Multi-objective Optimization of Layout of Detectors and Floating Car Datum Requirement for Higher Efficiency of Traffic State Prediction

To optimize the prediction error of speed field and the efficiency of traffic state prediction, a multi-objective optimization method considering the different physical and statistical properties of static detector data (SDD) and floating car data (FCD) is proposed to optimize the layout (the number and the corresponding location) of static detectors and the percentage of connected automated vehicles (CAVs) simultaneously. The optimization result is a set of Pareto optimal solutions providing the best trade-off between the layout of detectors and the percentage of the CAVs with reasonable prediction accuracy for different situations. In the detailed analysis, the comparative results of the predicted speed field before and after optimization indicates that: (1) the proposed optimization method improves the prediction accuracy by optimizing the layout of the detectors and the percentage of CAVs, (2) contradicting to intuitive knowledge, the increasing of the percentage of the CAVs may lead to the deterioration of the prediction accuracy as the FCD is lack of statistical representativeness.