Research On Intelligent Soft Sensing Of Thermal Efficiency And Energy Consumption Distribution Diagnosis Method Of Coal-fired Boiler In Power Plant

With the adjustment of the country’s energy development policy,improving the level of energy conservation and emission reduction of large coal-fired power plants has become one of the important technical approaches for China to achieve sustainable energy development and a "low-carbon economy." At the same time,due to the continuous upgrading of the power structure,coal-fired power plants have begun to assume more power grid peaking tasks,and coal-fired power plants are facing greater pressure to reduce energy consumption and improve efficiency.From this point of view,further curbing the unreasonable consumption of coal and achieving refined production of electricity have important practical significance for the healthy development of the energy industry.As the energy conversion place in the process of coal-fired power generation,the boiler’s thermal efficiency is directly related to the energy-saving level of coal-fired power units.This paper chooses to study the boiler thermal efficiency of the current active coal-fired power generation unit,combined with the actual operation of the power plant.An intelligent soft-sensing of thermal efficiency and thermal energy distribution diagnosis method of boiler was designed.Firstly,the power production technology of coal-fired power generation is briefly described,the working method and efficiency calculation method of coal-fired boilers are explained,and the information system of power plant and data sources are introduced.It also analyzes the characteristics of thermal power production data.For the abnormal data that appears as outliers,a local outlier factor algorithm is selected to identify and remove them.For the phenomenon of aliasing noise of measurement point data,three times five The filtering method performs smooth noise reduction processing on the production data to complete the preprocessing of the production data.Secondly,considering that the oxygen content of the flue gas is an important indicator reflecting the combustion status of the boiler and the high failure rate of the oxygen sensor,a training soft-sensor model of "oxygen content of flue gas-thermal efficiency of the boiler" is proposed.That is,the soft measurement of the oxygen content of the flue gas is performed first,the abnormal data is repaired,and then the soft measurement model of the boiler thermal efficiency is trained in combination with other characteristic parameters.Considering that the feature parameters selected in the engineering experience preliminary include weakly correlated features that have little effect on the predictor and redundant features that contain other feature information,they are eliminated by the garson neural network sensitivity algorithm and mutual information method.Then,training a soft-sensing model based on generalized regression neural network,and an improved strategy of the Drosophila algorithm was proposed to optimize the structural parameters of the generalized regression neural network.Finally,the performance of the soft-sensing model was verified through comparative experiments.Thirdly,in order to diagnose the energy consumption distribution of the boiler,a sliding time window is used to identify the steady-state operating conditions,and the energy consumption distribution of the boiler under the steady-state operating conditions is mainly analyzed.The fuzzy C-means clustering and curve fitting methods are used to obtain the reference curve of boiler thermal efficiency and characteristic parameters under full working conditions.By comparing the deviation of real-time production data and the reference value,the distribution diagnosis of boiler energy consumption is completed with the help of a soft measurement model.Finally,in order to solve the problem of insufficient monitoring and low intelligence of power plants by Energy Group,Combined with cloud platform technology,a remote intelligent management system on the energy group side was designed and developed,and the overall architecture and specific functional modules were analyzed and introduced.