Abstract: In the light of the ambiguity， variety， and complexity of traditional Chinese medicine(TCM) pulse conditions， and the shortcomings of traditional fuzzy cluster methods and backpropagation(BP) neural network methods， a novel method using the extreme learning machine(ELM) was proposed to detect the pulse conditions. This method identifies pulse condition by using the ELM to train and classify the characteristic vectors obtained by the pulse condition. The experimental results show that comparing with the traditional fuzzy cluster methods， BP neural network method and support vector machine method， the accuracy of proposed method is respectively increased by 21 percent， 9 percent and 5 percent， which shows that this is a better pulse condition estimation using proposed method.

Key words: traditional Chinese medicine pulse condition, pulse wave, pulse characteristics extraction, extreme learning machine (ELM), pulse conditions identification