In this paper, an analysis of existing automated test systems (ATS) and corresponding information technologies (IT) of parameters processing was carried out.
At the heart of the IT and processing methods (identification, segmentation, clustering, classification) is the optimization of the corresponding functional. The analysis showed that for small-scale production the objective function is multimodal and noisy. In gradient processing techniques in such circumstances there is low noise immunity, methods are sensitive to local extreme and the initial search point, in subgradient methods – high error.
To reduce these shortcomings authors proposed methods of identification, classification, clustering, pre-processing and evaluation of product parameters. They are based on the evaluation of the respective functional extremum coordinates using wavelet transform (WT).
The results showed that the proposed methods have improved the clustering quality and classification reliability under a priori uncertainty of diagnostic parameters. On the basis of these methods is proposed to develop the selection of components for the IT equipment in critical applications in radio equipment manufacturing.