A construction algorithm of WLAN fingerprint positioning databas e based on improved deep belief network (DBN) is presented in this paper.First,select some reference points from the reference points that need to be measured in the field to measure the position coordinates and the received signal intensity,and input the m into the DBN as training data to improve the performance of the DBN.Then,input the position coordinates of other reference poi nts into the trained DBN,and use the data in the output of the DBN as the received signal intensity of those reference points,so as to c onstruct the fingerprint positioning database.Finally, the data of some measured reference points are combined with the data of the rem aining reference points predicted based on the DBN to make up the completed fingerprint positioning database,and the KNN and WKNN positioning algorithms are used to evaluate the construction performance.The experimental results show that the improved D BN algorithm has shorter training time and better comple- mentation of fingerprint database when using the same training dataset.