Optimal Scheduling Of Blast Furnace Gas System Based On Multi-objective Differential Evolution Algorithm

The by-product gas is an important secondary energy in the iron and steel industry,and its efficient and rational utilization is the key to improve the economic efficiency and the level of energy saving and emission reduction.Blast Furnace Gas(BFG)is an important part of the by-product gas,with the characteristics of low calorific value,large fluctuation and easy release.Therefore,it is necessary to achieve the optimal scheduling of BFG system.However,the production process of BFG system is too complex to model accurately.With the improvement of the information for the iron and steel enterprises,a large amount of industrial data has been accumulated in the energy management system.Then the model can be established based on the data,and the scheduling can be optimized.Finally,the detailed and reasonable scheduling plan can be get.In view of the complexity of BFG system,the difficulty of modeling and forecasting real-time traffic,an optimized scheduling method based on improved Multi-Objective Differential Evolution(MODE)algorithm is proposed.Based on the expert’s experience and the system structure,the Dynamic Bayesian Networks(DBN)model is established based on the relationship between the gas cabinet and the gas production and elimination users.Then in order to improve the accuracy of the model,the model’s parameters are trained through the historical data.After obtaining the total adjustment amount by the DBN model,it’s necessary to choice the adjusting user and the adjustment allocation plan accordingly.We choose the users according to the user’s adjustment ability in the paper,which can be obtained by the adjustment speed and adjustable range.And the weight factor is determined by the Gauss function which varies with the gas tank overrun.And in order to solve the problem of uncertainty of the scheme caused by the wide range of adjustment,a Crowding Distance-based MODE(CDMODE)algorithm is proposed,aimed at avoiding the local optimum and improving the search precision.The algorithm combines two different mutation strategies in the process of mutation,and the population is divided into crowding population,general population and sparse population,according to the crowding distance of particles.In the early stage,the random particle is dominant,and the later,the best particles is dominant.And the weights and cross factors are determined according to the population of the dominant particles.In this paper,the actual industrial data of a steel plant is used to verify the effectiveness of the DBN model and the optimal scheduling algorithm.In the model validation phase,the prediction and inference experiments are carried out through the DBN model,and the simulation results are given.The experimental result shows that he DBN model with high accuracy can meet the industrial requirement.In the optimization algorithm verification phase,we carry out the experiment firstly by standard test functions,comparing with the existing MODE algorithm.Secondly,we optimize the scheduling through CDMODE algorithm,and compared with other optimization algorithms.The experimental result shows that the algorithm has better convergence and distribution,and the cabinet can operate safely after optimization.Therefore,the method proposed in this paper can provide scientific guidance for the balanced scheduling of BFG systems.