Modeling and Error Compensation of Robotic Articulated Arm Coordinate Measuring Machines Using BP Neural Network

Table 2

Simulation results of different parameters of BPNN for the AACMM.

Simulation number	TD	HL	NN	LR	GA	Epoch  [times]	Training time [seconds]	Maximum error [mm]	Average error [mm]
	10000	1	10	0.1		1000	124	25.9080	8.511
	8000	1	10	0.1		516	51	20.4858	8.6132
	5000	1	10	0.1		1000	64	22.6937	9.4585
	20000	1	10	0.1		586	253	22.3574	8.9189
	9000	1	10	0.1		1000	115	20.6444	8.3117
	10000	1	10	0.1		1000	127	26.887	8.8126
	10000	1	10	0.1		1000	126	25.8964	8.1825
	10000	1	10	1		360	123	21.5603	8.7369
	10000	1	10	0.01		1000	125	18.1312	8.1551
	10000	1	10	0.001		1000	125	27.6223	8.8921
	10000	1	10	10		136	17	31.2093	10.1348
	10000	1	3	0.1		1000	93	90.7123	34.1409
	10000	1	4	0.1		293	25	68.0939	26.4297
	10000	1	13	0.1		589	89	14.8641	6.3235
	10000	1	50	0.1		40	21	6.1157	2.5537
	10000	1	100	0.1		17	12	9.4408	2.8841
	10000	2		0.1		38	23	11.5531	2.4639