Automatic sedimentary microfacies identification from logging curves based on deep process neural network

An automatic sedimentary microfacies identification technique is developed based on the deep process neural network (DPNN), which consist several neurons and general non-time-varying neurons arranged in a certain topological structure. In this technique, the features of the shape and amplitude of logging curves are considered to form the category outputs. Combined with the deep learning theory, the diversity of the process features of logging curves and the complexity of combined features of multiple geophysical logging information are considered, and DPNN is created through the stacked superimposition of deep belief network and BP classifier. The technique maintains the structure and information relevance of process signal data and can characterize the distribution features of logging curves automatically, and classify the process signals directly. The theoretical nature and performance of the improved algorithm is tested and validated by some field examples.