Authors Papers Year of conference Themes Organizations To MES conference
|An adaptive random search algorithm for parametric identification of electronic componentsí models
| ||Biryukov V.N.|
| ||Pilipenko A.M.|
|Date of publication|
| ||The features of application of the modern optimization techniques for the problems of parameter extraction of electronic components models have been considered. By the example of parameter identification of SPICE-models of a semiconductor diode it was shown that it is impossible to obtain even a rough estimate of the models parameters while using standard optimization techniques of the first and second orders based on the calculation of derivatives of the objective function. Application of the known zero-order random search algorithm, which does not require calculation of derivatives of the objective function, allows determining the parameters of SPICE-models with an acceptable accuracy but under a sufficiently large time of analysis.|
To improve the efficiency (accuracy and speed of convergence) of the random search algorithm it was proposed the adaptive algorithm based on the use of new non-uniform law of distribution of random numbers and takes into consideration the information about the changes of the parameters defined on each successful iteration for reducing the size of the area of parameters.
The proposed adaptive random search algorithm allows improving the speed of parametric identification by 2 Ė 20 times in comparison with the known algorithm. The relative errors of parameters determination by means of the proposed algorithm are significantly less than the error of the original data. This opens up new possibilities for the development of more accurate models of electronic components.
| ||electronic components, model, optimization, random search, stiff problems.|
| ||Biryukov V.N., Pilipenko A.M. An adaptive random search algorithm for parametric identification of electronic componentsí models // Problems of Perspective Micro- and Nanoelectronic Systems Development - 2016. Proceedings / edited by A. Stempkovsky, Moscow, IPPM RAS, 2016. Part 1. P. 101-108.|
|URL of paper|