基于二型模糊逻辑的交通流量预测

提出了一种改进的模糊c均值聚类算法,该算法将模糊聚类的对象从单值扩展到区间,在构造二型模糊系统时,通过对历史数据的学习提取二型模糊规则,克服了专家方法不能对未知领域提取规则的不足.在此基础上,针对智能交通系统,提出一种新的基于二型模糊逻辑的交通流量预测方法.该方法应用区间型二型模糊集具有上下限隶属度函数的性质构造预测区间,适合于处理具有复杂不确定性的情况.通过隶属度函数可以反映出该区间中预测值的可靠性,从而克服了其他预测方法仅给出单值且稳定性不高的缺点.仿真结果表明,基于二型模糊逻辑的流量预测区间具有较高的准确度,其平均相对误差低于6%.

Traffic Flow Forecast Based on Type-2 Fuzzy Logic Approach

An improved fuzzy c-means clustering algorithm is proposed,in which the fuzzy clustering object is extended from single value to the interval data.In constructing the type-2 fuzzy system,the type-2 fuzzy rules are extracted from learning history data and the deficiency that human expert method can not extract rules in unknown systems is avoided.Based on the proposed algorithm,a novel traffic forecasting method based on type-2 fuzzy logic is proposed for intelligent traffic systems,in which the property that the interval type-2 fuzzy logic set possesses membership functions with upper and lower limit is utilized to create forecasting interval,which are suitable for handling the situations with complicated uncertainties.The reliability of the predicted values in that interval can be reflected through the membership function,thus the limits produced by other forecasting methods that only single value is given and are lack of stability is overcome.Simulation results show that the proposed algorithm has high accuracy and the average relative error is less than 6%.