Research On Energy Saving Optimization Of Power Station Units Based On Big Data Technology

With the continuous development of China's economy,the energy consumption also keeps growing.As the pillar industry of energy supply,and under the effect of China's unique energy structure,Coal-fired Power Generation will continue to hold dominate position,and is the key to achieve optimal energy consumption and emission reduction for China.The rise and rapid development of Big data and AI promote the deep integration of intelligentialize,informatization and industrialization,and bring new ways and opportunities for Chinese power generation enterprises to transform from the extensive development mode with "High energy consumption,High emission,Low efficiency" to the green development mode with "Low energy consumption,low emission,High efficiency".With the popularization of power system information's integration,power station units have accumulated a large quantity of operating data.It has become an important research field of current power generation industry that how to excavate the potential value of these data and make use of them.To promote the multi-angle,deep-seated,wide-range excavation and application on power station units of Big data technology has significance to improve the power stations efficiency and deepening the optimal energy consumption.Firstly,through the summary and analysis to definition and characteristics of power big data,the definition and value of power station units big data are expounded.Aiming at the excavating process of big data of power station units,the hierarchical structure of big data of power station units is proposed;meanwhile,based on the key technology of big data processing,the application framework of power stations data ecosystem is constructed,and the big data storage and batch processing technology is introduced to realize the information excavating and acquisition of big data of power station units.Secondly,the characteristics and data quality of the massive operating data of power station units are deeply analyzed,quality defects and causes are clarified,and data processing strategies and methods are clarified.In the detection of power station units data,the multi-parameter threshold method is used to screen the steady-state condition data.In the data preprocessing,bilinear interpolation method and joint probability density method are used to deal with data gaps and discrete values respectively;meanwhile,aiming at the characteristics of high latitude and non-linearity of big data of power station units,the fuzzy rough set theory is introduced to establish the method of feature parameter selection of big data of power station units,and to eliminate redundant or irrelevant parameters,simplify the feature parameter set,and ensure the high efficiency and high precision of big data excavating process.Thirdly,the energy consumption analysis method of power station units based on big data excavating technology is carried out.Based on the attribute reduction of fuzzy rough set,the k-means clustering algorithm is improved by Canopy algorithm,and the parallel calculation of the improved k-means clustering algorithm on Hadoop platform is carried out,so as to realize efficient optimization under all working conditions and determine the target reference value of power station units operation.At the same time,the energy consumption sensitivity analysis model of the support vector machine is established,and based on the dependence and correlation between energy consumption and input characteristic parameters,the sensitivity itv coefficient of key characteristic parameters to power supply coal consumption under different load conditions is analyzed.Then.the status quo of two-level optimal load distribution in the power plant network is analyzed,and a multi-target optimal load distribution strategy based on big data technology considering boundary conditions and pollutant emissions is proposed for the development of clean,flexible and intelligent power demand.Based on the massive operating data of power station units.the big data analysis method is introduced.the support vector machine is improved through particle swarm algorithm.then the multi-objective load optimization and prediction model of factory-level rapidness.economy and environmental protection can be established:with the use of MapReduce parallel programming model to realize the parallel process of the NSGA-II optimization algorithm.the plant-level multi-objective optimal load distribution calculation is accomplished.Based on this,the optimal load distribution at the plant level can effectively reduce the power-supply coal consumption and pollutant emission of the power station units,and has reference significance for the energy-saving power generation scheduling of power system.Finally,a comprehensive evaluation study on bench-marking management of gas power station units is carried out.On the basis of analyzing and studying the characteristics and characterization parameters of bench-marking management evaluation for gas power plants,a comprehensive evaluation index system of bench-marking manacement for gas power plants is established from five aspects including safety and environmental protection,units reliability,equipiment management.cconomic operation and production tcchnology:meanwhile.based on the rough set attribute reduction principle of big data analysis method,combined with vector Angle cosine and principal component analysis method,a comprehensive evaluation model for bench-marking management of gas power station units is established.Aiming at the model empowerment scheme,the sensitivity analysis model based on index weights is established,and test results show that the established comprehensive evaluation model of gas power plant units in the management of the weight distribution of sensitivity is low,the model evaluation result is stable and good robustness,which provides guidance and help for bench-marking management of gas power plant enterprises and competitions between the power plant units.