Study on Bayesian Network Based Fault Diagnosis Method for Drilling Site Data Collector
Focusing on the fault characteristics of data collector in drill well site,fault nodes and causal relationships are determined with correlation function analysis and prior knowledge,a Bayesian networks(BN)structure is preliminarily constructed.BN structure is optimized and parameters is learned with K2 learning algorithm and expectation maximization algorithm(EM),respectively.The fault diagnosis model of drilling data collector is obtained,the model is validated with historical data.On the basis,with comprehensively consideration of multiple factors affecting the failure of drilling data collectors,temperature and humidity are confirmed as the affect factors,the prediction model for drilling data collectors is established,and the influence mechanism of collector fault is simulated and analyzed.
NETL
NSTL
万方数据
