Fault Intelligent diagnosis of reversible pumped storage Unit based on Bayesian networks and counterfactual reasoning

The fault type and fault characteristics of reversible pumped storage units are intertwined and complex, and how to infer the possible fault type from the existing fault characteristics is the key problem of fault diagnosis research. This paper first completes the construction of a Bayesian network through the collection of fault information in order to complete the tracing of faults.Then, in combination with the counterfactual reasoning method, the calculation formula for each fault is intervened by introducing sufficient and necessary causes. Finally, the exact inference of Bayesian networks is used to calculate the faults and to determine the type of faults by comparing the magnitude of the sufficient and necessary causes between different faults.Taking a pumped storage power plant hydraulic imbalance fault as an example, a method based on the combination of Bayesian network and counterfactual reasoning is used to realize the intelligent diagnosis of its fault, and the diagnosis results match with the actual situation of the unit, which proves the effectiveness of the method in this paper.