Crash risk measuring and analysis based on vehicle kinematic volatilities: a study on a highway in a hilly area

Abstract Measuring crash risks and identifying high-risk sections are crucial for reducing traffic injuries and damage. This study proposed a framework for measuring risks by the volatilities of vehicle kinematics. Generalized autoregressive conditional heteroscedasticity models were developed to calculate kinematic volatilities and identify high-volatility sections. The Technique for Order Preference by Similarity to an Ideal Solution and clustering methods were employed to classify section risks based on historical crash data. The study utilized the generative adversarial network to address data imbalance and compared the performance of three single and three ensemble classifiers. The volatility analysis accurately identified 61, 64, 61, 48, 51, 51, 63, 75, and 69 high-volatility sections for three-axis accelerations, angular velocities, and attitude angles. eXtreme Gradient Boosting exhibited the best overall performance, with an accuracy of 0.899 and macro-average precision, recall, and F 1 score of 0.898, 0.900, and 0.899. Volatilities in <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mi>a</mml:mi> <mml:mrow> <mml:msub> <mml:mi>g</mml:mi> <mml:mi>x</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mi>a</mml:mi> <mml:mrow> <mml:msub> <mml:mi>g</mml:mi> <mml:mi>z</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mrow> <mml:msub> <mml:mi>a</mml:mi> <mml:mi>y</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mrow> <mml:msub> <mml:mi>ω</mml:mi> <mml:mi>y</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mrow> <mml:msub> <mml:mi>ω</mml:mi> <mml:mi>z</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mrow> <mml:msub> <mml:mi>ω</mml:mi> <mml:mi>x</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> , and <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mrow> <mml:msub> <mml:mi>a</mml:mi> <mml:mi>x</mml:mi> </mml:msub> </mml:mrow> </mml:mrow> </mml:math> had high feature importance and showed strong positive effects on the high-risk level. These effects are closely linked to kinematic responses on alignment characteristics and abnormal variations during accidents. The study proposed a method to facilitate the measurement and early inspection of risks by management departments, thus avoiding passive assessment and safety improvements after multiple accidents occur.