Low Latency Data Acquisition Design for Real-Time Disruption Prediction

DL-based prediction models have been used to predict plasma disruption. The real-time running of the prediction model needs real-time diagnostic data transferred from Real-time data acquisition (DAQ). In this paper, a real-time data acquisition system is implemented. It supports high sample rates, multi-analog input channels, and low data transmission latency. The key feature of this real-time DAQ system is that it can stream sample data in very low latency over traditional UDP ethernet. The sample rate can reach 2M per channel, and analog input channels can reach 32. Data transmission latency is less than 5.5μs. This data acquisition equipment has good DC and AC accuracy and good real-time performance.