Quality Prediction and Control in Wire Arc Additive Manufacturing via Novel Machine Learning Framework

Wire arc additive manufacturing (WAAM) is capable of rapidly depositing metal materials thus facilitating the fabrication of large-shape metal components. However, due to the multi-process-variability in the WAAM process, the deposited shape (bead width, height, depth of penetration) is difficult to predict and control within the desired level. Ultimately, the overall build will not achieve a near-net shape and will further hinder the part from performing its functionality without post-processing. Previous research primarily utilizes data analytical models (e.g., regression model, artificial neural network (ANN)) to forwardly predict the deposition width and height variation based on single or cross-linked process variables. However, these methods cannot effectively determine the optimal printable zone based on the desired deposition shape due to the inability to inversely deduce from these data analytical models. Additionally, the process variables are intercorrelated, and the bead width, height, and depth of penetration are highly codependent. Therefore, existing analysis cannot grant a reliable prediction model that allows the deposition (bead width, height, and penetration height) to remain within the desired level. This paper presents a novel machine learning framework for quantitatively analyzing the correlated relationship between the process parameters and deposition shape, thus providing an optimal process parameter selection to control the final deposition geometry. The proposed machine learning framework can systematically and quantitatively predict the deposition shape rather than just qualitatively as with other existing machine learning methods. The prediction model can also present the complex process-quality relations, and the determination of the deposition quality can guide the WAAM to be more prognostic and reliable. The correctness and effectiveness of the proposed quantitative process-quality analysis will be validated through experiments.

A Model of Catalytic Cracking: Catalyst Deactivation Induced by Feedstock and Process Variables

Changes in the quality of the feedstocks generated by involving various petroleum fractions in catalytic cracking significantly affect catalyst deactivation, which stems from coke formed on the catalyst surface. By conducting experimental studies on feedstocks and catalysts, as well as using industrial data, we studied how the content of saturates, aromatics and resins (SAR) in feedstock and the main process variables, including temperature, consumptions of the feedstock, catalyst and slops, influence the formation of catalytic coke. We also determined catalyst deactivation patterns using TG-DTA, N2 adsorption and TPD, which were further used as a basis for a kinetic model of catalytic cracking. This model helps predict the changes in reactions rates caused by coke formation and, also, evaluates quantitatively how group characteristics of the feedstock, the catalyst-to-oil ratio and slop flow influence the coke content on the catalyst and the degree of catalyst deactivation. We defined that a total loss of acidity changes from 8.6 to 30.4 wt% for spent catalysts, and this depends on SAR content in feedstock and process variables. The results show that despite enriching the feedstock by saturates, the highest coke yields (4.6–5.2 wt%) may be produced due to the high content of resins (2.1–3.5 wt%).

Learning Processes in Interactive CALL Systems: Linking Automatic Feedback, System Logs, and Learning Outcomes

Interactive digital tools increasingly used for language learning can provide detailed system logs (e.g., number of attempts, responses submitted), and thereby a window into the user’s learning processes. To date, SLA researchers have made little use of such data to understand the relationships between learning conditions, processes, and outcomes. To fill this gap, we analyzed and interpreted detailed logs from an ICALL system used in a randomized controlled field study where 205 German learners of English in secondary school received either general or specific corrective feedback on grammar exercises. In addition to explicit pre-/post-test results, we derived 19 learning process variables from the system log. Exploratory factor analysis revealed three latent factors underlying these process variables: effort,accuracy focus, and time on task. Accuracy focus and finish time (a process variable that did not load well on any factors) significantly predicted pre-/post-test gain scores with a medium effect size. We then clustered learners based on their process patterns and found that the specific feedback group tended to demonstrate particular learning processes and that these patterns moderate the advantage of specific feedback. We discuss the implications of analyzing system logs for SLA, CALL, and education researchers and call for more collaboration.

Locally Linear Embedding Orthogonal Projection to Latent Structure

Quality variables are measured much less frequently and usually with a significant time delay by comparison with the measurement of process variables. Monitoring process variables and their associated quality variables is essential undertaking as it can lead to potential hazards that may cause system shutdowns and thus possibly huge economic losses. Maximum correlation was extracted between quality variables and process variables by partial least squares analysis (PLS) (Kruger et al. 2001; Song et al. 2004; Li et al. 2010; Hu et al. 2013; Zhang et al. 2015).

