Abstract: Considering that the drawbacks of the distance coefficient and the Fisher information matrix for optimal sensor placement, a mode contribution and distance coefficient are used to modify the model error covariance in the Fisher information matrix. Firstly, the relationship between the Fisher information matrix and the information entropy is illustrated. Secondly, considering the impacts of the prediction error on the Fisher information matrix, the Euclidean distance and the mode contribution coefficient are employed to modify the model error covariance matrix. Finally, the sensor placements are obtained by maximizing the determinant of the so-modified Fisher information matrix by using the sequential forward algorithm. According to three evaluation criteria, the efficiency of the different modification methods is compared using a truss model. The results showed that the proposed method could obtain better evaluation values and effectively avoids the aggregation of the sensor placements at the same time.