Integration of knowledge is one of the most important tasks for knowledge management in distributed environments. For the same subject in the real world, each different source may generate different versions of data. Of course, with local integrity constraints these data are consistent, but they may be inconsistent with global integrity constraints. This is a popular phenomenon in processing data from distributed sources in the real world. In this work we will investigate processing inconsistency of knowledge and its integration process using the temporal model of indeterminate valid time and probability. This data model is used to describe events that take place in the future with some certainty degrees. With this model, conflict is considered as a situation in which for the same event agents give different time intervals and probabilities about the occurrence of that event. In the integration process we need to determine a proper time interval and probability (based on consensus method) that properly represent that event. With this aim, two kinds of distance functions are defined as well as analyzed. In addition, some postulates and algorithms for knowledge integration are also worked out and analyzed.