Incorporation of prior information on parameters in inverse transient analysis for leak detection and roughness calibration

Leak detection and calibration of hydraulic models for water distribution systems (WDS), including the pipe roughness and other parameters, is undoubtedly a critically important issue. This paper first briefly reviews the relevant literature. After addressing some general issues regarding prior information (PI) on calibration parameters, an inverse transient model, capable of detecting leaks and internal pipe roughness in WDS, is presented. The classic objective function of the weighted least squares type is generalized by adding terms accounting for PI on calibration parameters. The methodology presented is general in that it can be applied to any WDS inverse model. It is then shown that PI, when used in a certain way, conditions the inverse problem and thus improves the solution obtained. As a part of the suggested approach, a procedure for careful and effective use of PI on the parameters in inverse transient problems is suggested. Finally, the methodology is applied to three case studies, leading to some general conclusions about the value and use of PI.     

基于深度SSD改进模型的传动设备状态在线监测研究

针对现有传动设备在线监测算法存在的检测精度地、效率差等问题,提出一种基于改进SSD网络模型的在线检测算法。先对故障集进行预处理,通过滤波调制、共振解调等环节滤除原始故障集的噪声干扰;以VGG-16为基础设计了SSD网络结构,同时增加了辅助卷积层和预测层;对SSD网络模型进行改进,引入了注意力机制模块和特征增强模块,改善模型各层的数据共享性能同时提高了模型的数据训练效率;基于通道拼合方式对故障数据进行多尺度特征融合,并优化SSD模型的各层金字塔结构,以更好的匹配先验框及选择最佳的损失函数。实验结果显示,提出算法的传动设备故障检测率达到98.8%,同时算法的检测效率也优于现有算法。     

验前信息的决策方法

在一些实际的可靠性工程中,对产品提出可靠性指标(如可靠度、可靠度置信下限等)要求的同时,试验经费又决定了只能做少量的试验.在这种小子样数据情况下,Bayes方法所获的可靠性评估结论依赖于验前信息.为了评定产品是否满足可靠性指标,须给出验前信息的决策域.文中提出一种验前信息决策域的计算方法,并将其用于计算成败型和指数寿命分布型产品验前信息的决策域.通过两个算例说明以上方法是有效可行的.     

Incorporating prior model into Gaussian processes regression for WEDM process modeling

Sufficient sampling is usually time-consuming and expensive but also is indispensable for supporting high precise data-driven modeling of wire-cut electrical discharge machining (WEDM) process. Considering the natural way to describe the behavior of a WEDM process by IF-THEN rules drawn from the field experts, engineering knowledge and experimental work, in this paper, the fuzzy logic model is chosen as prior knowledge to leverage the predictive performance. Focusing on the fusion between rough fuzzy system and very scarce noisy samples, a simple but effective re-sampling algorithm based on piecewise relational transfer interpolation is presented and it is integrated with Gaussian processes regression (GPR) for WEDM process modeling. First, by using re-sampling algorithm encoded derivative regularization, the prior model is translated into a pseudo training dataset, and then the dataset is trained by the Gaussian processes. An empirical study on two benchmark datasets intuitively demonstrates the feasibility and effectiveness of this approach. Experiments on high-speed WEDM (DK7725B) are conducted for validation of nonlinear relationship between the design variables (i.e., workpiece thickness, peak current, on-time and off-time) and the responses (i.e., material removal rate and surface roughness). The experimental result shows that combining very rough fuzzy prior model with training examples still significantly improves the predictive performance of WEDM process modeling, even with very limited training dataset. That is, given the generalized prior model, the samples needed by GPR model could be reduced greatly meanwhile keeping precise.     

Paying attention to adult learners online: The pedagogy and politics of community

Our article profiles the evolution of a fully online writing course designed for adult learners in our university's Prior Learning Assessment Program. Based on our own observations and experiences teaching adult learners online, we question if the virtual learning environment presents different challenges and prospects for the adult learner versus the traditional student learner, along with an extension and complication of the more social metaphors of “virtual community.” Moreover, because of the changing demographic from traditional to adult students, we argue that this change also fosters a change in the relationship between teachers and students. In chronicling this relationship, we note problems when the labor of adult education becomes invisible to those supervising online instructors. Because of these “invisible” labor issues, we argue that successful online instruction must include a range of interactions between students and instructors that extend the more public concept of community to better acknowledge the importance of personal, private interaction. Thus, we conclude with a call to rethink our online writing pedagogies to be more flexible to adult learner needs and learning styles, simultaneously recognizing the impact of adult online education on faculty workload.     

