关键词： Machine learning method   Infrared thermography   Local temperature   Temperature field   Experimental validation  

摘要： Heat distribution in the submerged arc welding (SAW) process has a significant impact on the quality of welds. In this paper, a machine learning method is proposed to predict and analyze temperature in the SAW process. Thermal video data is obtained from an infrared camera at the bottom surface of the workpiece. Programs written in MATLAB are used to extract the temperature history and to generate transient isotherm using image processing techniques. The evolution of the transient isotherm throughout the heating and cooling cycle is analyzed quantitatively. The experimental datasets are suitably prepared for use in machine learning. Prediction of local temperature, temperature-time curve, and temperature field map is made through the proposed machine learning method. Various additional characteristics of the temperature field map are analyzed and predicted. The proposed method is experimentally validated and the predicted results agree closely with the experimental data.

关键词： Gaussian Process Regression   Support Vector Regression   Decision Tree   Tunneling  

摘要： This research intends to use machine learning approaches to predict tunnel geology and its construction time and costs. For this purpose, the Gaussian Process Regression (GPR), Support Vector Regression (SVR), and Decision Tree (DT) have been utilized. An estimation of the geological conditions of the Garan road tunnel and its construction time and cost has been conducted. In addition, after constructing about 200 m from the inlet and outlet sides of the tunnel, using the field-observed data of these sectors in the tools, all the previously forecasted results were updated for unconstructed parts. Fivefold cross-validation has been applied to assess the performance of each model. The obtained models are used to predict construction time and cost in real scenarios, and the accuracy of each model was investigated through different statistical evaluation criteria. Finally, it turns out that all the models provide relatively high performance and reduce the uncertainties of tunnel geology. However, the GPR provides more accurate results compared to the SVR and DT tools. Thus, we recommend the GPR for the prediction of geology and construction time and costs in future levels of a tunnel.

摘要： Drought is one of the most environmentally impactful hydrologic processes with devastating economic consequences for many rural communities in arid and semi-arid countries all over the world. In this research, we have employed satellite data and a stochastic approach for forecasting the changes in lake surface areas and demonstrated for the application of the new technique for the case study of the Lake Gregory in Australia. High-resolution Landsat satellite images are used on a monthly time scale from Landsat 5, 7, and 8, on days that are not cloudy. The software ENVI 5.3, using normalized difference vegetation index (NDVI), and modify normalized difference water index (MNDWI) indices were employed to obtain the lake surface maps, and satellite images have been split into water and non-water using a decision tree. The ArcGIS 10.3 software was used to calculate the area of the Lake monthly. The overall trend data shows that from 2004 to 2019, the LS is steadily declining, reaching its lowest area in *** TRMM satellite monthly precipitation (P) and temperature (T) measurement were obtained to investigate the correlation between these changes and regional precipitation. We developed a novel generalized group method of data handling (GGMDH) to forecast lake surface (LS) fluctuations, in which the LS time-series database is extracted from the satellite imagery. For downscaling, precipitation and three different scenarios are defined based on climate change projections to forecast the LS in the 2020-2060 period. The comparison of the GGMDH with stochastic models integrated with preprocessing scenarios indicates the GGMDH in long-term LS forecasting outperforms the stochastic model. The result showed GGMDH is the best model among other ones to modeling lake surface by R-2 (%) = 94.16, RMSE = 8.77 for the forecasting stage. The forecasted surface of the Lake Gregory fluctuated from 226 to 0.008 km(2) in the future.

关键词： Machine learning  

摘要： Machine learning methods are considered a promising approach for improving operations and processes in manufacturing. However, the application of machine learning often requires the expertise of a data scientist combined with thorough knowledge of the manufacturing processes. Small and medium-sized companies that specialize in certain high value-added, variant rich production processes often lack an in-house data scientist and therefore miss out on generating a deeper data driven insight from their production data streams. This paper proposes a three-step machine learning methodology to empower process experts with limited knowledge in machine learning: 1) data exploration through clustering, 2) representation of the production systems behaviour through specially structured neural networks and 3) querying this representation through evolutionary algorithms to achieve decision support through online optimization or scenario simulation. The chosen algorithms focus on parameter-light, well-established, general use algorithms in order to lower knowledge requirements for their application.

