Robust PSD determination of micro and meso-pore adsorbents via novel modified U curve method

Proper estimation of pore size distribution (PSD) for various solid adsorbents is essential for efficient characterization of such materials. PSD determination is inherently an ill-posed problem which can be solved by resorting to Tikhonov regularization technique. Automatic selection of the optimal regularization level is crucial for efficient extraction of PSD from experimental adsorption or condensation isotherms. A detailed comparison of Leave One Out Cross Validation, L-curve, U-curve and modified L-curve criteria are presented using several real adsorption and condensation case studies. A modified U-curve criterion is also proposed and its superior performances are thoroughly validated. The new method employs a relatively simple minimization procedure, which is much easier than quite complex and tedious procedures used by conventional methods to compute their internal optimum parameters. Other practical issues, such as efficient computation of L-corner have been received proper attention in this article and a simple algorithm is presented for this purpose.     

基于L曲线的自适应时空正则化视频超分辨率重建

为了解决视频超分辨率重建的病态问题,以得到良好的重建效果,提出了一种新颖的视频超分辨率重建算法。在算法中引入了时空联合正则化算子,通过视频帧本身的空间平滑信息和视频相邻帧的帧间相关先验信息的引入,提高了解的质量;同时,为了选择合适的时空正则化系数,提出了基于L曲线的自适应时空正则化系数计算方法,可以自适应地计算合适的正则化系数。通过对模拟图像序列和真实视频序列的实验结果表明,算法能得到较为精确的解,重建出具有良好视觉效果的高分辨率视频。     

结构识别计算中基于L曲线的模型确认方法研究

模型确认方法的判断标准是考察包含识别参数的模型对待识别结构进行描述的精确程度。但由于对待识别结构系统的认识不足以及计算效率的考虑,通常会产生识别参数的不当选取问题,导致模型对待识别结构的描述总是存在不完善之处;当这种不完善较显著时,传统的模型确认方法通常会失效。为此,本文结合结构识别问题的求解可采用正则化技术来改善其不适定性的特点,引入不同识别结果 所对应目标函数与 的范数之间的对应规律作为先验条件,提出了一种基于L曲线的模型确认方法,有望弥补传统模型确认方法的上述不足。以一弹簧-质量系统、BENCHMARK结构分别作为仿真算例、试验算例,对所得模型进行了确认;结果表明：传统确认方法不能准确指出试验算例的合理模型,而基于L曲线的模型确认方法对仿真算例与试验算例均可指出合理模型;从而证明了本文研究的基于L曲线的模型确认方法的有效性。     

Inverse combustion force estimation based on response measurements outside the combustion chamber and signal processing

Exposure to vibration has various physiological effects on vehicle passengers. Engine is one of the main sources of vehicle vibration. The major causes of engine vibration are combustion forces transmitted through the pistons and connection rods. Evaluation of sources is the first step to attenuate this vibration. Assessment of these sources is not an easy task because internal parts of machinery are not accessible. Often, instrumentation for such systems is costly, time consuming and some modifications would be necessary. Aim of the first part of this paper was to validate an inverse technique and carry out mobility analysis on a vehicle crankshaft to achieve matrix of Frequency Response Functions (FRFs). Outcomes were implemented to reconstruct the applied force for single and multiple-input systems. In the second part, the validated inverse technique and FRFs were used to estimate piston forces of an operating engine. Bearings of crankshaft were chosen as nearest accessible parts to piston connecting rods. Accelerometers were connected to the bearings for response measurement during an ideal engine operation. These responses together with FRFs, which were estimated in the previous part, were utilised in the inverse technique. Tikhonov regularization was used to solve the ill-conditioned inverse system. Two methods, namely L-curve criterion and Generalized Cross Validation (GCV), were employed to find the regularization parameter for the Tikhonov method. The inverse problem was solved and piston forces applied to crankpins were estimated. Results were validated by pressure measurement inside a cylinder and estimating the corresponding combustion force. This validation showed that inverse technique and measurement outcomes were roughly in agreement. In presence of various noise, L-curve criterion conduces to more robust results compared to the GCV method. But in the absence of high correlation between sources (f>600 HzHz), the GCV technique leads to more accurate results. This research shows that inverse techniques have great ability to estimate vibration sources inside the machinery.     

