Research On Anomaly Detection And Anomaly Pattern Recognition Technology Of Energy Efficiency State Of Heavy Gas Turbine Unit

With the rapid development of big data technologyand artificial intelligence technology,data gradually become new means of production.Big data technology bring great changes to the society,and they bring great opportunities as fas as industry.China has put industrial big data and artificial intelligence technology on the development strategy,and proposed ’Made in China 2025’.For the power generation industry,big data and artificial intelligence technology can push the power generation industry to a new production form,and build a more efficient,safer and more intelligent power plant.In this paper,the technology of anomaly detection and anomaly pattern recgnition which are about heavy gas turbine unit is studied.Based onn the study of the thermal system and equipment composition of gas turbine unit,a reasonable anomaly pattern system is established.And carry out reserch on each anomaly pattern,analyze the causes,symptoms,consequences and treatment measures of each anomaly pattern,and lay a solid mechanism foundation for the follow-uo process.Then the LSTM-AE model is put forward,which can combine the advantages of long short memory neural network which is good at long term modeling and autoender which can represent the distribution of data,and it can effectively solve the problem of anomaly detection of energy efficiency state of gas turbine unit.At last,the anomaly pattern recognition of gas turbine unit is studied.On the one hand,on the basis of sufficient expert knowledge,this paper use the concept of graph theory to express knowledge rules,and then build a knowledge reasoning network,which can well infer anomaly patterns in the cause to pattern,pattern to symptom relationship graph.On the other hand,the artifical intelligence technology for anomaly pattern recognition is studied.By establising the artifical classification model,the task can be completed well.The study was completed in random forest with comprehensive consideration.And taking compressor fouling as an example,it shows the validity of the model.In this paper,a gas-steam combined cycle unit of Yuedian group is taken as the research object.The anomaly dectection and anomaly pattern recognition of energy efficiency state are studied.By using the actual historical operation data of the unit,carry out the application analysis of artificial intelligence technology in this field.I can be found that artificial intelligence technology can effectively solve the problem of anomaly detection and anomaly pattern recognition of energy efficiency status of gas turbine unit.