基于深度学习的火力发电厂锅炉故障诊断方法

A Fault Diagnosis Method for Boiler in Thermal Power Plants Based on Deep Learning

针对当前火电厂锅炉故障诊断准确度不足的问题,提出一种基于深度学习的火电厂锅炉故障诊断方法。在系统分析锅炉运行特征的基础上,选取重要参数作为锅炉故障特征量。设计基于深度卷积神经网络的作为基础的故障特征分类模型,用以对故障特征信息进行深度学习与分类识别,并针对DCNN参数寻优效率低和稳定性低的问题,采用引入避险原则改进的缎蓝园丁鸟优化算法实现DCNN参数的自适应寻优,从而建立基于ISBO-DCNN的火电厂锅炉故障诊断模型。实验结果表明:提出的ISBO-DCNN模型相比于其他计算模型具有更高的诊断准确性,能够较为准确地实现对火电厂锅炉的故障诊断,在火电厂运检工作方面具有良好的工程实际应用能力。

A deep learning based fault diagnosis method for thermal power plant boilers is proposed to address the issue of insufficient accuracy in current boiler fault diagnosis.On the basis of systematic analysis of boiler operation characteristics,important parameters are selected as boiler fault characteristic quantities.Design a fault feature classification model based on Deep Convolutional Neural Networks(DCNN)for deep learning and classification recognition of fault feature information.In response to the low efficiency and stability of DCNN parameter optimization,the Satin Bowerbird Optimization(SBO)algorithm improved by introducing the principle of avoidance is used to achieve adaptive optimization of DCNN parameters,Thus,a fault diagnosis model for thermal power plant boilers based on ISBO-DCNN is established.The experimental results show that the proposed ISBO-DCNN model has higher diagnostic accuracy compared to other computational models,and can accurately diagnose faults in thermal power plant boilers.It has good engineering practical application ability in the operation and inspection work of thermal power plants.