Abstract: Accurately forecasting the operational performance of a tunnel boring machine (TBM) in advance is useful for making timely adjustments to boring parameters, thereby enhancing overall boring efficiency. In this study, we used the Informer model to predict a critical performance parameter of the TBM, namely thrust. Leveraging data from the Guangzhou Metro Line 22 project on the big data platform in China, the model’s performance was validated, while data from Line 18 were used to assess its generalization capability. Results revealed that the Informer model surpasses random forest (RF), extreme gradient boosting (XGB), support vector regression (SVR), k-nearest neighbors (KNN), back propagation (BP), and long short-term memory (LSTM) models in both prediction accuracy and generalization performance. In addition, the optimal input lengths for maximizing accuracy in the single-time-step output model are within the range of 8–24, while for the multiple-time-step output model, the optimal input length is 8. Furthermore, the last predicted value in the case of multiple-time-step outputs showed the highest accuracy. It was also found that relaxation of the Pearson analysis method metrics to 0.95 improved the performance of the model. Finally, the prediction results were most affected by earth pressure, rotation speed, torque, boring speed, and the surrounding rock grade. The model can provide useful guidance for constructors when adjusting TBM operation parameters.

利用Informer模型预测隧道掘进机性能：广州地铁项目案例研究

[1]AsterisPG, TsarisAK, CavaleriL, et al., 2016. Prediction of the fundamental period of infilled RC frame structures using artificial neural networks. Computational Intelligence and Neuroscience, 2016:5104907.

[2]AsterisPG, NozhatiS, NikooM, et al., 2019. Krill herd algorithm-based neural network in structural seismic reliability evaluation. Mechanics of Advanced Materials and Structures, 26(13):1146-1153.

[3]AsterisPG, LemonisME, LeTT, et al., 2021a. Evaluation of the ultimate eccentric load of rectangular CFSTs using advanced neural network modeling. Engineering Structures, 248:113297.

[4]AsterisPG, SkentouAD, BardhanA, et al., 2021b. Soft computing techniques for the prediction of concrete compressive strength using non-destructive tests. Construction and Building Materials, 303:124450.

[5]BenestyJ, ChenJD, HuangYT, et al., 2009. Pearson correlation coefficient. In: Cohen I, Huang YT, Chen JD, et al. (Eds.), Noise Reduction in Speech Processing. Springer, Berlin, Germany, p.1-4.

[6]BengioY, SimardP, FrasconiP, 1994. Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks, 5(2):157-166.

[7]BilginN, YükselA, 2023. The effect of EPB face pressure on TBM performance parameters in different geological formations of Istanbul. Tunnelling and Underground Space Technology, 138:105184.

[8]CarterNJ, SchwertmanNC, KiserTL, 2009. A comparison of two boxplot methods for detecting univariate outliers which adjust for sample size and asymmetry. Statistical Methodology, 6(6):604-621.

[9]ChenXS, FuYB, ChenX, et al., 2022. Progress in underground space construction technology and technical challenges of digital intelligence. China Journal of Highway and Transport, 35(1):1-12 (in Chinese).

[10]EmadW, MohammedAS, BrasA, et al., 2022. Metamodel techniques to estimate the compressive strength of UHPFRC using various mix proportions and a high range of curing temperatures. Construction and Building Materials, 349:128737.

[11]GaoBY, WangRR, LinCJ, et al., 2021. TBM penetration rate prediction based on the long short-term memory neural network. Underground Space, 6(6):718-731.

[12]HajihassaniM, AbdullahSS, AsterisPG, et al., 2019. A gene expression programming model for predicting tunnel convergence. Applied Sciences, 9(21):4650.

[13]HassanpourJ, 2018. Development of an empirical model to estimate disc cutter wear for sedimentary and low to medium grade metamorphic rocks. Tunnelling and Underground Space Technology, 75:90-99.

[14]LiJH, LiPX, GuoD, et al., 2021. Advanced prediction of tunnel boring machine performance based on big data. Geoscience Frontiers, 12(1):331-338.

[15]LiZM, Yazdani BejarbanehB, AsterisPG, et al., 2021. A hybrid GEP and WOA approach to estimate the optimal penetration rate of TBM in granitic rock mass. Soft Computing, 25(17):11877-11895.

[16]LiuQS, HuangX, GongQM, et al., 2016. Application and development of hard rock TBM and its prospect in China. Tunnelling and Underground Space Technology, 57:33-46.

[17]LuZP, ShiKB, 2023. A novel VMD-LHPO-KELM machine learning-based TBM boring parameter prediction. Earth Science Informatics, 16(3):2925-2938.

[18]MahmoodzadehA, NejatiHR, MohammadiM, et al., 2022. Forecasting tunnel boring machine penetration rate using LSTM deep neural network optimized by grey wolf optimization algorithm. Expert Systems with Applications, 209:118303.

[19]PanYC, LiuQS, LiuQ, et al., 2020. Full-scale linear cutting tests to check and modify a widely used semi-theoretical model for disc cutter cutting force prediction. Acta Geotechnica, 15(6):1481-1500.

[20]PsyllakiP, StamatiouK, IliadisI, et al., 2018. Surface treatment of tool steels against galling failure. MATEC Web of Conferences, 188:04024.

[21]RostamiJ, 1997. Development of a Force Estimation Model for Rock Fragmentation with Disc Cutters Through Theoretical Modeling and Physical Measurement of Crushed Zone Pressure. PhD Thesis, Colorado School of Mines, Colorado, USA.

[22]RostamiJ, 2016. Performance prediction of hard rock tunnel boring machines (TBMs) in difficult ground. Tunnelling and Underground Space Technology, 57:173-182.

[23]ShiQH, SongPF, TanZW, et al., 2022. GA-BP neural network prediction model for tunneling speed of shield machine with composite formation dual mode (TBM-EPB). Proceedings of the International Conference on Computational Infrastructure and Urban Planning, p.1-4.

[24]XuQH, HuangX, ZhangBG, et al., 2023. TBM performance prediction using LSTM-based hybrid neural network model: case study of Baimang River tunnel project in Shenzhen, China. Underground Space, 11:130-152.

[25]XueYD, LuoW, ChenL, et al., 2023. An intelligent method for TBM surrounding rock classification based on time series segmentation of rock-machine interaction data. Tunnelling and Underground Space Technology, 140:105317.

[26]YagizS, 2008. Utilizing rock mass properties for predicting TBM performance in hard rock condition. Tunnelling and Underground Space Technology, 23(3):326-339.

[27]YangY, ZhangQ, 1997. A hierarchical analysis for rock engineering using artificial neural networks. Rock Mechanics and Rock Engineering, 30(4):207-222.

[28]ZhangL, GaoJR, ZhangB, et al., 2015. Application status and prospects for the casting support tunnelling system using TBM. Modern Tunnelling Technology, 52(5):24-31 (in Chinese).

[29]ZhengYL, ZhangQB, ZhaoJ, 2016. Challenges and opportunities of using tunnel boring machines in mining. Tunnelling and Underground Space Technology, 57:287-299.

[30]ZhouJ, Yazdani BejarbanehB, Jahed ArmaghaniD, et al., 2020. Forecasting of TBM advance rate in hard rock condition based on artificial neural network and genetic programming techniques. Bulletin of Engineering Geology and the Environment, 79(4):2069-2084.