Simulation of Micro Stage System Based on BP Neural Network Control - Advanced Research on Computer Education, Simulation and Modeling - page 137

3.2 Simulation Stage

The simulation stage of the algorithm is given as follows:

(i) A three layered neural network based on the BP method used is
shown in Fig.3. The initial values of weight coefficients in each layer are given while
1

k =

.

i

l

j

x

K

P

x

K 2

I

x

K

M

D

Fig. 3. BP network structure

yk are obtained by sampling and the error can be
calculate by equation (6) at the moment.

rk and () ()

(ii) The values of

e ( k ) = r ( k ) − y ( k )

.

(6)

(iii) The inputs and outputs of neurons in each layer are calculated. The three
adjustable parameters

K , K and K of PID controller are the output of output

layer of neural network.
(iv) The output

uk of PID controller is calculated by equation (5).
(v) The self-adaption adjustment of the control parameters in PID controller
are achieved by learning and adjusting weight coefficients on line in neural
network.
(vi) Let

()

k = k + 1

and return to (i).

3.3 Simulation Results

The simulation is done using Matlab software according to the above control
algorithm and the simulation result is shown in Fig. 4. It can be seen from the
simulation curve in Fig. 4 that there are no oscillation, no overstrike and smooth, and
that the adjustment time is less than 0.05 second. So the control performance of PID
controller is good and the expected effect is achieved.