Guass-Newton algorithm based on l_p data fitting and its applications

Robust regression analysis and minimal residual error analysis are two aspects of data processing of dynamic measurement. The Gauss-Newton algorithm and its application on lp data fitting are described. The connection is established between lp criterion and least square criterion based on Gauss-Newton method when Lp norm criterion changed. Experimental results show that the algorithm can be easily realized. Having a clear structure, this algorithm is easily to be programmed by computers and it can be successfully applied in data processing of dynamic measurement, of which on dynamic measurement, its get reasonable globe optimization value.