The purposes of the research were to develop a data error detection algorithm and to recommend an improved database that will flag the errors detected by the new algorithm for the WSDOT's electronic surveillance system located at the Traffic Systems Management Center in Seattle. The algorithm compares 20-second volumes and 20-second volume/occupancy ratios with threshold values to determine the reliability of detector data. The algorithm will operate on the central computer system for the electronic surveillance data.
Tests of the algorithm proved it to be very successful at detecting data errors caused by short pulses (hanging off) from detectors, chattering detectors, and intermittent malfunctions of detectors. The algorithm did not detect errors caused by long pulses (hanging on) from detectors. The false alarm rate was very low (12 false alarms out of 3,510 observations).
The research team surveyed existing and potential users of the data collected by the electronic surveillance system. On the basis of the user responses, the research team developed a technique to flag erroneous data in the existing system, recommended flagging procedures for the new computer system being designed, and recommended a database configuration that will meet the needs of WSDOT and outside users.