Management and prediction of traffic crashes on residential streets in Iraq using the expert system (MPTCRSI-ES)

Expert system techniques have proven effective in predicting traffic accidents, outperforming traditional shallow models. Residential streets, particularly in low- and middle-income countries like Iraq, face significant safety issues, including speeding, cut-through traffic, inadequate infrastructure, and the lack of specialized zones in the current decision-making process. This results in a subjective, expert-dependent approach to traffic management. This paper presents the development of an integrated knowledge-based system (KBS) for predicting and managing traffic accidents on residential streets in Iraq, using the expert system (MPTCRSI-ES). The system consists of eight modules with forward chaining functions and fuzzy rules—logical statements that use linguistic variables (e.g., “high risk,” “low traffic”) to handle uncertainty and imprecision, derived from expert evaluations of street scenarios. Equivalent Property Damage Only (EPDO), a weighted index that accounts for fatalities, injuries, and property damage, was used to measure crash severity more effectively than simple crash counts. Experts assessed the scenarios using Cronbach’s alpha and internal consistency reliability (ICR), achieving a Cronbach’s alpha of 0.960 and excellent ICR reliability. The system was validated through case studies on eight residential street sections in Iraq, with matching results between expert answers and system outputs (Cronbach’s alpha = 0.917, ICR = excellent). End-user evaluations showed an overall rating above 4 (Cronbach’s alpha = 0.932, excellent ICR). The Verification, Validation, and Evaluation results confirmed that the system met its objectives. Simulation results indicate that the proposed system can assist traffic authorities in managing and predicting accidents effectively. Quantitative comparisons show that the MPTCRSI-ESsystem outperforms traditional models by approximately 27 % in expert agreement and predictive accuracy, providing a more reliable and efficient approach to traffic management.