一种基于最优输出跟踪的多源动态载荷识别方法

基于最优输出跟踪的基本思想,提出了一种时域内多源动态载荷识别的方法。该方法从结构动力响应出发,设计一个最优输出跟踪器并构造性能指标,将载荷识别问题转变为最优输出跟踪问题。通过Adams法求解微分方程,实现了多源动态载荷的识别,并采用L曲线法确定了性能指标中的关键参数。数值算例表明,所述的载荷识别方法能够在响应数据含有噪声的情况下,有效稳定地实现多源动态载荷的重构,具有较强的抗噪能力。     

Numerical boundary identification for Helmholtz-type equations

We study the identification of an unknown portion of the boundary of a two-dimensional domain occupied by a material satisfying Helmholtz-type equations from additional Cauchy data on the remaining known portion of the boundary. This inverse geometric problem is approached using the boundary element method (BEM) in conjunction with the Tikhonov first-order regularization procedure, whilst the choice of the regularization parameter is based on the L-curve criterion. The numerical results obtained show that the proposed method produces a convergent and stable solution     

包装系统非线性特性识别方法

目的建立一种能同时识别包装系统中非线性弹性和非线性阻尼特性的方法。方法首先将单自由度包装系统运动方程式推导为第1类Volterra非线性积分方程式,为在包含噪声的条件下准确求出方程的解,采用Landweber迭代正则化方法求解该积分方程,由于迭代次数对求解结果的准确性有很大的影响,文中采用L曲线准则确定迭代的次数。在求解Volterra方程后,利用方程的解和系统自由响应数据同时识别包装系统非线性弹性和非线性阻尼。结果文中采用一个数字实例验证了该方法的准确性。结论文中提出的方法可有效识别包装系统的非线性特性。     

位场向下延拓的波数域正则—积分迭代法

本文将正则方法与积分迭代法相结合,提出波数域正则—积分迭代法用于位场向下延拓。该迭代法以位场正则向下延拓值为迭代初值,并用正则向下延拓算子处理迭代过程中的剩余谱,使得在加快迭代收敛的同时,压制噪声谱,以提高延拓的稳定性;计算了改进迭代法的最优步长,对比分析了改进迭代法的收敛性和滤波特性,同时提出采用L-曲线准则计算正则向下延拓的正则参数,并将其扩展到波数域以提高计算速度和适应波数域迭代法。基于理论模型和航磁数据的对比试验分析表明,改进迭代法位场向下延拓的精度、稳定性及收敛速度都有提高,其延拓效果优于现有广义逆法、积分迭代法、泰勒级数迭代法和Landweber迭代法。     

基于径向谱的位场向下延拓正则参数选取方法

Tikhonov正则算法是解决位场向下延拓不适定性广泛采用的一种有效方法,最优正则参数的选取对下延精度和计算时间都有影响.为了快速、有效地选取正则参数,提出一种基于位场径向平均功率谱的Tikhonov正则参数选取方法.该方法通过建立由径向平均功率谱确定的截止波数和Tikhonov正则低通滤波函数的截止波数的关系来确定正则参数.基于理论重力模型数据及航磁实测数据将新方法同以往普遍采用的L-曲线法、C-范数法进行了正则向下延拓效果的对比实验.实验结果表明,新方法实现简单、物理意义明确,且选取的正则参数优于另外两种经典方法.     

正则化算法在遥感影像外定向中的应用

正则化的目的是在求解过程中对方程的解进行约束,得到稳定可用的解.遥感影像外定向是确定外方位元素,它是正射纠正和三维信息提取的基础.在计算卫星遥感影像外方位元素时存在病态问题.本文使用两种正则化方法求解这一病态问题:Tikhonov正则化和两步解法.这两种方法与主成分估计法和岭-压组合估计法进行了比较,实验结果表明,基于GCV参数选取方法的Tikhonv正则化方法和主成分估计及两步解法求解结果较优.由不同精度的控制点进行外定向的检验精度表明,除了应用较优的算法,精确的控制点对外定向也很重要.