Intelligent early warning model of early-stage overflow based on dynamic clustering

During conventional drilling, the early warning of overflow can be realized by monitoring the total variation of drilling fluid. However, this may pose potential safety risks due to the technical challenges it is faced with such as the serious lag, the low accuracy and the incapability to give an early warning in terms of early-stage overflow. In this paper, a new model for early warning of early-stage overflow has been creatively proposed. The proposed model is characterized by surface detection techniques that are different from existing surface inspection techniques. It is able to give an earlier, faster and more accurate warning. In addition, real-time correction of the instantaneous discharge flow can be achieved. The proposed model is established by employing pattern identification and K-mean dynamic clustering. After clustering and linear fitting, the fitting results are compared with the overflow identification sensitivity so as to determine the occurrence of overflow. The experimental results show that the early warning model proposed has overcome the hysteresis and low accuracy of conventional overflow monitoring methods and is capable of early warning of early-stage overflow.