Study On Simultaneous Optimization Of Low Temperature Flow Properties And Oxidation Stability Of Biodiese

The application of renewable energy is an effective way to achieve the goals of carbon neutrality and timely carbon peak.The content of unsaturated fatty acid alkyl ester in biodiesel is more than half,low temperature flow properties of biodiesel improve while oxidation stabilization performance becomes worse with the increase of unsaturated degree,these two properties can not be unified and coordinated within the standard range,which seriously restricts the commercial application and rapid development of biodiesel.At present,the most mature optimization technologies for low temperature flow properties and oxidation stabilization performance are to add anticoagulants or antioxidants,but these additives can solely optimize one performance,and have a disadvantage of fast consumption,toxicity,bad versatility,high cost of synthesis,altering other biodiesel properties or yield reduction.In order to achieve maximum utilization rate of biodiesel at low cost,a new approach based on iodine value（Ⅳ）is proposed for synchronously optimizing low temperature flow properties and oxidation stabilization performance of biodiesel in this dissertation research,biodiesel with the best performance can be obtained by blending components to the optimal Ⅳ.In doing so,biodiesel and its blends are used as the research object,combined with machine learning algorithms and meta-heuristic optimization algorithms to explore the complex nonlinear correlations between biodiesel components and low temperature flow properties,oxidation stability and Ⅳ.The main contents are as follows:（1）The effects of composition on low temperature property were investigated by multi-layer perceptron neural network（MLPNN）and meta-heuristic optimization algorithms,and a novel hybrid model is established for estimating biodiesel cold filter plugging point（CFPP）.The results showed that the accuracy of the improved gray wolf optimization（IGWO）-MLPNN model was high with a root mean square error（RMSE）and the correlation coefficient（R²）of 1.2093 and 0.9598,respectively.Compared to other CFPP prediction models,it is illustrated that the IGWO-MLPNN model proposed in this thesis is novel and precise,and it covers a wide range of modeling data.Additionally,it is suitable to predict the CFPP of unknown biodiesel and the complex relationship between components and CFPP is clarified.（2）Kinematic viscosities（Kv）of 55 fatty acid methyl esters（FAME）blends at 40℃and biodiesels at 10-80°C were measured,combined with multivariate linear regression（MLR）method and Andrade equation to study the effects of composition and temperature on the flow property.The results showed that the correlation of Kv based on composition had a R² of 0.996,and compared with the five models previously published in literature,the average absolute relative error（AARD）was the lowest of 1.69%.Meanwhile,the correlation of Kv based on composition and temperature had a R² of0.996,and application of Kvs of seven biodiesels from literature proved a low overall AARD of 0.94%,this correlation can better explain the relationship between composition and temperature and Kv.（3）The induction periods（IP）of 13 biodiesels and 80 biodiesel binary mixtures determined by Rancimat method were used to analyze the effects of composition and its molecular structure on oxidation stabilization performance,and two novel hybrid models are proposed for estimating biodiesel IP.The results showed that the RMSE and Pearson correlation coefficient（PCC）of a random forest-hybrid genetic algorithm-particle swarm optimization-least squares support vector machine（RF-HGAPSO-LSSVM）hybrid model based on 9 important FAME were 0.2628 and 0.9920,and the order of effects of components on the IP is C18:3>C18:2>C18:1>C16:0.Reactivity of unsaturated fatty acid chains and double bonds in molecules with oxygen is high;the position of the double bond significantly affects the oxidation rate of the fat component.The PSO-LSSVM model based on five parameters（including bis-allylic position equivalent（BAPE）and topological index,etc.）had a RMSE and PCC of 0.319 and0.9886,respectively.The oxidation process was analyzed at the microscopic level based on this model,and the results reveal that the BAPE is the main cause of biodiesel deterioration.The distance from double bond to the ester group,the hydroxyl characteristics,and the cis-,trans-isomers have significant effects on the oxidation stability.（4）The Ⅳs of 14 biodiesels and 38 binary mixtures of biodiesel were determined by Wij’s method,combined with feature selection method,whale optimization algorithm（WOA）and LSSVM to study the effects of composition on Ⅳ,and a novel approach is proposed for estimating the Ⅳ as a function of FAME profiles.The results showed that the RMSE and PCC of RF-WOA-LSSVM model were 1.1893 and 0.9977,respectively,and C18:2 and C18:1 have a significant effect on the Ⅳ.This hybrid model had an outlier of 3 in the Willam plot,an average absolute error of 0.2791%,and a Gaussian distribution histogram of the residual distribution.,the results verify the applicability of the modeling data and the accuracy of the model.Comparison with previous biodiesel Ⅳ machine learning models and application of other new experimental datasets from the literature prove the feasibility of the proposed hybrid RF-WOA-LSSVM model for estimating the Ⅳ of different biodiesels types.（5）Aiming at the contradictory characteristics between Ⅳ and low temperature flow properties and oxidation stabilization performance,A theoretical method for synchronously optimizing these two properties through Ⅳ is constructed.On the basis of the above studies,the CFPP,IP,and Ⅳ of biodiesel mixture were simulated.Through standard limits and image analysis,the optimal Ⅳ range of 94.26～113.06 g I₂/100g was obtained,and the best composition form of biodiesel in this range was given.Furthermore,a multi-factor analysis of the optimal Ⅳ range was carried out,and it is found that the standard limit of the CFPP is the main reason for influencing the optimal Ⅳ range.