Provides Sensitivity Analysis, Self-Calibration, and Optimization of Traffic Analysis Tool Frameworks Using a Database of Input Parameters
This architecture for software-assisted calibration facilitates the use of traffic analysis tools that manage a significant number of input and output parameters, such as traffic simulation programs. U.S. Federal Highway Administration guidelines for applying micro-simulation modeling software encourage traffic engineers to embrace calibration, in which they reconcile simulated and field-observed traffic performance. Calibration is crucial to producing accurate forecasts from a model, but the process requires expertise and time. Simulation users nationwide, in various comprehensive surveys, consistently cite the difficulty of calibrating complex simulation models as their biggest challenge. Researchers at the University of Florida have developed an architecture for software-assisted calibration that is practical, flexible, and easy to use. The architecture, called Sensitivity Analysis, Self-Calibration, and Optimization, or SASCO, provides users easy-to-use menus to prioritize relevant input and output parameters; eliminating the need for hours of research, and minimizing the possibility of human error.
Architecture for software-assisted calibration that facilitates the use of traffic analysis tools
- Reduces the amount of time, money, and expertise typically required for calibration and sensitivity analysis
- Presents a user-friendly, pre-vetted menu of input parameters and ranges to choose from, eliminating the need for hours of research or guesswork and the possibility of human error in selecting inappropriate input parameters and ranges
- Seamlessly allows users to customize the amount of calibration to be performed on each input parameter, maintaining unique and customized calibration levels required in most real-world projects
- Presents a user-friendly library of output parameters, allowing users to select and prioritize output parameters quickly and easily
- Provides an easy-to-use library of output parameters which allows calibration efforts to be based on any combination and quantity of outputs, meeting the consistent need for unique and customized levels of calibration
- Allows the user to specify the computer run time in advance, maintaining control of desirable run times
- Automatically produces documentation of all calibration settings requested by the user; providing transparent, step-by-step process documentation coveted by decision-makers and clients
- Facilitates sensitivity analysis of the individual input parameters, providing researchers and practitioners information about how each input parameter affects the analysis
This architecture optimizes calibration by minimizing the discrepancy between simulated and field-measured results. Menus containing pre-vetted input choices and custom selection of relevant outputs and run times allow users to accurately calibrate models based on their selection of the most meaningful parameters. SASCO automatically documents all decisions made during the calibration process, which facilitates standardized output reporting.