CAJ | 학술논문

采用BP神经网络，对停车场的空余泊位数量进行了预测，并针对BP神经网络训练过程中易震荡、收敛速度过慢和容易陷入局部最小的缺点，采用BP动量法与调节学习速率相结合的方法对其进行改进。通过仿真验证了BP神经网络对停车场空余泊位数预测的有效性，为区域的停车诱导提供了依据。
채용BP신경망락，대정차장적공여박위수량진행료예측，병침대BP신경망락훈련과정중역진탕、수렴속도과만화용역함입국부최소적결점，채용BP동량법여조절학습속솔상결합적방법대기진행개진。통과방진험증료BP신경망락대정차장공여박위수예측적유효성，위구역적정차유도제공료의거。
This paper used the BP neural network for the prediction of parking lot free park-ing number.In the process of the BP neural network training, used the BP momentum meth-od combined with self-adjusting learning rate to improve the basic BP neural network which was easy to appear shocking, convergence speed slowing and trapping in local minimum. The effectiveness of the improved BP neural network on the free parking number prediction was verified by simulation providing a basis for regional parking guidance.