Instruments and Systems: Monitoring, Control, and Diagnostics Annotation << Back Increasing the Efficiency of the Radar Equipment Based on Quality Forecasting of the Power Amplifier Unit V.O. KUSHNAREV, V.N. KLYACHKIN The effectiveness of aerospace defense systems is largely determined by the quality of the functioning of radar equipment. To increase the efficiency of its operation, it is necessary to be able to predict the expected values of the power of the amplifier unit with the specified characteristics of the electrovacuum device – klystron – received by the enterprise. This is the task of regression analysis. If standard least-squares regression analysis methods do not provide the necessary accuracy, neural networks or machine learning methods such as support vector analysis, decision tree boosting, or random forest can be used. By adjusting the hyperparameters of these methods, the necessary forecasting accuracy is ensured and the best of the four listed methods is selected. In this example, gradient boosting turned out to be the best method: the prediction error on the test sample decreased by 1.4–1.9 times compared to the least squares method at different levels of the frequency range. When a new klystron arrives at the enterprise according to its characteristics, the developed program allows you to predict the value of the output power and assess how efficiently the radar equipment will work in this case. Keywords: klystron, regression model, machine learning, neural network, boosting, random forest, support vector method, hyperparameters. DOI: 10.25791/pribor.8.2025.1605 Pp. 33-39.