Bridge structures are often of random characters due to the uncertainty of materials and other factors. Therefore， the identified moving loads on bridges based on the responses of the bridge structure are also uncertain. However， there are only a few of relevant researches. In this paper， its proposed to explore the effective identification of the statistical moments of the identified moving loads on bridge structures with random fields. The proposed method is based on the combination of Karhunen-Loeve （KL）expansion of the random fields of bridge structural parameters， Kalman filter with unknown input （KF-UI） and the improved twopoint estimation method （I2PEM）. First， the spatially correlated random fields of bridge structural parameters are transformed into a combination of multiple random variables through KL expansion. Then， in the uncertainty propagation of multiple random variables， the I2PEM is adopted so the uncertain inverse problem is transformed into several deterministic inverse problems. Finally，the identification of moving load based on KF-UI is performed on the selected points of each random variable， and the statistical moments of the moving loads are efficiently estimated. The proposed estimation method is successfully verified by a numerical simulation example.