Evolutionary algorithm for vehicle driving cycle generation.

Area of Science:

• Engineering
• Environmental Science
• Computer Science

Background:

• Modeling transit bus emissions and fuel economy requires extensive experimental data across various operational conditions.
• Chassis dynamometer tests use driving cycles composed of "microtrips" (intervals between stops) to represent real-world operation.
• Current methods of generating test cycles are expensive and time-consuming due to the vast combinations of operational parameters.

Purpose of the Study:

• To propose a novel methodology for constructing transit bus driving cycles with specific parameter values.
• To reduce the experimental cost and time associated with chassis dynamometer testing.
• To enable the creation of customized driving cycles from a limited set of standard tests.

Main Methods:

• A methodology is presented for building driving cycles by concatenating "microtrips" from standard test cycles.
• A customized evolutionary algorithm is employed to select appropriate "microtrips" based on desired cycle parameters.
• The algorithm utilizes genetic representation, specific mutation/crossover operators, and Roulette-Wheel selection with elitism to optimize microtrip selection.

Main Results:

• The proposed methodology allows for the generation of driving cycles with prescribed independent variable values.
• The evolutionary algorithm effectively selects "microtrips" to minimize deviations from desired overall cycle parameters.
• This approach facilitates the creation of a wide range of driving cycles from a limited experimental dataset.

Conclusions:

• The developed methodology offers an efficient and cost-effective way to generate diverse transit bus driving cycles.
• This technique supports more comprehensive modeling of bus emissions and fuel economy under various operational conditions.
• The utility is integrated into the West Virginia University's Integrated Bus Information System.