Real-time estimation method of reactor state transition probability based on ensemble learning

Abstract

The invention provides a real-time estimation method of reactor state transition probability based on ensemble learning. The real-time estimation method comprises the following steps: evaluating uncertainty of a real-time measurement signal, and establishing a state transition model Fensemble of a reactor based on ensemble learning; establishing an observation model of the measurement signal; andcarrying out real-time prediction on the reactor state based on the uncertainty, the state transition model Fensemble and the observation model, and obtaining the uncertainty of a reactivity prediction state. According to the method, a mathematical model-based model is replaced by a learning-based model, so that rapid reactor state prediction can be realized, the real-time requirement is met, andthe reactor state prediction efficiency can be effectively improved; accurate estimation and quantization uncertainty of a current state and a future state of the system are obtained by combining measurement data and a state transition model.

Description

technical field [0001] The invention belongs to the field of nuclear reactor monitoring and operation support, in particular to a real-time estimation method of reactor state transition probability based on integrated learning. Background technique [0002] Such as figure 1 As shown, the reactor core containing fuel assemblies is deployed in a steel pressure vessel. The coolant flows in from the inlet of the loop, flows down the wall of the pressure vessel, enters the lower part of the core, and flows along the axis of the core. In the height direction, the core heat is gradually raised while taking away the core heat, and then mixed at the core outlet, enters the loop outlet, and enters the heat exchanger for heat exchange. Therefore, from the perspective of coolant temperature monitoring, a large number of thermocouples are arranged at the loop inlet, loop outlet, and core outlet to monitor temperature changes in real time. From the perspective of neutron detection, fixe...

AI Technical Summary

Problems solved by technology

Invention patent "On-line measurement method for three-dimensional power distribution of reactor core" (ZL201610478643.6) achieves conservative estimation of some key safety parameters such as power distribution by means of simplified approximation or conservative estimation, but with great uncertainty In the process of operation, a large penalty factor needs to be superimposed on the basis of conservative estimates, which brings about the conservatism of reactor operation, limits the scope and flexibility of operation, and loses economic efficiency

Images