Process Optimization For Enzymatic Saccharification Of NaOH Pretreated Cotton Stalk Based On Artificial Neural Network

Producing reducing sugars from lignocellulosic bimass through pretreatment and enzymatic hydrolysis,is a very important way for high-valued biomass utilization.Among many pretreatment methods for lignocellulosic biomass,hydrolysis in alkaline has attracted much attention by researchers at home and abroad,since it has the advantages of mild reaction conditions,low equipment requirements,and high yield of enzymatic hydrolyzation,etc.In this study,cotton stalk collected from the outskirts of Shihezi city in Xinjiang was used as the starting materials.The processes of hydrolysis of cotton stalk materials in dilute alkaline solution,cellulase solution,and cellulase/xylase mixed solution were investigated.And back propagation?BP?neural network was used to simulate and predict these hydrolysis processes.Through the modeling and optimization of these hydrolyis processes of cotton stalk,the efficiency of hydrolysis was enhanced,which provided a significant application basis for the effective utilization of cotton stalk.The effects of different dilute alkalis?NaOH,KOH,Ca?OH?₂,NH₃·H₂O and CH₃COONa?on the components of pretreated cotton stalk were studied.The results showed that cellulose content in cotton stalk hydrolyzed in NaOH solution was more than hydrolyzed in other alkalis,which was up to 60.35%.The process of hydrolysis of cotton stalk in NaOH was investigated,and BP neural network was used to simulate and predict the hydrolysis process.The results showed that the weight loss rate and lignin removal rate of cotton stalk increased,and the hemicellulose content decreased,with the strengthening of NaOH pretreatment conditions?concentration,reaction temperature,and time?.Taking NaOH concentration,reaction temperature,time and solid content as the input layer,reducing sugar content and lignin content as the output layer.Through repeated training,a back propagation neural network?BPNN?model?4-15-2?for the hydrolysis of cotton stalks in NaOH was established.The linear regression values of reducing sugar content and lignin content in NaOH prereated cotton stalk were 0.9227 and 0.9002,respectively.The MSE of BPNN was small,indicated that the neural network model had a good performance and could be used to optimize the process of hydrolysis of cotton stalks in NaOH solution.Based on the optimized weights in the neural network model,the effects of different pretreatment conditions on reducing sugar content and lignin content were analyzed.The influence of NaOH pretreatment conditions on the content of reducing sugar was:alkali dosage>reaction time>temperature>solid content,and the influence on the lignin content in cotton stalk was:temperature>solid content>alkali dosage>reaction time.The reducing sugars produced from hydrolysis of cotton stalk materials in cellulase solution after pretreated in different dilute alkalis?NaOH,KOH,Ca?OH?₂,NH·H₂O and CH₃COONa?were compared,and the maximum reducing sugar content was obtained for cotton stalk pretreated by NaOH.Then hydrolysis of NaOH-treated cotton stalk materials in cellulase solution was studied.And BP neural network was used to simulate and optimize the enzymatic hydrolysis process conditions?temperature,enzyme dosage,solid content,and enzymolysis time?.The results showed that the reducing sugar yield of cotton stalk could reach 28.58%,while the concentration of NaOH,reaction temperature,treatment time and solid content were 6%NaOH,60°C,60 h and 10%,respectively.After many times training,a neural network model?4-7-1?for the hydrolysis of NaOH treated cotton stalks in cellulase was established.The MSE of the model was small,R was 0.9571,indicated that the model had a good performance.The optimal enzymolysis conditions were obtained,with the temperature,the enzyme dosage,the amount of substrate added and the time were 46.31°C,70 FPU·g^-1,7.5%and 72 h,respectively.At these conditions,the reducing sugar yield of cotton stalk was62.14%.The yield of reducing sugar obtained from the verification experiment was very close to the predicted value of the model equation,indicated that the prediction model was reasonable and effective in the range of the NaOH-treated cotton stalk digestion experiment.The factors affecting the enzymolysis of cellulase on the yield of reducing sugars were compared as follows:enzymolysis temperature>hydrolysis time>enzyme dosage>solid content.The process of hydrolysis of NaOH-treated cotton stalk materials in the cellulase/xylanase mixed enzyme was investigated.Selecting total enzyme dosage,cellulase addition ratio,reaction time and solid content as input factors,BP neural network was used to simulate and optimize the enzymatic hydrolysis of cotton stalk in the mixed enzyme solution.The MSE of the model was small,R was 0.9571,indicated that the model?4-6-1?was reasonable and effective,and could be used to predict the reducing sugar yield.The optimized reaction conditions were written as follows:the enzyme dosage was 35 g/L,the proportion of cellulase was 43%,the solid content was 5%,the hydrolysis time was 62.6 h,respectively.At these conditions,the reducing sugar yield was 85.54%,which was greater than 62.14%?hydrolysis in cellulase solution?,indicated that the addition of xylanase could promote the hydrolysis of NaOH treated cotton stalk in cellulase.The influence of the four factors on reducing sugar yield were compared as follows:hydrolysis time>enzyme dosage>ratio of cellulase addition>solid content.The surface morphology and structure of the material were characterized by XRD,SEM and FT-IR.The results showed that lignin and part of hemicellulose were removed in cotton stalk after pretreated by NaOH,which resulted in the relative increase of cellulose content.And the surface structure of cotton stalk was destroyed and the cellulose was exposed,which facilitated the enzymatic reaction.After enzymatic hydrolysis,the carbohydrates in pretreated cotton stalk were hydrolyzed into sugar,so,the content of cellulose and hemicellulose content decreased,and lignin content increased relatively.