Research On Air-Condition Fault Detection Algorithm Based On Deep Learning

Air conditioning system is an indispensable part of modern buildings,which can provide people with a comfortable and safe working and living environment.However,various faults inevitably occur in the air conditioning system,which will lead to higher energy consumption and poor indoor thermal comfort.Therefore,it is of great significance to detect and diagnose faults in the air conditioning system.Therefore,based on the data-driven fault detection and diagnosis process,this thesis combines the deep learning neural network model to detect and diagnose the fault of the air conditioning system.The experimental results show that the neural network model proposed in this thesis can achieve high fault detection and diagnosis accuracy.The main research contents are as follows:(1)The disadvantage of this traditional manual design extraction method is that the feature extraction process and the subsequent fault detection and diagnosis process are independent of each other,and there is no joint optimization between the two.In response to this problem,a framework SDCNN,which combines shallow convolutional neural network and deep convolutional neural network,is proposed.Convolutional neural networks can integrate feature transformation,feature extraction,and pattern recognition steps into a single model,which is jointly optimized.At the same time,in order to improve the feature extraction ability of the convolutional network,the shallow convolutional network and the deep convolutional network are combined to form a dual-channel convolutional neural network,which further improves the fault detection and diagnosis accuracy.(2)The feature extraction ability of convolutional neural networks is excellent,but it is not optimal for pattern recognition.First,the trainable parameters in the fully connected layer of the convolutional neural network model account for 80-90% of the model.Secondly,as the classifier of the convolutional neural network framework,the Soft Max function has the problem of calculation overflow.Aiming at the above two problems,a hybrid model DCCNN-LGBM,which combines two advanced classifiers,two-channel convolutional neural network and optical gradient booster,is proposed to perform accurate fault detection and diagnosis for air handling units.Among them,DCCNN is a two-channel convolutional network combining shallow convolutional network and deep convolutional network for feature extraction.Among them,LGBM is the classifier,LGBM replaces the fully connected layer and Soft Max,and uses the features extracted by DCCNN to detect and diagnose the fault of the air handling unit.The experimental results show that the method can effectively detect and diagnose the faults of air handling units,and the detection accuracy is better than most existing models.(3)Deep learning-based models require a large number of balanced datasets to achieve good results.But in the real world,faults are usually fixed in a short period of time,and it is difficult to collect enough faulty samples to construct a balanced dataset.In order to solve the problem of less fault data,a method to generate fault data using Conditional DRAGAN(CDRAGAN)based on convolution/deconvolution is proposed.And use the generated fault data as a dataset to train a classifier for fault detection and diagnosis.The good experimental results prove that the method proposed in this paper can still achieve high fault detection and diagnosis accuracy under limited wrong training samples.