Aiming at the phenomenon of non-Gaussian noise and mising sensor measurement information in complex sensor network environment,a distributed Kalman filter is designed based on the fusion weights.In this paper,the maximum correntropy criterion is utilized to design the Kalman filter to address the effects of non-Gaussian noise.To describe the phenomenon of missing sensor measurement information,a set of binary random variablesis modeled as a Bernoulli distribution.In order to achieve consensus among sensor nodes,covariance intersection method is used for distributed fusion.In addition,considering that it is difficult to improve the accuracy of information fusion with the traditional weight design method,the weights are fully distributed and adaptively designed in the form of proportions of the sensor's confidence level through the introduction of the virtual estimation error and confidence level function.The simulation experiment shows that the proposed algorithm exhibits good performance in non-Gaussian noise and packet drops,and improves the estimation accuracy by 27.68%compared with the traditional distributed Kalman filter.