A fuzzy logic ramp metering algorithm will address the needs of Seattle's freeway system and overcome limitations of the existing ramp metering algorithm. This project progressed toward implementing and testing a fuzzy neural ramp metering algorithm on-line at the Traffic Systems Management Center (TSMC) for the Washington State Department of Transportation's Northwest District. Improvements were made to neural network predictors to allow better generalization. Code was written for the fuzzy ramp metering algorithm and its interface with the pre-existing TSMC code. Of the new code written, approximately 95 percent of it was for the interface, and only 5 percent of it was for the ramp metering algorithm itself. Interfacing the fuzzy controller with the existing TSMC software required modification of 16 pre-existing files related to the ramp metering database, real-time skeleton, and ramp metering and data collector communications. A method was developed and code was written to directly send metering rates from the VAX to the 170 and to implement them, whereas previously only a metering rate adjustment had been possible. The operator interface was designed and code was written to enter fuzzy tuning parameters and fuzzy equations. The specifications for each new parameter were designed. Although this code was written, it has not yet been implemented on-line because of time constraints. Preparation for on-line implementation required more time than anticipated because of the unexpected complexity of the pre-existing TSMC code. On-line implementation and testing will proceed on a WSDOT/TransNow project that begins in September 1997. In addition to software design, further planning was necessary to ensure smooth implementation and quality performance. The testing plan was developed in greater detail to include software quality testing. Primary and backup study sites were chosen, and an evaluation technique was selected. A risk assessment plan was developed to mitigate future problems.
December 10, 2007
Cynthia E. Taylor, Deirdre R. Meldrum.
Washington State Transportation Center (TRAC)
- # of Pages: 70 p., 1,785 KB (PDF)
- Subject: Algorithms, Fuzzy logic, Implementation, Neural networks, Ramp metering, Risk assessment, Testing, Traffic control centers.
- Keywords: Artificial neural networks (ANN), fuzzy logic controller (FLC), traffic data prediction, ramp metering, research, Seattle (Wash.)
- Related Publications:
This abstract was last modified April 29, 2008