基于神经网络的船舶主动力装置系统可靠性优化分配  

Optimal Apportionment of Marine Power Plant System Reliability based on Neural Networks

作　　者：安卫[1] 吴晓平[2] 陈国钧[1] 

机构地区：[1]海军工程大学动力工程学院,湖北武汉430033 [2]海军工程大学基础部,湖北武汉430033

出　　处：《中国造船》2003年第1期59-64,共6页Shipbuilding of China

摘　　要：船舶主动力装置系统可靠性分配是船舶动力可靠性工程设计中的重要内容 ,它常常构成有约束条件的非线性优化问题。探讨了基于神经网络理论求解该类问题的方法。建立系统可靠性优化分配的数学模型 ,采用罚函数法将不等式约束转化为无约束的优化问题 ,构造出 Hopfield神经网络求解的能量函数 ,求得动力系统微分方程组。用数值计算的方法求解微分方程组 ,所得平衡点即为最优解。该方法通用性好 ,收敛速度快 。The reliability apportionment of marine power plant system is important content in the design of marine power. It often consists of nonlinear optimization problem. This paper discusses the method solving the above problem based on neural networks theory. Firstly, the mathematical model of system reliability optimal apportionment is established. Through the penalty function method, the problem of inequality restriction is transformed to nonrestraint optimizsation problem. Then, the power function to solve the problem with Hopfield neural networks is formed. Secondly, the differential equation group of dynamic system is obtained. Thirdly, the differential equation group is solved with the method of numeral calculation. The attained balance point is just the optimal solution. The method does well in common use, and the speed of convergence is fast. It provides a new solution to reliability apportionment of marine power plant system.

关 键 词：船舶 主动力装置 优化分配 系统可靠性 神经网络 

分 类 号：U664.1[交通运输工程—船舶及航道工程]