Research On Performance Forecast And Operation Scheduling Of Pumping Station Of Blast Furnace Cooling System

Blast furnace ironmaking is an important part of steel production.As a guarantee for maintaining the safe and stable operation of the blast furnace,the blast furnace cooling system is effectively controlled and optimally dispatched,which is an important means to ensure the normal operation of the blast furnace and reduce energy waste.This paper models the centrifugal pump in the cooling system,and studies the performance index prediction and energy efficiency optimization of the centrifugal pump.Aiming at the problem of centrifugal pump performance index prediction,the structural parameters of the mechanism model are often difficult to obtain and drift with the equipment operation,and the data-driven machine learning method is difficult to adapt to the equipment operation multi-condition scenario,and a least square induction method is proposed.Performance prediction method of centrifugal pump based on Least Square Induction Transfer Learning(LSITL).This method uses the least squares method to extract the characteristics of the centrifugal pump performance curve,and adopts the induction method to establish the migration model under multiple working conditions.The inverse solution of the least squares support vector machine(Least Squares Support Vector Machines,LSSVM)is used to achieve the correction.Performance prediction of centrifugal pumps.Aiming at the energy efficiency optimization problem of parallel water pump units,considering the economy and energy conversion efficiency indicators of the scheduling scheme when working conditions change,a least square inductive transfer learning and an improved particle swarm algorithm(Least Square Induction Transfer Learning-Particle swarm algorithm)are proposed.,LSITL-PSO)combined method.This method obtains the optimized objective function through migration learning,and then optimizes the optimization of the particle swarm algorithm for the objective function to realize the energy efficiency optimization of parallel pump sets.The data of the apros simulation platform water pump and the mechanism model water pump are verified by experiments,and compared with the mechanism modeling method and traditional machine learning methods,it shows the applicability and accuracy of the LSITL algorithm in the prediction of pump performance indicators.The comparison with the optimization results obtained by the simulation platform enumeration method shows the accuracy and efficiency of the improved PSO algorithm.Compared with the mechanism model-PSO algorithm,the LSITL-PSO algorithm is feasible and accurate.