Predicting the Reliability of Electronic Modules Using Machine Learning Methods

The graduation thesis describes the development of a new methodology for assessing the reliability of electronic components for electronic device development, in the context of the main elements used in the design of modern electronic devices. The reliability of components and devices plays a critical role in ensuring the long-term and uninterrupted operation of technical systems. The key feature of the proposed methodology lies in the application of machine learning methods to generate new reference data on the basic indicators of reliability, durability, and maintainability for calculating the reliability of electronic devices, taking into account the conditions of their application by forecasting reliability indicators based on the physical and electrical parameters of the elements themselves. The work begins with a review of current and existing handbooks on the reliability of electronic components, identifying limitations and imperfections in existing approaches. Next, information on the reliability parameters of electronic components is collected and analyzed with the aim of developing a more accurate and informative mathematical model, validated by physical tests, using machine learning methods to forecast basic reliability indicators with high accuracy. The final result of the work is a software package for determining the reliability of basic electronic circuits. This work has practical significance for manufacturers of modern electronic devices, helping them improve the quality of reliability forecasting for their products, thereby contributing to the advancement of technical development and competitiveness in the market.