为准确计算动车组牵引能耗,提出BP神经网络模型和改进牵规法预测动车组牵引能耗.选取机车类型、坡度、目标速度、停站方案等8个因素作为动车组牵引能耗的BP神经网络输入变量,建立3层BP神经网络模型.采用增加动车组运动方程和优化基本阻力公式方式对牵规法进行优化.利用正交实验法对动车组牵引能耗影响因素进行分析,并对111组实测能耗进行模拟验证.研究结果表明:BP神经网络模型的实测能耗与计算能耗相对误差在4.26%以内,改进牵规法的实测能耗与计算能耗相对误差基本在10%以内,证明BP神经网络模型比改进牵规法模型能更好地预测动车组的牵引能耗,而且当目标速度增大时,BP神经网络模型的计算精度明显比改进牵规法的计算精度高;目标速度和坡度对牵引能耗有显著影响.%In order to predict the traction energy consumption of the high-speed trains, the back-propagation (BP) artificial neural network model and the optimized train traction calculation procedures were proposed. The input variables of the back-propagation (BP) artificial neural network model were locomotive properties, slope, target speed, plan of stop and so on. And the output variable of the back-propagation (BP) artificial neural network model was the traction energy consumption of the high-speed trains. Compared with the train traction calculation procedures, the optimized train traction calculation procedures considered the train motion equation model and changed the coefficient of the resistance formula equation. The method of orthogonal experiment was used to analyze the influence factors of traction energy consumption, about 111 groups data were calculated by the two models. The result shows that the BP artificial neural network model is more accurate than the optimized train traction calculation procedures. The error between the BP neural network model and the measured value is within 4.26%, and the error between optimized train traction calculation procedures and the measured value is about 10%.When the target speed increases, the precision of BP artificial neural network model is obviously higher than that of the optimized train traction calculation procedures. The target speed and the slope have significant influence on the traction energy consumption.
展开▼
