【作者】 董翠英；

【导师】 陈娟；

【作者基本信息】 北京化工大学 ， 控制科学与工程， 2019， 博士

【摘要】 超声波作为一种能量形式在液体介质中传播会引起空化效应、机械效应、热效应,这些效应已经在物理、化学、生物和医学等学科的基础研究和应用技术中展示出十分广阔的前景。超声波在植物秸秆发酵、植物药有效物质提取等植物学领域的应用是近年来超声波应用研究的新热点。单频超声波在介质中传播易产生驻波从而使能量不均匀,影响了作用效果。为解决这一问题本文研究了双频超声波在介质中传播的作用机理,研究了双频超声及其参数对空化的影响,解释了双频超声能量效应更高的原因。在双频超声作用机理的研究基础上,提出了双频超声与碱联合对玉米秸秆发酵预处理的新工艺,并对工艺参数进行优化。并针对小样本问题,研究了基于试验设计的虚拟样本生成技术,扩大了虚拟样本生成技术的适用范围。研究了双频超声植物药有效成分提取,采用试验样本与虚拟样本构成的合成样本建立提取率的预测模型,并利用此模型优化了提取工艺参数。本文的主要研究内容如下:(1)双频超声机理分析。本文分析了超声促渗作用机制,并对主动机制超声空化进行研究。研究了双频超声作用下的声场并对声场进行建模,研究了双频声场下空化泡的动力学模型。对单双频超声作用下空化泡半径的变化情况进行数值仿真,结果表明双频超声的空化效果比单频超声更强烈,在声强相同的情况下,双频超声作用下的空化效果是超声单频的4倍,双频超声的能量效率更高。分析了双频超声参数如声压幅值、频率、频率差及空化泡初始半径对空化的影响,根据理论分析指导了双频超声促渗装置的设计。(2)玉米秸秆预处理工艺。针对单频超声波在介质中传播会产生的驻波,导致超声能量不均匀等问题,本文提出采用双频超声波与碱联合对玉米秸秆进行预处理的新工艺,通过与单频超声联合碱预处理、仅碱预处理和未进行预处理这三种预处理方法进行实验对比,结果表明经双频超声波与碱联合预处理工艺处理的玉米秸秆,其微观结构、累积沼气产量、产气速度和净能量等指标均优于上述三种方法。(3)厌氧发酵预处理工艺参数优化。提出了将正交实验设计与最小二乘支持向量机(LS-SVM)相结合,建立优化厌氧发酵预处理工艺参数的预测模型。选择玉米秸秆重量、超声波持续时间、碱预处理时间和单/双频超声作为正交试验设计的试验因素。LS-SVM和正交试验结合建立累积产气量与预处理工艺参数之间的预测模型,利用模型分析了各因素对产气量的影响及各因素的约束条件,并采用遗传算法进行优化得到厌氧发酵产沼气预处理的最佳工艺参数,在此参数下实验验证了方法的有效性。(4)虚拟样本生成技术。小样本训练集加入虚拟样本可以提高机器学习能力从而解决小样本建模精度差的问题,但虚拟样本一般是根据先验知识产生,这使虚拟样本的适应范围受到一定程度的限制。针对已有虚拟样本生成方法的适应性不高的问题,提出了基于试验设计的全因子扩散生成虚拟样本的方法。该方法产生虚拟样本的适应领域可以扩大到试验设计应用领域,有效地扩大了虚拟样本生成方法的应用范围,并利用文献数据证明本文提出的虚拟样本可以提高建模精度。(5)双频超声植物药提取工艺参数优化。研究双频超声作用下从植物药金银花中提取绿原酸和木犀草苷有效成分。对影响提取率的单因素进行研究,并确定了响应面法(BBD-RSM)的实验因素及水平。按照BBD-RSM设计试验并获得试验样本,根据设计结果进行全因子扩散产生虚拟样本,虚拟样本与试验样本共同构成了合成样本;并利用LS-SVM对合成样本建立提取率的预测模型;通过对该预测模型分析,确定了各试验因素对提取率的影响与相互关系,得到优化约束条件;最后,采用改进自适应粒子群算法对模型在约束条件下寻优,得到最高提取率下的最佳工艺参数,并通过实验进行了验证。更多还原

