**Abstract**

The proper design of underground powerhouse support is the key engineering technique to guarantee the safe construction and operation of underground works. By regression analysis of the surrounding rock support parameters of 29 underground powerhouses with a span of 18.0–34.0 m, the empirical formula of the relationship between the support strength of anchor bar, strength-stress ratio, and plant span and the relationship among the support strength of the anchor cable, strength-stress ratio, and plant span are proposed. Furthermore, an intelligent design model for the anchor support of the underground powerhouse was trained by a BP (back propagation) neural network. Research shows that the support strength index of the anchor bolt and the anchor cable of these 29 plants are all distributed around 1.0. Therefore, a support strength index of 0.8–1.2 can be used as a reference for practical engineering support design. Finally, the reliability of the intelligent design model for the anchor support of the underground powerhouse was verified by comparison with actual engineering and support strength index. This shows that the intelligent design model can provide a reference for engineering design and construction.

**Keywords:** underground powerhouse, support strength criteria, strength-stress ratio, BP neural network, intelligent design
