【作者】 柴琴琴；

【导师】 阳春华； Kok Lay Teo；

【作者基本信息】 中南大学 ， 控制科学与工程， 2012， 博士

【摘要】 氧化铝蒸发过程是回收有用资源、排除杂质和维持整个氧化铝生产循环系统中水量平衡的关键工序,也是主要的耗能工序。蒸发过程的母液质量直接影响磨矿和溶出过程的碱粉消耗量,以及全流程生产的稳定性。随着资源、能源危机和市场竞争的日益加剧,在现有工艺设备条件下,利用优化控制技术来降低蒸发过程能耗、保证母液质量,对提高氧化铝生产过程生产效率和降低能耗具有重要意义。然而,氧化铝蒸发过程流程长,过程参数检测存在大滞后,参数间非线性耦合关系强；且过程受其它工序排出溶液、设备结垢等不确定因素的干扰,因此,蒸发过程仍采用人工控制的方法,存在母液和冷凝水质量不合格、生蒸汽使用过量、能耗高等问题。针对上述问题,论文以四效逆流氧化铝蒸发过程为研究对象,开展蒸发过程母液浓度的预测、蒸发过程能耗分析及优化、以达到最优能效状态为目标的蒸发过程控制方法的研究。主要研究工作和创新性成果如下：1)针对长流程的蒸发过程工况复杂,末效出口母液浓度检测滞后的问题,提出了基于最小二乘支持向量机的末效出口母液浓度预测方法。首先采用独立成分分析方法提取蒸发过程非平稳性数据的主要特征信息,在此基础上,建立了基于最小二乘支持向量机的末效出口母液浓度预测模型。工业现场数据实验结果表明预测精度满足现场实际生产工艺要求,为生产操作提供依据。2)针对影响设备性能的关键因素无法检测的问题,提出了基于鲁棒估计函数的蒸发过程数值计算方法,获得各单元母液浓度值。该方法利用末效出口母液浓度预测结果和过程参数检测结果,以测量误差的鲁棒估计函数为评价指标,基于平衡机理级联模型修正数值计算误差。蒸发过程数值计算结果为过程能耗分析奠定了基础。3)针对蒸发过程能耗高,一般优化问题无法兼顾过程能耗和母液质量的问题,构建了最大化蒸发过程用能效率的优化问题。首先,结合蒸发过程工艺机理和能源消耗特点,基于数值计算结果和有效能分析法分析得到了蒸发过程各部分能源损耗情况。在此基础上,建立了以最大化目的(?)效率和最小化(?)损失率为目标的蒸发过程能效优化模型,求解获得能效最优操作参数设定值。实际数据实验结果表明,该优化方法保证了母液质量,且过程目的(?)效率平均提高了3%左右。4)针对数值计算误差修正问题和蒸发过程能效优化问题中约束条件的复杂性特点,设计了基于不可行度的涡旋粒子群算法来求解约束优化问题。该方法将基于不可行解有效信息的不可行度计算函数增广到目标函数中,形成近似优化问题的适应度函数,然后采用具有自组织特性的涡旋粒子群算法来求解近似优化问题。测试函数优化结果表明了优化算法求解约束优化问题的有效性。5)为使蒸发过程过渡到能效最优的操作条件,在研究蒸发过程动态特性的基础上,建立了各单元蒸发室的动态特性模型,并针对动态模型中滞后时间未知的问题,提出了基于多特征时间点的滞后时间辨识方法。该方法以最小化多采样点时刻模型输出与实测值的偏差为目标,通过求解一组辅助时滞系统方程来获得未知时滞参数的梯度信息,然后采用信赖域内点优化方法求解参数辨识问题。数值算例表明了该方法的准确性和快速性,可有效辨识蒸发过程动态模型滞后时间参数。6)提出了以达到能效最优工艺参数设定值和降低汽水比技术指标为目的蒸发过程控制问题及其求解方法。该方法基于控制参数化技术将控制向量用分段常数函数来近似,通过求解一系列近似控制参数优化选择问题实现了带连续状态不等式约束、多时滞系统的优化控制问题的求解。实验结果表明,优化控制有效降低了蒸汽消耗同时实现了能效最优设定参考轨迹的跟踪。更多还原

