first I have been control the speed of dc motor using 8051 microcontroller (open loop control ) to get the reading , this is the reading:
pwm values:
Input={32,37,42,47,52,57,62,67,72,77,82,92,97,102,107,112,117,122,127,132, 137,142,152,162,172,182,192,202,212,228,233,238,243,248,253}
speed values: output={0,38,52,72,84,104,115,133,141,158,165,192,204,212,225,231,244,248,254,261,270,276,286,306,315,330,336,348,357,366,371,377,379,380,380}
Then , these reading gave it to neural network to learning , the structure of ANN is one input(PWM data have 35 data) and one output (speed data have also 35) and the hidden neuron number between3 to 20 , and I used backpropagation method to train NN after trained I wanna give the update weight and bias after many epochs get to microcontroller , till now i use NN as identification , then after that i use NN as control by inverse identification in microcontroller so speed it will be the input to NN and the output is pwm .