Performance Study of Hard Rock Cantilever Roadheader Based on PCA and DBN

With the wide application of cantilever roadheader in urban subway tunnel construction, accurate prediction of excavation performance of cantilever roadheader in rock stratum has become a research hotspot. Accurate prediction of tunneling performance of cantilever roadheader in rock stratum is the key to its successful application in tunnel engineering. Based on Guiyang Rail Transit Line 1 and Line 3, this paper conducts field investigation and statistical analysis of data on the construction performance and tunneling characteristics of roadheader, and establishes a prediction database of tunneling performance of hard rock cantilever roadheader. The principal component analysis (PCA) was introduced into the deep belief network (DBN) to optimize the input parameters of the DBN model, and the PCA-DBN model for the performance prediction of hard rock cantilever roadheader was proposed. The new model is trained and predicted based on the data of Guiyang Rail Transit Line 1, and the rationality and feasibility of the model are verified through the field data test and analysis of Guiyang Rail Transit Line 3. The results show that the performance prediction model of hard rock cantilever roadheader based on PCA-DBN can realize real-time and continuous prediction of tunneling performance of ground roadheader in front of tunnel face according to engineering measured data. The comparative analysis with the DBN model shows that the accuracy of the PCA-DBN prediction model is better than that of the DBN model, which can better adapt to complex and changeable geological conditions. The new model provides a new method and possibility for accurately predicting the tunneling performance of cantilever roadheader in hard rock. A prediction database of hard rock cantilever roadheader tunneling performance was established.The performance prediction model of hard rock cantilever roadheader based on PCA-DBN was established.The performance prediction model realizes real-time and continuous prediction of tunneling performance of ground roadheader.The prediction accuracy of PCA-DBN is higher than DBN in predicting the performance of cantilever roadheader.