摘要
船舶各种设备故障的早期诊断和预测,对船舶的安全运行具有非常重要的意义。由于船舶上设备繁多,运行环境特殊,因此,各种设备的故障症状与故障原因之间关系十分复杂,使用传统诊断方法在实际应用中效果不理想。BP神经网络在故障诊断中有广泛的应用,但由于BP网络采用的是沿梯度下降的搜索求解算法,存在收敛速度慢,且容易陷入局部极小等问题。而遗传算法具有全局搜索速度快的优点。为此,采用自适应遗传算法来优化BP神经网络,并以船舶主机轴系的故障诊断为实例,证明遗传算法优化的BP网络方法非常适用于对船舶各种设备故障的早期诊断和预测。
The early stage diagnosis and forecast of marine equipment failures is important for ship＇s safe operation. Because of various equipment and special running conditions, the relationship between symptoms and causes of fault is very complicated and traditional fault diagnosis methods are not ideal in practice. The BP nerve network has been widely applied in fault diagnoses, but it has some trouble with slow convergence rate and easy getting into local infinitesimal due to adopting search algorithm along grads drop. Genetic algorithms have the advantage of rapid searching rate, auto-adapt genetic algorithms are then adopted to optimize the BP nerve network. With an example of marine main engine shafting fault diagnosis, it is proved that BP network optimized by genetic algorithms can satisfy early stage diagnosis and forecast of marine various equipment failures very well.
出处
《中国航海》
CSCD
北大核心
2007年第2期85-88,共4页
Navigation of China
关键词
船舶
舰船工程
故障诊断
神经网络
遗传算法
主机
轴系
Ship, Naval engineering
Fault diagnosis
BP network
genetic algorithm
Main engine
Shafting system
