Operation of the Neural z-Vertex Track Trigger for Belle II in 2021 - a Hardware Perspective

Abstract To reduce the background z -Vertex Track Trigger estimates collision origin in Belle II experiment using neural networks. The main part is a pre-trained multilayer perceptron. task of this perceptron to estimate -vertex suppress from outside interaction point. For this, low latency real-time FPGA implementation needed. We present an overview architecture and neuronal network preprocessing. also show handling missing input data through preprocessing with specially trained networks implemented hardware. we will results estimation for trigger system. Major update stage utilizing 3D Hough transformation processing step ongoing.