Abstract: MDM(Minimum Distance Method) is a very familiar algorithm in state recognition. But it has a presupposition, that is, the distance within one class is short and the distance between classes is long. When this presupposition is not satisfied, a mistake is made. In order to overcome the shortcomings of MDM, an im-proved minimum distance method based on ANN(Artificial Neural Networks) is presented. The simulation re-sults demonstrate that this method has two advantages, that is, the rate of recognition is fast and the accuracy of recognition is high.