摘要
减速器出口速度的合理性直接影响着驼峰溜放作业的效率和安全。目前 ,国内外常用的驼峰车辆溜放速度控制模型主要是基于车辆走行阻力的统计特性。但车辆走行阻力是随机、离散的复杂变量 ,难以准确测定 ;而且 ,基于这个统计模型的出口速度计算法比较机械 ,没有自适应能力 ,使得某些溜放环境变化后 ,溜放作业的安全连挂率有所下降 ,安全状况恶化。为此 ,本文基于模糊神经网络 (FNN)理论 ,提出了一种新的计算车辆减速器出口速度的智能控制模型。该模型采用五层的前向神经网络来构造模糊系统 ,以模拟熟练的调速作业员给定出口速度的模糊和自适应策略 ,并在相关的先验知识的基础上 ,使用了改进的误差反向传播学习算法 ,具有自学习和自适应能力。在驼峰溜放环境变化时 ,控制系统能通过自学习 ,自动校正减速器计算出口速度模型 ,改善控制品质 ,使系统保持设计的安全连挂率。
The humping efficiency and operation safety are directly affected by the rationality of the exit speed of rolling cuts. The model of hump rolling speed control on the rolling resistance of cuts is widely used in the world at present. But the resistance is a discrete and stochastic parameter, which is hard to measure accurately, and the arithmetic of calculating the exit speed based on the statistic model is not self-adapting. As a result, the safe coupling rate of cuts within classification tracks drops when factors of humping are changed, which influences the operation safety. So an intelligent control model used to calculate the exit speed of rolling cuts based on fuzzy neural networks is introduced in this paper. This model uses five-layer forward neural networks to construct a fuzzy logic system to simulate the fuzzy and self-adapting strategy. The exit speed is fed by the skilled operators. On the basis of prior knowledge accumulated from experience and sample data, it uses an improved BP algorithm to train the model, so the model is self-learning and self-adapting. The control system can adapt the change of humping factors and modify the parameters of the model automatically so as to keep the safe coupling rate of cuts as required by the system design. The computer simulated results prove that the model is very effective.
出处
《中国铁道科学》
EI
CAS
CSCD
北大核心
2001年第3期27-30,共4页
China Railway Science
基金
铁道部科技研究开发计划项目 (2 0 0 0X0 2 0 )
