A Hybrid Method for Short-Term Traffic Congestion Forecasting Using Genetic Algorithms and Cross Entropy

López-García, Pedro; Onieva, Enrique; Osaba, Eneko; Masegosa, Antonio D.; Perallos, Asier · 2015 · OpenAlex-citations

DOI: 10.1109/tits.2015.2491365

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of short-term traffic congestion forecasting, a critical component of Intelligent Transportation Systems (ITS) aimed at reducing noise, pollution, and travel time. The authors propose a novel hybrid optimization method, termed GACE, which combines Genetic Algorithms (GA) and the Cross Entropy (CE) method to optimize Hierarchical Fuzzy Rule-Based Systems (HFRBS). The motivation stems from the limitations of existing forecasting techniques, such as Neural Networks and Support Vector Machines, which often struggle with local optima, generalizability, or high-dimensional data. The specific goal is to predict congestion levels along a 9-km stretch of the I5 freeway in California with time horizons of 5, 15, and 30 minutes. The methodology employs a Parallel HFRBS (PHFRBS) to handle the "curse of dimensionality" by structuring fuzzy systems into layers, where each unit processes only two inputs. The GACE algorithm optimizes three components of the PHFRBS: the hierarchy of variables, membership functions, and rule bases. In each iteration, the population is split into two sub-populations. One sub-population undergoes standard GA operations (selection, crossover, mutation) to explore the solution space, while the other uses the CE method to exploit promising areas by updating mean and variance parameters based on the best-performing individuals. This hybrid approach leverages the exploration capabilities of GA and the exploitation efficiency of CE. The fitness of each solution is evaluated using the Mean Absolute Error (MAE) between predicted and actual congestion levels. The study utilizes real-world data from the California Performance Measurement System (PeMS), derived from sensors along the I5 freeway. Congestion is classified into four levels (free, slight, moderate, severe) based on traffic density and vehicle speed. The authors conducted comparative experiments evaluating different levels of hybridization, ranging from pure GA to pure CE, and various weightings of the combined techniques. The results demonstrate that the GACE method achieves higher accuracy in predicting short-term traffic congestion than either GA or CE used independently. The hybrid approach effectively optimizes the fuzzy system parameters, providing more reliable forecasts across the tested time horizons. The significance of this work lies in its contribution to the field of traffic forecasting by introducing a robust optimization framework for fuzzy systems. By successfully combining GA and CE, the method offers a solution to the trade-off between exploration and exploitation in complex optimization problems. Furthermore, the use of HFRBS provides interpretable linguistic rules, which are advantageous for operators compared to black-box models. The findings suggest that hybridizing metaheuristic algorithms can enhance the performance of ITS applications, leading to more effective traffic management and reduced societal costs associated with congestion.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• anticipation