The in situ stress distribution is one of the driving factors for the design and construction of underground engineering. Numerical analysis methods based on artificial neural networks are the most common and effective methods for in situ stress inversion. However, conventional algorithms often have some drawbacks, such as slow convergence, overfitting, and the local minimum problem, which will directly affect the inversion results. An intelligent inverse method optimizing the back-propagation (BP) neural network with the particle swarm optimization algorithm (PSO) is applied to the back analysis of in situ stress. The PSO algorithm is used to optimize the initial parameters of the BP neural network, improving the stability and accuracy of the inversion results. The numerical simulation is utilized to calculate the stress field and generate training samples. In the application of the Shuangjiangkou Hydropower Station underground powerhouse, the average relative error decreases by about 3.45% by using the proposed method compared with the BP method. Subsequently, the in situ stress distribution shows the significant tectonic movement of the surrounding rock, with the first principal stress value of 20 to 26 MPa. The fault and the lamprophyre significantly influence the in situ stress, with 15–30% localized stress reduction in the rock mass within 10 m. The research results demonstrate the reliability and improvement of the proposed method and provide a reference for similar underground engineering.

The underground powerhouse of a hydropower station, in the form of a cavern group, is generally characterized by a large scale and complicated spatial structure. During the construction phase, extensive excavation in limited underground space may cause a multi-cavern effect between adjacent caverns and thus lead to deformation and failure of the surrounding rock mass, which undoubtedly compromises cavern stability and construction safety. This paper takes the drainage gallery LPL5-1 in the Baihetan underground powerhouse (adjacent to the main powerhouse) as a case study. During the excavation of the main powerhouse, the shotcrete at the upstream arch of LPL5-1 cracked, ballooned and peeled off. After field investigation and numerical simulations, the stress evolution induced by excavation is studied and the failure mechanism is analyzed. The results indicate that the multi-cavern effect led to the surrounding rock mass failures in LPL5-1, which is related to the continuous excavation of the main powerhouse and the resultant extensive stress adjustment. During the main powerhouse excavation, a stress concentration zone was generated at the upstream arch and was intensified with the excavation progressed. The expanded stress concentration zone affected LPL5-1 and made its surrounding rock mass split, thus causing the shotcrete cracking.

To investigate the potential use of the complex function theory in displacement back analysis, a novel method for parameter identification is proposed by combining the complex function theory and the back-propagation neural network optimized by the particle swarm optimization algorithm. The finite element method is replaced with the complex function method to establish a nonlinear relationship between parameters and deformations of the surrounding rock around a complex underground cavern. As demonstrated by a virtual example of an arched tunnel, the deformation parameters were identified by the proposed method and demonstrated to be approximately equal to the predefined parameters. Subsequently, the proposed method is further applied to the inverse analysis of mechanical parameters for the surrounding rock in the underground powerhouse of the Baihetan Hydropower Station. The computed displacements based on the back-analyzed parameters show excellent agreement with the monitoring displacements. The average error between them is about 4.17%, so the proposed method provides the potential to viably enhance the back-analysis technique.