【摘要】 对火电厂回热系统的故障进行分析 ,提出了基于Levenberg Marquardt神经网络算法的回热系统故障诊断模型 ,其算法是梯度法与高斯牛顿法的结合。仿真结果表明 ,该模型显著缩短了训练时间 ,降低了跟踪误差 ,具有很高的准确性 ,优于常规BP神经网络算法模型 ,适合用于在线学习与监测更多还原

【Abstract】 Faults in regenerative heat system of the power plant have been analysed, and a new fault diagnosis model based on levenbergMarquardt (LM) neural network for regenerative heat system being put forward. The LM algorithm is a combination of the gradient decent algorithm with the GaussNewton algorithm. Results of simulation show that the new model has greatly reduced the training time and the tracking error, having very high accuracy, being more superior than traditional BP neural network algorithm model, and suitable for online training and monitoring.更多还原