【Abstract】 The evaporation process is used to recycle the valuable materials, remove impurities contended in the mother liquor, and maintain water balance of the whole alumina production process. It is a key procedure and a high energy consumption procedure in the production of alumina. The quality of the mother liquor not only influences the alkaline powder consumption of the grinding and dissolution process, but also influences the stability of the whole alumina production process. Thus, it is essential to develop optimal control techniques for the evaporation process, under the existing equipment and the current procedure, such that the specific quality of the mother liquor is met with the highest energy efficiency. In addition, with the increasing resources crisis, energy crisis and competition pressure, optimal control of the evaporation process becomes more and more important in enhancing the productivity and energy efficiency of the whole alumina production process.The evaporation process is a long procedure, thus there is a long time delay in parameter measurement. Moreover, the relationships among the paremeters are highly nonlinear coupled. In addition, there are many uncertainties existed in the evaporation process, such as the fluctuation of the solution discharged from other processes and the scaring of the evaporators. Thus, the currently used manual control method usually leads to unacceptable product mother liquor and condensate, and excessive live steam consumption.Consider these problems arising in the study of the evaporation process; this dissertation takes a practical four-effect countercurrent alumina evaporation process as the research background. It focuses on the mother liquor concentration prediction, the energy consumption analysis and optimization, and the operation control of the process to achieve the optimal energy efficiency. The detailed content and the main contributions are arranged as follows. 1) Due to the long and complex working procedure, there is a large time delay in the measurement of the product mother liquor concentration. To deal with this problem, a least square support vector machine (LSSVM) based model is built to predict the product mother liquor concentration. Firstly, the independent component analysis method is used to extract the main information from the non-stationary process data. On this basis, LSSVM model is established to predict the concentration of the product mother liquor. Using the real data collected from the practical evaporation process, the experimental results show that the prediction precision satisfies the the measurement requirement of practical industrial process. Thus, the prediction result provides guidance for evaporation process operation.2) The key factors for evaluating the performance of the unit cannot be measured. Furthermore, there is no historical concentration data to use. To obtain these concentration values, a numerical calculation method based on robust estimation function is proposed. The proposed numerical calculation method contains two parts:one is the mechanistic model built on balance principle and cascading method; another part is the error correction model based on the robust estimation of the measurement error. In the error correction model, the result of the LSSVM predictive model and the collected data are used. The numerical calculation results are the bases of process energy consumption analysis.3) Since the evaporation process is a high energy consumption process, but the existing optimization problems arising in the study of evaporation process are either aimed at the product quality or purely designed for energy saving, thus an energy efficiency optimization model is built, in which both the product quality and the energy consumption are considered. To construct the optimization model, firstly, on the basis of the numerical calculation results, the energy efficiency and the cause of energy loss are analyzed using the exergy analysis method. Then, an optimization model, whose objective function aiming at minimizing exergy loss rate and maximizing the target exergy efficiency, is built. Solving this optimization model, the opertation parameter settings for optimal energy efficiency can be obtained. The optimization results of the practical industrial production demonstrate that the quality of the mother liquor is guaranteed. Moreover, the target exergy efficiency is increased by3.03%averagely.4) Both the error correction model and the energy consumption optimization model contain complex constraints. To solve the constrained optimization problems, a vortex motion based particle swarm optimization (VMPSO) algorithm is developed. Firstly, an infeasible degree function is constructed based on the information provided by infeasible solutions. The summation of all the infeasible degree function is then appended to the objective function to form an augment objective function. The VMPSO with self-organized characteristic is then developed to solve the approximate optimization problem. The optimization results of several benchmark functions demonstrate the advantage of the proposed optimization algorithm in finding the global optimum.5) To control the evaporation process, the dynamic behavior is investigated. Then, dynamic model in each of the evaporation vessels is established. The dynamic model is in the form of multiple time delays space model with unknown delays. Since the delays influence the behavior of the process and the control effect, thus the delays should first be estimated. For this, a time delay estimation method with multiple characteristic time points is proposed to determine the unknown time delays. The main characteristics of the proposed estimation method are that the cost function of the estimation problem measures the discrepancy between predicted and observed system output; the partial derivatives of this cost function can be computed by solving a set of auxiliary time-delay systems. Using the partial derivatives, the estimation problem is solved by trust region interior point technique effectively. Numerical simulations demonstrate the accuracy and effectiveness of the estimation method. By applying this method, the obtained dynamic system is satisfied.6) To deal with the problems of unacceptable product and high energy usage encountered in the control of evaporation process, an optimal control problem is studied. The aim of this optimal control problem is to follow the desired parameter settings and decrease the steam consumption. This optimal control problem is with both multiple delays and continuous inequality constraints. To solve this optimal control problem, the control is first approximated by a piecewise constant function using the control parameterization technique. Then the optimal control problem is solved effectively through solving a sequence of approximate optimization parameter selection problems subject to only boundedness constraints on decision variables. The developed optimal control method is used to study the problem arising in the evaporation process. The optimal control results obtained based on field data show that the steam comsumption is reduced and the operation parameter settings to achieve the optimal energy efficiency are followed.更多还原