An assessment of the susceptibility to rock slope failure by means of a back-propagation network is proposed for the eastern portion of the Southern Cross-Island Highway in Taiwan. The model was developed on the basis of six influence parameters of rock slope instability, which include the rock type, slope aspect, slope angle, joint set number, joint spacing and bedding-slope relationship. The values of these influence parameters were used as inputs for the network and were classified as nominal scales in terms of binary numbers, while the state of failure/non-failure of a given slope was assumed to be the output variable. Data on a total of 170 slopes along the highway was fed into the network for learning. According to the outputs of the network, the susceptibility to rock slope failure is categorized into four levels, namely low, medium-low, medium and high, which are mapped along the highway. Three highly susceptible regions are found, which can be viewed as hazardous sections requiring cautionary measures. Moreover, the proposed model can be used as a tool for determining the possible state of an unfamiliar rock slope in the context of devising management strategies to be applied to the investigated portion of the highway.
All Science Journal Classification (ASJC) codes