Two phase visualization by electrical impedance tomography with prior information

Numerical and experimental work was conducted to develop a visualization technique for the phase distribution in a two-phase flow field with known internal structures by electrical impedance tomography technique, which reconstructs the resistivity distribution with the electrical responses that are determined by corresponding excitations. The finite element method is employed to solve the electrical field induced by the currents through electrodes placed along the boundary and a modified Newton-Raphson iterative method is used to determine the search step minimizing the error between the calculated and the measured voltages at the electrodes. The locations and resisitivities of the known structures are considered as prior information. To mitigate the ill-posedness of inverse problem and to incorporate prior information, the modified Tikhonov regularization technique is employed. Also, with an apparatus developed for impedance imaging this study attempts to reconstruct the images of the simulated bubble distributions and the reconstructed images imply the potential possibility of the electrical impedance tomography for the two-phase flow visualization.     

A method for calibrated maximum likelihood classification of forest types

A modification to the maximum likelihood algorithm was developed for classification of forest types in Sweden's part of the CORINE land cover mapping project. The new method, called the “calibrated maximum likelihood classification” involves an automated and iterative adjustment of prior weights until class frequency in the output corresponds to class frequency as calculated from objective (field-inventoried) estimates. This modification compensates for the maximum likelihood algorithm's tendency to over-represent dominant classes and under-represent less frequent ones. National forest inventory plot data measured from a five-year period are used to estimate relative frequency of class occurrence and to derive spectral signatures for each forest class. The classification method was implemented operationally within an automated production system which allowed rapid production of a country-wide forest type map from Landsat TM/ETM+ satellite data. The production system automated the retrieval and updating of forest inventory plots, a plot-to-image matching routine, illumination and haze correction of satellite imagery, and classification into forest classes using the calibrated maximum likelihood classification. This paper describes the details of the method and demonstrates the result of using an iterative adjustment of prior weights versus unadjusted prior weights. It shows that the calibrated maximum likelihood algorithm adjusts for the overclassification of classes that are well represented in the training data as well as for other classes, resulting in an output where class proportions are close to those as expected based on forest inventory data.     

Computational Bayesian inference for estimating the size of a finite population

We develop computationally intensive Bayesian methods to estimate the size of a closed population and apply these methods to estimate the number of children born in upstate New York with spina bifida from 1969 to 1974. The names of these children may appear on three different administrative lists: medical; birth; and death records. We assume diffuse prior distributions on the marginal probabilities of a name appearing on each record and on the various odds ratios modeling the interactions of these lists. Samples from the posterior distribution are generated using a modified sample-resample technique. A Bayesian log-linear model is developed and the posterior distribution is sampled from a Markov chain generated using the Metropolis algorithm. These two approaches are compared in terms of their interpretability and computational complexity.     

Do prior online course outcomes provide more information than G.P.A. alone in predicting subsequent online course grades and retention? An observational study at an urban community college

In this study, prior online course outcomes and pre-course enrollment G.P.A. were used as predictors of subsequent online course outcomes, and the interaction between these two factors was assessed in order to determine the extent to which students with similar G.P.A.'s but with different prior online course outcomes may differ in their likelihood of successfully completing a subsequent online course. This study used a sample of 962 students who took an online course at a large urban community college from 2004 to 2010. Results indicate that prior online course experience is a very significant predictor of successful completion of subsequent online courses, even more so than G.P.A. For students with no prior online course experience, G.P.A. was a good predictor of future online course outcomes; but for students with previous online course experience prior online course outcomes was a more significant predictor of future online course grades and retention than G.P.A.     

Optimistic bias about online privacy risks: Testing the moderating effects of perceived controllability and prior experience

This study examined the ways in which Internet users construct their risk judgments about online privacy. The results, based on telephone survey data from a national probability sample in Singapore (n = 910), revealed that (a) individuals distinguish between two separate dimensions of risk judgment (personal level and societal level), (b) individuals display a strong optimistic bias about online privacy risks, judging themselves to be significantly less vulnerable than others to these risks, and (c) internal belief (perceived controllability) and individual difference (prior experience) significantly moderate optimistic bias by increasing or decreasing the gap between personal- and societal-level risk estimates. The implications of the findings for research and practice are discussed.