关键词： eddy current sensors   multifrequency eddy currents   machine learning   partial least squares regression   support vector machines   convolutional neural networks   deep learning   data driven approach  

摘要： Eddy current (EC) measurements, widely used for diagnostics of conductive materials, are highly dependent on physical properties and geometry of a sample as well as on a design of an EC-sensor. For a sensor of a given design, the conductivity and thickness of a sample as well as the gap between the sample and the sensor (lift-off) are the most influencing parameters. Estimation of these parameters, based on signals acquired from the sensor, is quite complicated in case when all three parameters are unknown and may vary. In this paper, we propose a machine learning based approach for solving this problem. The approach makes it possible to avoid time and resource-consuming computations and does not require experimental data for training of the prediction models. The approach was tested using independent sets of measurements from both simulated and real experimental data.

关键词： Micritic limestone   Carbonate rock   Machine learning   XGBoost  

摘要： It is important in geophysical applications to relate the compressional and shear ultrasonic wave velocities of micritic limestone to its porosity, volume fraction and density of micrite grains as well as the effective confining pressure. In this paper, this difficulty task is successfully realized by using the most relevant machine learning methods: The artificial neural network method, the support vector machine method and the extreme gradient boosting method (XGB). A relevant dataset available in literature is considered to train and test the models. It is observed that the XGB method significantly outperform the other methods in term of accuracy and training time. It allow obtaining a very high R-squared value of 0.96 and a very small relative root mean squared error of 3% while predicting the sonic velocities from other petrophysical properties. The robustness of the models is also confirmed by studying the sensitivity of the random splittings between the training and the testing sets (C) 2020 Elsevier B.V. All rights reserved.

关键词： knee arthroplasty   prediction   decision support   shared decision making   patient-reported outcomes  

摘要： Background: Approximately 15%-20% of total knee arthroplasty (TKA) patients do not experience clinically meaningful improvements. We sought to compare the accuracy and parsimony of several machine learning strategies for developing predictive models of failing to experience minimal clinically important differences in patient-reported outcome measures (PROMs) 1 year after TKA. Methods: Patients (N = 587) in 3 large Veteran Health Administration facilities completed PROMs before and 1 year after TKA (92% follow-up). Preoperative PROMs and electronic health record data were used to develop and validate models to predict failing to experience at least a minimal clinically important difference in Knee Injury and Osteoarthritis Outcome Score (KOOS) Total, KOOS JR, and KOOS subscales (Pain, Symptoms, Activities of Daily Living, Quality of Life, and recreation). Several machine learning strategies were used for model development. Ten-fold cross-validation and bootstrapping were used to produce measures of overall accuracy (C-statistic, Brier Score). The sensitivity and specificity of various predicted probability cut-points were examined. Results: The most accurate models produced were for the Activities of Daily Living, Pain, Symptoms, and Quality of Life subscales of the KOOS (C-statistics 0.76, 0.72, 0.72, and 0.71, respectively). Strategies varied substantially in terms of the numbers of inputs required to achieve similar accuracy, with none being superior for all outcomes. Conclusion: Models produced in this project provide estimates of patient-specific improvements in major outcomes 1 year after TKA. Integrating these models into clinical decision support, informed consent and shared decision making could improve patient selection, education, and satisfaction. Published by Elsevier Inc.

