Incremental Completion Algorithm Based on Traffic Big Data

Due to the failure of detectors and communication, the lack of traffic data is inevitable, and the phenomenon of data lack in traffic will have a great impact on the performance of intelligent transportation systems. For missing traffic data, scholars have proposed many completion methods, but most of the completion methods are static algorithms. For this reason, this paper proposes an incrementally updated tensor completion algorithm for dynamic data completion. The algorithm uses random singular value decomposition and incremental update principles in the construction phase of the data model, taking into account time complexity and computational efficiency, when new data comes in, it can perform incremental completion on the original data. Experiments on real data sets show that the tensor completion algorithm based on incremental update can save computing resources as new data enters, and is better than other existing completion methods in terms of computing speed.