This study addresses the problem of accurate distribution parameter estimation for reliability assessment within technical diagnostic systems for electronic components. We propose a novel adaptive clustering approach using wavelet transformation to improve the robustness and accuracy of parameter estimation for exponential and Weibull distributions in electronic component defect testing. The method specifically addresses challenges in high-noise, small-sample environments by dynamically adapting to cluster center drift over time, thus reducing errors in the testing process. Validation experiments on resistor batches demonstrate the method’s resilience to noise, achieving a signal-to-noise ratio of 1.17 with reduced relative error to a level suitable for practical applications. This robust approach is particularly valuable for automated selection of electronic components used in critical systems, where failure tolerance is low and reliability is paramount. Our results indicate the proposed method significantly enhances the reliability and accuracy of diagnostic processes, establishing it as a practical tool for reliability testing and parameter estimation in automated diagnostic systems for electronic components.