Prediction And Optimization Of Coiling Temperature For Laminar Cooling Of Hot-Rolled Strip

Laminar flow cooling technology is currently one of the most used cooling methods in domestic hot strip mill accelerated cooling units,and the key to its cooling is the accurate control of the strip coiling temperature.However,due to the harshness of the production site environment,there are two main difficulties in ensuring the quality of the strip,one is that due to the presence of a large amount of water mist in the cooling process of the strip,it is difficult to achieve accurate detection of the strip temperature in real time.Accurate prediction of mid-laminar cooling and coiling temperatures is the key to achieving coiling temperature control.Secondly,in the production of hot rolled strip,the cooling collector tube is in a high temperature production environment for a long time,and there are problems such as aging,deformation,rusting and blockage of the collector tube nozzles,which can lead to the predetermined cooling pattern in laminar cooling becoming less accurate and reduce the control accuracy of the strip in laminar cooling.significance.Based on the production data collected in the field over a period of 2 months,the paper aims to improve the control accuracy of the coiling temperature and the quality of the produced strip steel based on the optimization algorithm.(1)ELMAN curl temperature prediction model based on an improved Atom Search AlgorithmThe Atom Search Algorithm(ASO)is based on the laws of motion of atoms and guides the population through the interaction forces between individual atoms in the population to perform intelligent and optimal search.It is a simple algorithm with few parameters.To address its shortcomings of being prone to premature maturation and slow convergence in the later stages of computation,the algorithm is improved by adding adaptive weights and levy flight strategies,and is applied to the search for the optimal weights and thresholds of the ELMAN curl temperature prediction model,which improves the shortcomings of the ELMAN network in terms of slow training speed and the tendency to fall into local optima in the training process and the difficulty of reaching the global optimum in the search process.The results show that the ELMAN curl temperature prediction model based on the atomic search algorithm can predict the curl temperature of the strip more accurately based on the existing strip data,The hit rate of the improved model at±5℃ is 92.3%,which is higher than the selected comparison model,and the effect is good.(2)The DELM layer cold time series prediction model based on the improved chimpanzee optimization algorithmThe chimpanzee algorithm is a new type of meta-heuristic optimization algorithm proposed in recent years.For the disadvantages of slow convergence speed and poor global search ability,the chimpanzee population is initialized using Latin hypercube sampling,and the Corsi variation strategy and somersault foraging strategy are introduced in the position update.better solution accuracy.As the strip moves very rapidly in the layer cooling region,the network model for its prediction must have the ability of fast computation if the overall strip temperature is to be accurately controlled.In deep learning models,DELM has the advantage of fast training,but its effect is affected by the random input weights and random bias of each ELM-AE that makes up the model,so the improved chimpanzee optimisation algorithm is used with The model was combined with the improved chimpanzee optimisation algorithm to optimise the parameters,preserving the training speed of the network model while improving its optimisation accuracy,and achieved good prediction results in the prediction of temperature time series in the mid-layer cold section.(3)Optimisation of laminar cooling header opening and closing strategyIn order to balance the relationship between coiling temperature and cooling rate and to improve the quality of strip steel,the laminar cooling header opening and closing mode is optimised in terms of header opening and closing position and header opening and closing quantity.Firstly,a multi-objective particle swarm algorithm incorporating multiple strategies is used to optimise the position in the laminar cooling header opening and closing mode,and the multi-objective problem is transformed into a multi-objective multi-choice knapsack problem.Compared with the multi-objective particle swarm algorithm,the multi-objective particle swarm algorithm incorporating the velocity update strategy of the atomic search algorithm and the position update strategy of the chimpanzee algorithm searches for a more comprehensive Pareto surface,and searches for the collector tubes with less loss in the cooling mode,laying the foundation for the subsequent optimisation of the collector tube quantity.Secondly,on the premise of collector location optimisation,a fuzzy expert compensation decision maker is established,using fuzzy theory,the coiling temperature and strip steel plate cooling rate predicted by the laminar cooling coiling temperature prediction model as input,combined with field expert experience,a dual-input single-output expert decision maker is designed to intelligently adjust the number of collector valves opening and closing in the laminar cooling section,and the coiling temperature is controlled within the target range;the laminar cooling The temperature predicted by the time series prediction model of the mid-section temperature and the cooling rate of the strip steel plate are used as inputs to control the overall temperature uniformity of the steel plate by means of a fuzzy expert compensation decision maker.The effectiveness of the optimisation method is also verified through MATLAB simulation experiments to provide effective guidance to the site.