Research On Coagulation Dosing Prediction Based On Improved Tianying Optimization Algorithm Combined With LST

In the process of coagulation,sedimentation,filtration and disinfection of water purification treatment in waterworks,the core is the coagulation administration process,which can effectively reduce the retention of suspended solids and colloidal substances in the water,and can directly affect the water quality and production and operation costs of the factory.Due to the diversity of water state changes and the influence of objective environmental factors,it is difficult to establish an accurate mathematical model according to its reaction mechanism at present,and it is difficult to accurately put the dosage of coagulant in actual production,which seriously limits the economy of coagulant dosing in waterworks.The prediction of coagulant dosage in waterworks can provide more accurate,reliable and economical prediction technology for waterworks,which is helpful to improve the water quality and increase the economic benefit of waterworks.In order to achieve the above objectives,this thesis explores the relationship between coagulation dosage and each water quality characteristic,and improves the shortcomings of the aquila optimization algorithm(AO).Then,the improved aquila optimization algorithm based on multi-strategy integration(MSIAO)is used to optimize the parameters of the long short-term memory network(LSTM),and the coagulation dose prediction model of MSIAO-LSTM is constructed.The model was used to train and test the historical data of coagulation dosage of a water plant in the southwest.The specific work of this thesis is as follows:(1)Firstly,the research status of coagulant administration in water plant was analyzed;Secondly,it describes the tap water treatment process and coagulant administration process and its mechanism.Through the historical coagulant administration data and historical water quality data,the various factors affecting the dosage of coagulant are deeply analyzed.The results showed that the seasonal type,raw water turbidity,temperature,p H and alkalinity had important effects on the dosage of coagulant.At the same time,the collected historical data was preprocessed,and the correlation analysis was carried out for each feature of coagulation administration,and irrelevant variables were eliminated to reduce the input dimension of the prediction model and improve the prediction accuracy.(2)By analyzing the optimization mechanism of aquila optimization algorithm,it is found that although aquila optimization algorithm has the advantages of high optimization accuracy,easy implementation and strong applicability,it still has some problems,such as easy to fall into local optimal and slow convergence speed.Therefore,the population is initialized with Tent chaotic map based on refraction reverse learning,which makes the population distribution more uniform.The inter-species mutual assistance and optimization strategy are proposed to improve the search mechanism to improve the global search performance of the algorithm.The adaptive weight based on Bernoulli sequence is used to improve the search rate of the algorithm,and the Cauchy-Gauss mutation operator is introduced to enrich the population diversity.Enhance the ability of the algorithm to escape local extreme value.Twelve benchmark functions were used for optimization test,and the function optimization of MSIAO algorithm was compared with classical and new intelligent optimization algorithms.Wilcoxon rank sum test was used for significance analysis,and CEC2014 complex test function was used for optimization comparison,so as to verify the effectiveness and superiority of the improved aquila optimization algorithm.(3)The single-layer and double-layer LSTM neural networks were used to carry out training tests on the historical data of coagulation administration according to seasons,and it was found that the model prediction accuracy was closely related to the hidden layer,the number of neurons,the learning rate,the iteration rounds and other parameters of the neural network.Therefore,the improved Eagle optimization algorithm based on multi-strategy fusion was used to optimize the LSTM parameters to improve the model prediction accuracy.In the prediction of coagulant dosage in spring,summer,autumn and winter,the average determination coefficients R~2 of the MSIAO-LSTM model were 0.9593,0.93735,0.96915 and 0.9793,respectively,which increased by16.49%,23.01%,20.07%and 11.47%compared with the basic LSTM prediction model,respectively.Compared with the AO-LSTM model,the LSTM model improved by4.93%,10.98%,10.07%and 2.08%respectively.The results show that the LSTM model optimized by MSIAO and AO algorithm has higher prediction accuracy,and the LSTM model optimized by MSIAO algorithm has the best prediction effect.Compared with the traditional model,The model proposed in this thesis can better track the variation rule of four seasons coagulant dosage in waterworks and provide a strong guarantee for the stable operation of waterworks.