| With the development of new energy industry in China,the power generation of thermal power industry is under great pressure.However,as a traditional energy source,thermal power generation still occupies an important position.How to save energy,reduce consumption and improve the operating efficiency of thermal power units is a problem that traditional power generation groups need to solve urgently.Nowadays,the thermal power plant has established its own production management and control center,but there is still a gap between the management of the power plants and the intelligence,problems such as cumbersome procedures become increasingly prominent with the expansion of thermal power plant scale.In addition,thermal power plant generates a large amount of highdimensional,multi-type data every day,how to excavate its deep and effective information is also a key problem faced by enterprises.With the development of cloud platform technology and the rapid application in the Internet industry,the construction of a platform based on distributed storage and cloud computing provides us with a new way of solving the power generation group’s supervision of its power plants.Aiming at the shortcomings of the existing information platform of an energy group,an intelligent thermal power analysis system of power generation group based on Hadoop cloud platform is designed,which can realize the centralized management and analysis of all kinds of data of power plants under the group.The specific research contents of this thesis are as follows:Firstly,the thesis investigates and analyzes the existing production information system of the group and one of its power plants to understand the internal information system architecture of the power plant.Based on this,the requirements of the power generation group intelligent analysis system are analyzed in detail,and the overall construction scheme of the intelligent analysis system is designed.The construction objectives,overall structure,cloud platform technology architecture and server deployment scheme are described respectively.Secondly,compared with the performance of common real-time database platforms and integrated the actual project needs,the openPlant real-time database software is selected as the data management tool for thermal power plants,the data acquisition module structure and database table structure based on the openPlant database are described in detail.On the basis of data collection and storage,the main modules of the system are designed.Thirdly,a transient thermal stress soft measurement method of turbine rotor based on the distributed Gradient Boosting Decision Tree algorithm is proposed.The MIV average impact value method is used to extract the features of the original data of the database,and the variables with strong correlation with the thermal stress value are selected.After uploading the extracted data to the HDFS system,the distributed GBDT algorithm under the Spark framework is used for model training to implement the soft measurement function of transient thermal stress.Through comparative analysis,it can be seen that the model has higher accuracy and can meet the actual needs of production.Finally,the thesis shows the main user function interface and background management interface of the system.Using the historical data of the generator set,the parallel acceleration ratio of the soft-sensing model is tested on the platform,and its superior performance is further verified. |