Application of machine learning to predict the thermal power plant process condition

This paper deals with the development of an algorithm for predicting thermal power plant process variables. The input data are described, and the data cleaning algorithm is presented along with the Python frameworks used. The employed machine learning model is discussed, and the results are presented.

Probabilistic Graphical Model for Continuous Variables

Most of the sampled data in complex industrial processes are sequential in time. Therefore, the traditional BN learning mechanisms have limitations on the value of probability and cannot be applied to the time series. The model established in Chap. 10.1007/978-981-16-8044-1_13 is a graphical model similar to a Bayesian network, but its parameter learning method can only handle the discrete variables. This chapter aims at the probabilistic graphical model directly for the continuous process variables, which avoids the assumption of discrete or Gaussian distributions.

Effects of process variables on the reactivity of slaked lime produced from Shuk quicklime

This is a research report of the effects of process variables on the reactivity of slaked lime produced from Shuk quicklime. It involved the calcination (at temperature of 1000 0C, particle size of 90 µm and time of 3 hrs) of Shuk limestone and subsequent slaking of its quicklime. The quicklime was characterized by x-ray diffractometer (XRD) and scanning electron microscopy (SEM) respectively to determine its mineral content and surface morphology respectively. Effects of process variables (quicklime/water ratio, particle size and time) on the reactivity of the slaked lime were determined. The reactivity was optimized using response surface methodology (RSM). The XRD analysis revealed calcite as the type mineral of the Shuk quicklime. The surface morphology of the quicklime sample showed that the particles are packed together in powdered form with visible pores that will allow passage of water. Reactivity of the lime was influenced by the quicklime/water ratio, particle size and time. Quadratic model appropriately explained the relationship between reactivity and considered slaking factors of quicklime/water ratio, particle size and time. The optimum reactivity value of the slaked lime was obtained as 59.3 oC at quicklime/water ratio of 0.24 g/ml, particle size of 88.2 µm and time of 15.1 minutes.

Optimization and Sensitivity Analysis of the Cutting Conditions in Rough, Semi-Finish and Finish Honing

Honing processes are currently employed to obtain a cross-hatched pattern on the internal surfaces of cylinders that favors oil flow in combustion engines or hydraulic cylinders. The main aim of the present paper is to optimize the machining conditions in honing processes with respect to surface roughness, material removal rate and tool wear by means of the desirability function. Five process variables are considered: grain size, density, pressure, linear speed and tangential speed. Later, a sensitivity analysis is performed to determine the effect of the variation of the importance given to each response on the results of the optimization process. In the rough and semi-finish honing steps, variations of less than 5% of the importance value do not cause substantial changes in the optimization process. On the contrary, in the finish honing step, small changes in the importance values lead to modifications in the optimization process, mainly regarding pressure. Thus, the finish honing phase is more sensitive to changes in the optimization process than the rough and the semi-finish honing phases. The present paper will help users of honing machines to select proper values for the process variables.

Artificial neural network-based modelling of optimized experimental study of xylanase production by Penicillium citrinum xym2

The industrial production of enzymes is generally optimized by one-factor-at-a-time (OFAT) approach. However, enzyme production by the method involves submerged or solid-state fermentation, which is laborious and time-consuming and it does not consider interactions among process variables. Artificial neural network (ANN) offers enormous potential for modelling biochemical processes and it allows rational prediction of process variables of enzyme production. In the present work, ANN has been used to predict the experimental values of xylanase production optimized by OFAT. This makes the reported ANN model to predict further optimal values for different input conditions. Both single hidden layered (6-3-1) and double hidden layered (6-12-12-1) were able to closely predict the actual values with MSE equals to 0.004566 and 0.002156, respectively. The study also uses multiple linear regression (MLR) analysis to calculate and compare the outcome with ANN predicted xylanase activity, and to establish a parametric sensitivity.