| Geomagnetic positioning has become a hotspot in the underground engineering positioning technology in recent years.And it is necessary to study the evaluation methods of underground adaptation because of the geomagnetic adaptability evaluation is the basis of geomagnetic matching positioning and navigation in the tunnels.In the experiment,different types of tunnels are selected as the research area such as the tunnels of simulated mines and the gold mine,and FVM-400 fluxgate magnetometer was used to measure geomagnetic data.Construction the geomagnetic reference map after denoising the geomagnetic data of 41 samples of the underground projects.Selected the geomagnetic features such as mean,variance,kurtosis coefficient,coefficient of skew,geomagnetic roughness,rough variance ratio,magnetic information entropy and coefficient of association to evaluate the adaptability of the tunnels.Then carried out the experimental research on the initial weight determination method of BP based on the equivalent weight analysis,multiple regression analysis,information entropy analysis and partial least square principle.The results show that the accuracy of BP network based on entropy weight and equivalence is approximately the same,but the training time is longer;The accuracy of the partial least squares BP network is low though it has simple structure and fast convergence;The BP evaluation method which based on multiple regressions that has the highest accuracy and short training time,and it can avoid falling into local optimality to some extent,and it is more suitable as the preferred method for the initial weight factor of the geomagnetic adaptive BP evaluation method.The evaluation result is the geomagnetic map in the weak adaptational tunnels.A convolution enhancement operator of the geomagnetic of tunnel is constructed in this paper,which is used to preprocess the convolution enhancement of the tunnels.The results indicate that convolution processing of Laplace operator can enhance the recognition characteristics of these data,and show strong robustness,which is suitable for data preprocessing model of underground geomagnetic matching.Figure 24;Table 28;Reference 44... |
