Main Machine Research Articles

Identification of Significant Features and Data Mining Techniques in Predicting Heart Stroke

The health-care industry generates a large amount of data, which is processed using certain methodologies. One technique that is frequently utilized is data mining. Heart disease is the leading cause of death on a global scale. This system foresees the risk of heart disease developing. The results of this system give you a % likelihood of getting heart disease. Medical parameters are utilized to categories the datasets. This system uses a data mining classification algorithm to analyze such parameters. The datasets are processed in Python programming using two main Machine Learning Algorithms: Decision Tree Approach and Naïve Bayes Algorithm, with the latter showing to be the best algorithm in terms of heart disease accuracy.

Model-Based Field Winding Interturn Fault Detection Method for Brushless Synchronous Machines

The lack of available measurements makes the detection of electrical faults in the rotating elements of brushless synchronous machines particularly challenging. This paper presents a novel and fast detection method regarding interturn faults at the field winding of the main machine, which is characterized because it is non-intrusive and because its industrial application is straightforward as it does not require any additional equipment. The method is built upon the comparison between the theoretical and the measured exciter field currents. The theoretical exciter field current is computed from the main machine output voltage and current magnitudes for any monitored operating point by means of a theoretical healthy brushless machine model that links the main machine with the exciter. The applicability of the method has been verified for interturn faults at different fault severity levels, both through computer simulations and experimental tests, delivering promising results.

Neutronics Modeling and Nuclear Analysis Based on the Full CAD Geometry of CFETR

Radiation field analyses of the fusion reactor are vital to machine design and personal/environmental irradiation protection. Owing to the complicated and toroidal symmetry of fusion reactors, these nuclear analyses have been performed based on a sector model with reflecting boundary conditions. However, not all sections of a fusion reactor are symmetrical in the toroidal direction, particularly the neutron flow channels introduced by auxiliary systems from which particles can leak directly from the plasma. Hence, the reflecting boundary conditions cannot accurately describe the particle transport. Consequently, radiation field analyses based on a full-sector model must be performed to verify the results obtained. In this regard, the neutronics model of CFETR has been built in 360 deg. Meanwhile, the development of the automatic geometry conversion platform cosVMPT has enabled an entire 360-deg model of the CFETR to be established. The model contains all primary components and the outer house building. Sixteen upper/lower ports and six equatorial ports are included, in which two of them are slanted for neutral beam injection, whereas the other ports are filled with a shielding block. The on-the-fly (OTF) global variance reduction method is utilized to accelerate neutron/photon coupling transport. The results show that cosVMPT and the OTF method are reliable, and that the obtained neutron/photon flux is asymmetric outside the main machine. The computational results of the 360-deg model are compared with those of the sector model such that the application scope of simplifying the modeling and calculation using the sector model can be further confirmed.

Exciterless Wound-Field Medium-Power Synchronous Machines: Their History and Future

The excitation system (ES) of the wound-field synchronous machine (WFSM) has remained largely unchanged over the past 50 years. Nonetheless, there has been significant research into exciterless solutions, which, through the integration of the ES with the main machine, aim to offer improved power density, simplified manufacture, and better cooling opportunities.

Sepsis biomarkers and diagnostic tools with a focus on machine learning.

Over the last years, there have been advances in the use of data-driven techniques to improve the definition, early recognition, subtypes characterisation, prognostication and treatment personalisation of sepsis. Some of those involve the discovery or evaluation of biomarkers or digital signatures of sepsis or sepsis sub-phenotypes. It is hoped that their identification may improve timeliness and accuracy of diagnosis, suggest physiological pathways and therapeutic targets, inform targeted recruitment into clinical trials, and optimise clinical management. Given the complexities of the sepsis response, panels of biomarkers or models combining biomarkers and clinical data are necessary, as well as specific data analysis methods, which broadly fall under the scope of machine learning. This narrative review gives a brief overview of the main machine learning techniques (mainly in the realms of supervised and unsupervised methods) and published applications that have been used to create sepsis diagnostic tools and identify biomarkers.

Sawing Force Prediction Model and Experimental Study on Vibration-Assisted Diamond Wire Sawing.

Diamond wire sawing is the main machining technology for slicing various brittle materials, such as crystalline silicon, SiC, and NdFeB. Due to their high hardness and high brittleness, as well as the ease with which the surfaces of machined materials are damaged, it is difficult to further improve the sawing efficiency and the surface quality based on research conducted on the original machining method. In this paper, a vibration-assisted diamond wire sawing method is proposed. We analyzed the impact of load on the ingot, motion trajectory, and sawing depth of the abrasive particles, and a macroscopic sawing force prediction model for the vibration-assisted sawing method was established and verified via experiments. Based on the single-wire-sawing experiment and prediction model, the influences of the vibration parameters and sawing parameters on the sawing force were determined. The influences of vibration assistance on the surface quality, including the roughness profile, waviness profile, thickness profile, Ra, and Rz, were explored through single-wire-sawing experiments, and the influences of vibration assistance on the geometric parameters of slices, such as the total thickness variation (TTV) and warp, were explored through multi-wire-sawing experiments. It was found that vibration-assisted sawing can reduce sawing force and improve surface quality.

