Artificial neural network regulator implementation for 50kW half controlled thyristor rectifier

Paper describes solution of regulator for 50kW one phase half controlled thyristor rectifier. Device stabilizes 620V DC link for 3-phase 400V/50Hz invertor for locomotive air compressor driven by induction motor. Wide range and unstable one phase input voltage are typical for application such like this. Conventional PI/PS regulator is not suitable due to long response time. Proposed solution uses feedforward multilayer perceptron artificial neural network for control phase angle estimation. Implementation based on selected method as well as method itself is described. Obtained results and comparison with conventional method are given.