To Separate overlapped peaks with a personal computer in BASIC language, an accurate peak-finding algorithm and an efficient parameter-estimation algorithm based on the curvefitting method were developed. The peak-finding algorithm consists of the automatic calculation process with the second and third derivatives and the manual correction process where an operator can reject noise peaks and/or change some parameters as he watches calculated peaks and the observed curve displayed on a CRT. For the parameter-estimation, the damped least-squares method was selected through a comparison of three other nonlinear optimization methods (Gauss-Newton, Davidon-Fletcher-Powell and Simplex) with respect to the convergence time, and then, the optimum damping-factor was found experimentally. The developed BASIC program was applied to separate an emission spectrum of Ni-HCL (Hollow-Cathod-Lamp) to Gaussian peaks.
抄録全体を表示