To determine the number of hidden layer nodes in single-hidden-layer feedforward neural network(SLFN), we proposed a new classfication algorithm based on self-organizing neural network using growing-combination structure, called GC-SSONN. Firstly, this method added the hidden nodes and employed PSO to initialize the network. Then a similar-nodes mechanism was used to adjust the SLFN structure dynamically. Finally, we tested the performance of GC-SSONN on four datasets, including three benchmark datasets and one gas-path fault datasets, respectively. The testing results show that this method can automatically determine the number of hidden layer nodes and improve the classification accuracy.