An ANN-Based Precision Compensation Method for Industrial Manipulators via Optimization of Point Selection

Table 2

The parameter setting of the proposed method.

No.	Stage	Parameter name	Parameter value
1	Point selection based on the QRS method	Sequence name	Halton
2		Total number of points	2000
3		Number of points used in the POE model	80
4	Training data selection based on trajectory similarity	Number of track division points	50
5		Number of the nearest distance points	10
6		Total number of point selection	500
7	Training process of the ANN model	Number of samples	450
8		Number of validation samples	50
9		Network type	BP
10		Neurons of input layer	6
11		Neurons of hidden layer	4
12		Neurons of output layer	6
13		Learning rate	0.12
14		Training times	1200
15		Activation function	Tan-sigmoid