关键词： Paleokarst reservoirs   Electrofacies   PCA   K-means   LDA  

摘要： The paleokarst system is one of the main carbonate reservoirs, which can form important super-large oil fields. There are many typical paleokarst reservoirs in the Tarim Basin Ordovician strata, mainly composed of caves, vugs, and fractures. Due to the deep burial depth and strong heterogeneity, qualitative identifying the different scale fracture-vuggy reservoirs from the tight limestone around the wellbore is a real challenge in the industrial community. In this paper, machine learning methods were used to classify electrofacies. Firstly, core samples and electrical imaging logging of the paleokarst reservoirs are observed in detail and a core-electrical imaging chart was established. Secondly, conventional logging data was optimized and preprocessed for data mining, using Principal Component Analysis (PCA) algorithm and K-means algorithm. High-resolution electrical imaging logging was chosen as a constraint to recognize electrofacies, and an electrofacies-lithology database was established. Thirdly, based on the electrofacies-lithology database, Linear Discriminant Analysis (LDA) algorithm was used to build an electrofacies prediction model, which can automatically identify the electrofacies in carbonate strata, with a coincidence rate of 92.2%. Finally, the model was used to quantitatively recognize paleokarst reservoirs and their distributions. The electrofacies machine learning workflow proposed in this paper could be used in Tarim Basin and other similar paleokarst reservoirs, which can improve exploration efficiency and save economic cost.

关键词： Nitrogen   RGB image   Simple nonlinear regression   Random forest regression   Backpropagation neural network  

摘要： Crop red-green-blue (RGB) images are powerful tools in nitrogen (N) nutrition estimation. Various regression models using crop N nutrition parameters and image indices have been suggested, but their accuracy and generalization performance for N estimation have not been thoroughly evaluated. In this study, a commercial digital camera was used to capture rice canopy RGB images in a 2-year field experiment, and three regression methods (simple nonlinear regression, SNR;backpropagation neural network, BPNN;and random forest regression, RF) were used for rice shoot dry matter (DM), N accumulation (NA), and leaf area index (LAI) estimation. A repeated random subsampling validation method was performed 1000 times on all three regression methods for the evaluation of model performance and stability. The RF regression models had the highest accuracy for the validation dataset, with average testing prediction accuracy (ATPA) of 80.17%, 79.44%, and 81.82% for DM, LAI, and NA estimation, respectively, followed by BPNN and SNR models. According to the distribution of ATPA in 1000-time calculations, the highest standard deviation (SD) and interval range (5%- 95%) of ATPA was observed in BPNN models, which indicated that the BPNN model was most susceptible to dataset splitting. The lower SD and interval range of ATPA were followed by RF and SNR models, which indicated that the RF and SNR models were less affected by dataset splitting and were able to produce robust regression models consistently. In conclusion, the ensemble algorithm of the RF model effectively prevents overfitting when dealing with different dataset segmentations;thus, the RF model has strong generalization performance. A combination of digital imagery and appropriate machine learning methods facilitates convenient and reliable estimation of crop N nutrition.

关键词： Machine learning   Multiclass classifier   Distance-based classifier  

摘要： Today, machine learning algorithms are an important research area capable of analyzing and modeling data in any field. Information obtained through machine learning methods helps researchers and planners to understand and review systematic problems of their current strategies. Thus, it is very important to work fully in every field that facilitates human life, such as early and correct diagnosis, correct choice, fully functioning autonomous systems. In this paper, a novel machine learning algorithm for multiclass classification is presented. The proposed method is designed based on the Minimum Distance Classifier (MDC) algorithm. The MDC is variance-insensitive because it classifies input vectors by calculating their distances/similarities with respect to class-centroids (average value of input vectors of a class). As it is known, real-world data contains certain proportions of noise. This situation negatively affects the performance of the MDC. To overcome this problem, we developed a variance-sensitive model, which we call Standardized Variable Distances (SVD), considering the standard deviation and z-score (standardized variable) factors. To ensure the accuracy of the SVD, we used Wisconsin Breast Cancer Original (WBCO) and LED Display Domain (led7digit) datasets, which we obtained from UCI machine learning repository, with 5-fold cross validation. It was compared and analyzed classification performance of the SVD with Decision Tree (DT), Random Forest (RF), k-Nearest Neighbor (k-NN), Multinomial Logistic Regression (MLR), Naive Bayes (NB), Support Vector Machine (SVM), and the Minimum Distance Classifier (MDC), which are well-known in the literature. It has also been compared thirteen different studies using the same datasets over the past five years. Our results in the experimental studies have shown that the SVD can classify better than traditional and state-of-the-art methods, compared in this study. The proposed method reached over 97% classification accurac