Type-2 fuzzy logic approach for short-term traffic forecasting
The performance of many components in intelligent transportation systems depends heavily
on the quality of short-term traffic forecasts. We propose a new method for forecasting traffic
based on type-2 fuzzy logic. Type-2 fuzzy logic is powerful in handling uncertainties,
including uncertainties in measurements and data used to calibrate the parameters. In our
formulation, the value of a membership function corresponding to a particular traffic state is
no longer a crisp value. Rather, it is associated with a range of values that can be …
on the quality of short-term traffic forecasts. We propose a new method for forecasting traffic
based on type-2 fuzzy logic. Type-2 fuzzy logic is powerful in handling uncertainties,
including uncertainties in measurements and data used to calibrate the parameters. In our
formulation, the value of a membership function corresponding to a particular traffic state is
no longer a crisp value. Rather, it is associated with a range of values that can be …
The performance of many components in intelligent transportation systems depends heavily on the quality of short-term traffic forecasts. We propose a new method for forecasting traffic based on type-2 fuzzy logic. Type-2 fuzzy logic is powerful in handling uncertainties, including uncertainties in measurements and data used to calibrate the parameters. In our formulation, the value of a membership function corresponding to a particular traffic state is no longer a crisp value. Rather, it is associated with a range of values that can be characterised by a function that reflects the level of uncertainty. Day-to-day traffic information is combined with real-time traffic information to construct fuzzy rules. The performance of the prediction procedure based on type-2 fuzzy logic is encouraging. The mean relative error is in the neighbourhood of 12% for occupancies and 5% for flows. A distinct advantage of a type-2 fuzzy logic-based traffic forecasting approach is that it can produce prediction intervals as a by-product of the fuzzy reduction process. Another desirable property of the proposed model is that the fuzzy engine formulated is usually tractable at every step, making it easy to incorporate site-specific information into the model-building process to obtain more accurate results.
IET