Research On Multiobjective Optimization Of Byproduct Gas System In Iron And Steel Industry

Byproduct gas in the important secondary energy in iron and steel industry. It is useful to take full advantage of byproduct gas for energy saving in iron and steel industry. Most of time, the byproduct gas system is unbalanced as the production and consumption of byproduct gas connecting to steel production. As a result, research on optimal scheduling is important.Based on an actual iron and steel enterprise, this dissertation's research combine actual demands and previous experience to study related problems. The main research contents and achievement of the thesis are listed as follows:(1) Research on the forecasting models of the supply and consumption of byproduct gasBased on the research of factors connected to the supply and consumption of byproduct gas, all equipments in the byproduct gas system are divided into three types which is forecasted with different methods. To the most complicated equipments affected by multiple factors, an ARMA forecasting model amended by BP neural network is proposed. This mixed model is based in forecast results of ARMA model. Furthermore, these results are amended by BP neural network whose inputs are reduced by grey relativity analysis. This mixed model not only can realize multi-cycle prediction as ARMA model, but also can reflect changing production status. This model improves the prediction accuracy compared with traditional ARMA model while production status changing.(2) Research on the gas consumption characteristics of captive power plantAs a result of lack of research on the gas consumption characteristics of captive power plant, a model mixed by mechanism modeling based on design data and amending based on operation data is proposed in this paper aimed at extraction steam heating units in the captive power plant. This method uses cycling function method to get the gas consumption characteristics based on the design data and then amends model parameters with typical operation data. This mixed model has less demend for operation data and can get accurate gas consumption characteristics of different units, which is basis of optimum between units.(3) Research on multi-objective genetic optimization algorithm based on improved gene designing methodIn the solution for optimization model, as a result of too many constraints in the model, it is difficult for traditional multi-objective genetic optimization algorithm. Therefore, this thesis improves gene designing method in view of the actual characteristics of the optimal scheduling problem of the byproduct gas system. This method uses heat values and heat as genes replacing traditional gas flow. The improved gene designing method turns heat value constraints and heat quantity constraints into definition domain of genes. In this way, constraints of model are reduced significantly and it is easy to solve the model.(4) Research on two scale optimization strategy of byproduct gas systemA model of two scale optimization strategy of byproduct gas system is proposed in this paper, which takes different features of two main buffer methods of byproduct gas into consideration. The long time scale (1h) optimization strategy considers capative power plant as main buffer method, which realizes gas balance in the long time scale and the highest gas utilization rate in power plant. The short time scale (10min) optimization strategy considering gasholders as buffer method. In this way, the feature of captive power plant, big buffer capacity and slow load change, and the feature of gasholders, small buffer capacity and rapid load change, can be utilized fully. Depended on analysis of the actual optimization results, the mothod in this paper can realize balance in the byproduct gas system. What's more, the load of captive power plant is more stable and gas levels of gasholders have less fluctuation compared with the traditional method.