The collapse of surrounding rock mass of tunnels is affected by many factors; the identification of the collapse type is a very complicated nonlinear system identification and it can not be solved by traditional methods. The problem of complicated nonlinear system identification can be solved very well using neural network (NN) model. Considering the existing problems of the traditional NN model and traditional evolutionary neural network(ENN) model and combining the immune evolutionary programming (IEP) proposed by authors with NN

a new ENN model whose architecture and connection weights evolve simultaneously is proposed. Using this new ENN model

the problem to identify the collapse type of surrounding rock mass of tunnel is studied. The results of an example show that the performance of the new ENN is very well and it is a very good method to identify the collapse type of surrounding rock mass of tunnels.