CAJ | 학술논문

提出利用浮动车出行数据对城市路网多OD对logit随机路径选择模型参数估计方法。首先采用相对阻抗，实现logit模型参数无量纲化和单一化，将离散多项logit模型转化为相对阻抗的连续概率模型；然后对多OD对的浮动车数据进行分区和权重处理，计算得到概率分布；最后在推导连续型概率密度和累积概率函数基础上，利用最小二乘法进行参数估计。以广州市浮动车数据对方法进行验证，结果拟合优度高，所得参数可有效反映实际路径选择行为。方法简单实用，特别适用于利用海量浮动车数据进行路径选择行为分析。
제출이용부동차출행수거대성시로망다OD대logit수궤로경선택모형삼수고계방법。수선채용상대조항，실현logit모형삼수무량강화화단일화，장리산다항logit모형전화위상대조항적련속개솔모형；연후대다OD대적부동차수거진행분구화권중처리，계산득도개솔분포；최후재추도련속형개솔밀도화루적개솔함수기출상，이용최소이승법진행삼수고계。이엄주시부동차수거대방법진행험증，결과의합우도고，소득삼수가유효반영실제로경선택행위。방법간단실용，특별괄용우이용해량부동차수거진행로경선택행위분석。
This study proposes an algorithm to estimate the parameter of Logit route choice model in urban road network with multinomial OD pairs utilizing taxi-floating car data. The relative route impedance is used and the Logit diversion parameter for whole network can be described by one unified dimensionless parameter. The discrete route choice probabilities are converted to continuous form with consideration of the weight of the numbers of trips, and a cumulative function relationship between route choice probability and relative impedance is re-derived from the original Logit route choice model. The least squares method is utilized to estimate of the unified parameter. The numerical example from taxi GPS data of Guangzhou shows that the goodness fit is excellent and the proposed algorithm is effective and convenient to model the route choice behavior with available and massive floating car data.