Study on the PSA-ANFIS approach for inverse design in slope engineering

摘要

As the most complex materials, the design object researched by slope engineering can not be described detailedly by any mechanical model. So, a reverse design approach has been established by the adaptive neuron-fuzzy inference system (ANFIS) which tries to make full use of the successful design experiences in slope engineering in the past. It has been found that the ANFIS-based approach has many disadvantages of artificiality, randomness and irregularity. Therefore, the convergence process of training ANFIS is lengthy and trivial and repetitive. It is not optimized. So it is hardly to be used in the reverse problems in slope engineering. For this reason, the problems on setting the training parameters and model of ANFIS was studied by using orthogonal test design method and PSA, and the approach based on PSA-ANFIS for reverse design in slope engineering was put forward and established. Furthermore, this approach is applied to inverse design the angles and heights of some slopes, and the result shows that the new approach is an excellent inverse design approach.