Pattern Recognition for Oilfield Output Decline Based on Discrete Hopfield Neural Network

Based on the discrete hopfield neural network (DHNN), the decline pattern recognition of oilfield output was researched. A new method to recognize different patterns is proposed in this paper. Firstly, the original output data are assorted into four types corresponding to four patterns of decline by use of a fuzzy C mean value cluster. Then, clustering center vectors are changed to unit vectors and sample sets memorized in the DHNN synchronously and equably are established based on symmetry of the net attractor graph, and spurious stable states can be avoided. The trained DHNN can recognize the decline patterns. The application results show that the decline pattern for a set of data can be recognized exactly.