Study on Downhole Geomagnetic Suitability Problems Based on Improved Back Propagation Neural Network

The analysis of geomagnetic suitability is the basis and premise matching navigation positioning. A evaluation model using mixed sampling an improved back propagation neural network (BPNN) based on gray wolf optimization (GWO) algorithm by incorporating dimension learning-based hunting (DLH) search strategy was proposed in this paper to accurately assess suitability. Compared with traditional model, its generalization ability accuracy were better improved. Firstly, key indicators labels used for analyzed, system established. Then, a method synthetic minority over-sampling technique (SMOTE) Tomek Links employed extend original dataset construct new dataset. Next, divided into training set test set, according 7:3. standard deviation, kurtosis coefficient, skewness information entropy, roughness, variance correlation coefficient as input put DLH-GWO-BPNN output. Accuracy, recall, ROC curve, AUC value taken indexes. Finally, PSO (Particle Swarm Optimization)-BPNN, WOA (Whale Optimization Algorithm)-BPNN, GA (Genetic GWO-BPNN algorithms selected compared methods verify predictable DLH-GWO-BPNN. ranking five models follows: PSO-BPNN (80.95 %) = WOA-BPNN (80.95%) < GA-BPNN (85.71%) (95.24%). results indicate that can be reliable underground research, which applied research navigation.