类BP神经网络的公交一卡通跨区消费推算模型

结合BP神经网络的建模方法,首次提出跨区消费分布矩阵的推算模型,并根据实际应用情况来设置误差阀值和学习率等关键参数,通过训练实际一卡通消费数据,得到跨区消费分布矩阵,实际输出数据误差在允许的范围内.实验表明该模型适用于跨区消费分布情况的预测,可为省域公交一卡通提供重要的决策支撑数据.

Regional Consumption Computing Model of Bus Card Imitating Based on BP Neural Network

The paper proposes a regional consumption distribution estimation model based on the modeling method of BP neural network. It sets the key parameters such as error threshold and learning rate according to the actual situation. By the actual training of card consumer data, it gets the regional consumption distribution matrix, and the error is in the allowable range. The actual test data shows the method can forecast the regional consumption situation, and prove important decision support data ior provincial bus card.