First, the Ball Forward Model Identification System

BEL model mill system identification algorithm

Use BEL model ball forward model and inverse model of the system.

First, the ball forward model identification system

BEL model based online identification ball forward model structure as shown in Fig .

SIF is a function of sensory input which , ECF is a function of emotional cues , u mill systems for the control input , y for the ball mill system output , y is the identification BEL model output , e is the identification error . BEL model in parallel with the mill , the output error is used to adjust BEL model link weights.

BEL away from the actual system must first determine the signal input SI and REW function is formed , which the system error of the input and output . Select input SI system generally u, the error signal e and the system output y of the function , REW the error signal e, BEL model output MO or function of the system output y , REW is the error signal e, BEL model output MO or the system output y function , select the SI and REW form are as follows :

SI = F (u) = k1u (t) REW = J (e) = k2e (t) + k3 ∫ e (t) dt where , k1, k2, k3 is the weight coefficient adjustment .

Second, the ball against the cross -type identification system

BEL model based on the structure of the model identification ball forward as shown in the following figure .

System output y and the error e BEL through SIF and ECF as input , after emotional learning , BEL model output and demand both system input signal o u converge. In the inverse model identification mill system configuration, select SI and REW form below:

SI = F (y) = {r1y (t) r2y (t)} REW = J (e) = r3e (t) + r4 ∫ e (t) dt where , r1, r2, r3, r4 is the weight adjustment coefficient .

BEL identification system using ball algorithm step-outs as follows :

1 , the input signal SI model set BEL REW and weight adjustment factor and k1 ~ k2, r1 ~ r4, BEL model learning rate weights a, β, and the orbitofrontal cortex Myolie body weights...