NNSOC adopted 3-layer structure of BP network, and network input / output samples were based on fuzzy rules and collected by simulation test. After doing repeated experiments, the experimental results show that that traingdx learning function and the best network structure can adjust network weights to memory and update the rules of NNSOC. The network testing results show that the response characteristics of NNSOC (braking torque and its overshoot of control current, network response time) can ensure that the braking deceleration is constant and the braking is smooth, safety, utility, energy-saving.