The detection of continuous outliers for sliding windows is an important problem in data stream management, which plays an important role in many fields such as credit card fraud detection, network intrusion prevention, and early warning for geological hazards. Most of the existing algorithms require the use of the range query to determine the positional relationship between objects, but the cost of the range query is usually high, which cannot meet real-time requirements. Therefore, this study proposes the grid-based excepted heap (GBEH), a query processing framework based on sliding windows. Specifically, GBEH proposes a grid queue based index (GQBI) on the basis of the grid to manage data streams, which maintains the positional relationship between data streams and the temporal relationship of data streams. Furthermore, GBEH proposes an outlier detection algorithm, namely, the priority based heap. This algorithm calculates the mathematical expectation of the number of objects in the cell that is included in the query range by use of the intersection area of the query range and the cell and on this basis, establishes an execution range query based on the min-heap. In this way, it effectively reduces the cost of range queries and achieves efficient detection. Theoretical analysis and experiments verify the efficiency and stability of GBEH.