Machine learning based approaches to solving EMI/EMC problems caused by BUCK-BOOST DC-DC converter

Companies apply EMI/EMC tests to the electronic devices they produce within the scope of civil and military standards. The emission test in these tests examines the electromagnetic interference emitted by the devices from the propagation and transmission route. In emission tests, it is obligatory that the emissions remain below the limit values determined by international regulations. If the devices fail these tests, afterwards solutions are both more limited and more costly. Therefore, the problem should be detected at the beginning of the design phase and its solution should be implemented. Within the scope of this thesis, it is aimed to examine the emission tests via machine learning methods in order to keep them below the limit values. In order to analyze here, the LED driver circuit design, which has problems in the literature on EMI / EMC, was made depending on certain input-output values. After the design, measurement results were taken and data sets were created. Emission prediction system was designed according to input-output values and semiconductor parameters for Buck-Boost converter based circuits using machine learning methods. In the designed system, an interface has been designed for displaying the results.