Intelligent Identification Method for Drilling Conditions Based on Stacking Model Fusion

Due to the complex and changing drilling conditions large scale of logging data, it is extremely difficult process data in real time identify dangerous working conditions. Based on multi-classification intelligent algorithm Stacking model fusion, 24 h actual an XX well are classified identified. The divided into standpipe connection, tripping out, in, Reaming, back circulation, drilling, other In fusion model, accuracy integrated base learner compared, confusion matrix multi-condition recognition results output, which verifies effectiveness fusion. variation parameter characteristics different conditions, a real-time condition diagram classification drawn, adaptation rules under summarized. stacking method has good effect connection condition, condition. These three conditions’ accuracy, recall rate, F1 value all above 90%. relatively poor ‘other conditions‘, reach less than 80%.