【Abstract】 Ultrasound,as a form of energy,can cause cavitation,mechanical and thermal effects in liquid media.These effects have shown very broad prospects in basic research and application technology of physics,chemistry,biology and medicine.The application of ultrasound in botanical fields such as plant straw fermentation and extraction of active substances from plant medicines is a new hotspot in the application of ultrasound in recent years.Standing wave,which is easily produced by single-frequency ultrasound propagation in the medium,makes the energy uneven and reduces the effect of ultrasound irradiation.In order to solve this problem,the mechanism of dual-frequency ultrasound propagation in medium was studied,and the influence of dual-frequency ultrasound and its parameters on cavitation was investigated,and the reason why dual-frequency ultrasound has higher energy effect was explained.Based on the study of the mechanism of dual-frequency ultrasound,a new technology of corn straw fermentation by dual-frequency ultrasound combined with alkali was put forward,and the technological parameters were optimized.Aiming at the problem of small sample size,the virtual sample generation technology based on experimental design was studied,which enlarged the application scope of the virtual sample generation technology.The main contents of this paper are as follows:(1)Dynamic study of dual-frequency ultrasonic cavitation.In this paper,the mechanism of ultrasound-enhanced infiltration was analyzed,and ultrasonic cavitation as the active mechanism of ultrasound was studied.The acoustic field under dual-frequency ultrasound was studied and the acoustic field was modeled.The dynamic model of cavitation bubble under this acoustic field was investigated.Cavitation bubble radius was simulated numerically under the action of single-frequency and dual-frequency ultrasound.The simulation results showed that the cavitation effect of dual-frequency ultrasound was stronger than that of single-frequency ultrasound.Under the same sound intensity,the cavitation effect of dual-frequency ultrasound is four times that of single-frequency ultrasound,so the energy efficiency of dual-frequency ultrasound is higher.The effects of dual-frequency ultrasound parameters such as acoustic pressure amplitude,frequency,frequency difference and initial radius of cavitation bubble on cavitation was analyzed.The design of dual-frequency ultrasound device was guided by theoretical analysis.(2)Pretreatment of corn straw by dual-frequency ultrasound combined with alkali.In order to solve the problems of standing wave generated by single-frequency ultrasound propagation in the medium and the uneven energy of ultrasound,a pretreatment process of corn straw using dual-frequency ultrasound combined with alkali was proposed.Compared with single-frequency ultrasound combined with alkali pretreatment,alkali pretreatment and untreated corn straw,dual-frequency ultrasound combined with alkali pretreatment was superior to the above three methods in the micro-structure,cumulative biogas yield,gas production rate and net energy.(3)Orthogonal experimental design(OED)and least squares support vector machine(LS-SVM)were combined to optimize the process parameters of anaerobic fermentation.In order to improve the gas production and gas production efficiency of anaerobic fermentation of corn straw,a model of cumulative gas production was established by combining OED with LS-SVM,and the parameters were optimized by using the model.The weight of corn straw,duration of ultrasound,alkali pretreatment time(2%NaOH)and single/dual frequency ultrasound were selected as the experimental factors for OED.LS-SVM combined with OED were used to establish the prediction model between cumulative gas production and pretreatment process parameters.The model was used to analyze the influence of each factor on gas production and the constraints of each factor.The genetic algorithm was used to optimize the optimal pretreatment process parameters of anaerobic fermentation biogas production.Under these parameters,the experiment validated the effectiveness of the method.(4)Virtual sample generation based on experimental design.Adding virtual samples to small sample training set can improve the machine learning ability of the sample set and solve the problem of small sample.However,virtual samples are usually generated according to prior knowledge,which limits the adaptability of virtual samples.To solve the problem of low adaptability of existing methods for generating virtual samples,a method of generating virtual samples by full factor diffusion based on experimental design was proposed.The applicability of this method was the same as that of the experimental design method,which effectively enlarged the application scope of the virtual sample generation method.The effectiveness of this method in improving the accuracy of modeling was proved by the literature data.(5)Process parameters optimization of dual-frequency ultrasound extraction based on virtual samples generated by response surface methodology(BBD-RSM).The extraction of chlorogenic acid and Luteolin from honeysuckle by dual-frequency ultrasound was studied.The experimental factors and levels of BBD-RSM were determined according to single factor experiments.The experiment was designed according to BBD-RSM and the test samples were obtained.According to the design results,the virtual samples were generated by all-factor diffusion.The virtual samples and the test samples together constitute the synthetic samples.LS-SVM was used to establish the prediction model of extraction rate for synthetic samples,and the model was used to analyze the influence of various factors on extraction rate and the relationship between them,and the optimal constraints were obtained..The improved adaptive particle swarm optimization algorithm was used to optimize the model under constraints,and the highest prediction extraction rate and the best process parameters were obtained.The experimental results showed that a higher extraction rate can be obtained under this parameter.更多还原