Abstract: The paper explores the principle and algorithm of Entropy-maximizing Model(EM Model) with gravitational prior probability. It regards transportation distribution with maximum probability as forecasting distribution and describes the traveler’s transportation action on the whole by considering the effect of random factors. The model parameters can be calibrated conveniently in applications. Its forecast is compared with that of Gravitation Model and Growth Factor method through practical example and the result indicates that EM Model overcomes the latter’s limit and its
application will have promising prospect for the forecast of transportation distribution.

Key words: entropy-maximising (EM)model, traffic distribution forecast, gravitational prior probability

摘要： 探讨了引入重力式先验概率的最大熵模型的原理和算法,该模型用于"四阶段
法"的交通分布预测,将发生概率最大的交通分布视为预测的交通分布,隐含考虑了随机因素的影响,从宏观上描述了出行者的交通行为,实际应用中模型参数容易标定.通过实例分析并与双约束重力模型、底特律增长系数模型的预测结果进行比较,结果表明最大熵模型克服了后者的局限,适用性较强,在交通分布预测中具有很好的应用前景.

关键词: 最大熵模型, 交通分布预测, 重力式先验概率 >