摘要
Rotationally symmetric triangulation(RST)sensor has more flexibility and less uncertainty limits because of the abaxial rotationally symmetric optical system.But if the incident laser is eccentric,the symmetry of the image will descend,and it will result in the eccentric error especially when some part of the imaged ring is blocked.The model of rotationally symmetric triangulation that meets the Schimpflug condition is presented in this paper.The error from eccentric incident laser is analysed.It is pointed out that the eccentric error is composed of two parts,one is a cosine in circumference and proportional to the eccentric departure factor,and the other is a much smaller quadric factor of the departure.When the ring is complete,the first error factor is zero because it is integrated in whole ring, but if some part of the ring is blocked,the first factor will be the main error.Simulation verifies the result of the a- nalysis.At last,a compensation method to the error when some part of the ring is lost is presented based on neural network.The results of experiment show that the compensation will make the absolute maximum error descend to half,and the standard deviation of error descends to 1/3.
Rotationally symmetric triangulation （RST） sensor has more flexibility and less uncertainty limits because of the abaxial rotationally symmetric optical system. But if the incident laser is eccentric, the symmetry of the image will descend, and it will result in the eccentric error especially when some part of the imaged ring is blocked. The model of rotationally symmetric triangulation that meets the Schimpflug condition is presented in this paper. The error from eccentric incident laser is analysed. It is pointed out that the eccentric error is composed of two parts, one is a cosine in circumference and proportional to the eccentric departure factor, and the other is a much smaller quadric factor of the departure. When the ring is complete, the first error factor is zero because it is integrated in whole ring, but if some part of the ring is blocked, the first factor will be the main error. Simulation verifies the result of the analysis. At last, a compensation method to the error when some part of the ring is lost is presented based on neural network. The results of experiment show that the compensation will make the absolute maximum error descend to half, and the standard deviation of error descends to 1/3.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2007年第9期1548-1553,共6页
Chinese Journal of Scientific Instrument
基金
Supported by China Scholarship Council(CSC),Deutscher Akademischer Austausch Dienst(DAAD),National Natural Science Foundation of China(60375011,60575028)
Natural Science Foundation of Anhui Province(04042044)
Supported by Program for New Century Excellent Talents in University(NCET-04-0560)
