The problem of sensor optimization placement in bridge structural health monitoring systems is investigated in this paper, and an improved COOT algorithm is proposed based on the COOT algorithm which is a new population intelligence optimization algorithm. The Logistic-improved chaos, the Opposition-based learning and the Metropolis criterion are introduced into the population initialization of the algorithm, random movement of ordinary individuals and following movement to improve population diversity and avoid the algorithm falling into locally optimal solutions, respectively. Finally, the effectiveness of the improved COOT algorithm is substantiated by studying the vibration sensor optimization placement of a cable-stayed bridge.