基于蚁群神经网络的凝汽设备故障诊断

BP算法在神经网络中应用较为广泛,但有收敛速度慢、易于陷入局部极小点的缺点.而蚁群算法是一种新型的模拟进化算法,有正反馈、分布式计算、全局收敛、启发式学习等特点.用蚁群算法优化神经网络,能使其权值快速准确的收敛于全局最优点.经比较,其优化性能要优于BP算法和遗传算法.凝汽设备是电厂汽轮机的重要辅助设备,把经蚁群算法优化的神经网络应用于凝汽设备故障诊断,仿真实例表明该方法对凝汽设备故障诊断效果良好.

Fault Diagnosis of Condensing Equipment Based on Ant Colony Neural Network

BP algorithm is widely used in neural networks. But it has some shortcomings, such as low convergent speed and easy convergence to the local minimum points. Ant colony system is a novel simulated evolutionary algorithm. It has positive feedback, distributed computation, and uses a constructive greedy heuristic. Using ant colony to optimize neural network can make the network connection weights converge to the global advantage rapidly and accurately. By comparison, its optimizing performance is better than t...