Machine learning models in people detection and identification

Introduction: This article is the result of research entitled "Development of a prototype to optimize access conditions to the SENA-Pescadero using artificial intelligence and open-source tools", developed at the Servicio Nacional de Aprendizaje in 2020. Problem: How to identify Machine Learning Techniques applied to computer vision processes through a literature review? Objective: Determine the application, as well as advantages and disadvantages of machine learning techniques focused on the detection and identification of people. Methodology: Systematic literature review in 4 high-impact bibliographic and scientific databases, using search filters and information selection criteria. Results: Machine Learning techniques defined as Principal Component Analysis, Weak Label Regularized Local Coordinate Coding, Support Vector Machines, Haar Cascade Classifiers and EigenFaces and FisherFaces, as well as their applicability in detection and identification processes. Conclusion: The research led to the identification of the main computational intelligence techniques based on machine learning, applied to the detection and identification of people. Their influence was shown in several application cases, but most of them were focused on the implementation and optimization of access control systems, or tasks in which the identification of people was required for the execution of processes. Originality: Through this research, we studied and defined the main machine learning techniques currently used for the detection and identification of people. Limitations: The systematic review is limited to information available in the 4 databases consulted, and the amount of information is variable as articles are deposited in the databases.

Practical recommendations for machine learning in underground rock engineering – On algorithm development, data balancing, and input variable selection

AbstractResearch has demonstrated that machine learning algorithms (MLAs) are a powerful addition to the rock engineering toolbox, and yet they remain a largely untapped resource in engineering practice. The reluctance to adopt MLAs as part of standard practice is often attributed to the ‘opaque’ nature of the algorithms, the complexity in developing them, and the difficulty in determining how the algorithms use the datasets. This article presents tools and processes for developing MLAs, input selection, and data balancing for practical underground rock engineering. MLAs for classification and regression – two main machine learning applications – are presented in terms of developing MLA to extract information from the dataset to obtain the desired output. Engineering verification metrics are selected based on their suitability for specific output. Methods for input selection and data balancing are discussed with a focus on selecting appropriate input data for the problem without introducing bias or excess complexity. Each tool and process for algorithm development, data preparation, and input selection is illustrated with a case study. This article demonstrates that geotechnical practitioners can extract additional value by applying MLAs to rock engineering problems. Once an understanding of the functions of MLAs is reached, the building blocks and open‐source code are available to be adapted to suit the rock mass behaviour of interest.

1-DREAM: 1D Recovery, Extraction and Analysis of Manifolds in noisy environments

Filamentary structures (one-dimensional manifolds) are ubiquitous in astronomical data sets. Be it in particle simulations or observations, filaments are always tracers of a perturbation in the equilibrium of the studied system and hold essential information on its history and future evolution. However, the recovery of such structures is often complicated by the presence of a large amount of background and transverse noise in the observation space. While the former is generally considered detrimental to the analysis, the latter can be attributed to measurement errors and it can hold essential information about the structure. To further complicate the scenario, one-dimensional manifolds (filaments) are generally non-linear and their geometry difficult to extract and model. Thus, in order to study hidden manifolds within the dataset, particular care has to be devoted to background noise removal and transverse noise modeling, while still maintaining accuracy in the recovery of their geometrical structure. We propose 1-DREAM: a toolbox composed of five main Machine Learning methodologies whose aim is to facilitate manifold extraction in such cases. Each methodology has been designed to address particular issues when dealing with complicated low-dimensional structures convoluted with noise and it has been extensively tested in previously published works. However, for the first time, in this work all methodologies are presented in detail, joint within a cohesive framework and demonstrated for three particularly interesting astronomical cases: a simulated jellyfish galaxy, a filament extracted from a simulated cosmic web and the stellar stream of Omega-Centauri as observed with the GAIA DR2. Two newly developed visualization techniques are also proposed, that take full advantage of the results obtained with 1-DREAM. This contribution presents the toolbox in all its details and the code is made publicly available to benefit the community. The controlled experiments on a purposefully built data set prove the accuracy of the pipeline in recovering the real underlying structures.

Using Classification Algorithms in Data Mining in Diagnosing Breast Cancer

Data mining is the process of extracting useful information from large-scale data in an understandable and logical way. According to the main machine learning techniques of data mining; classification and regression, association rules and cluster analysis. Classification and regression are known as predictive models, and clustering and association rules are known as descriptive models. In this study, the classification method was used. With this method, it is aimed to assign a data set to one of the previously determined different classes. The data set used in the study was obtained from the UCIrvine Machine Learning Repository database. The dataset named “Breast cancer” consists of breast cancer data consisting of 699 samples and 10 features collected by William H. at the University of Wisconsin hospital. The data content includes information about the characteristics of some cells analyzed in the detection of breast cancer, cell division, and whether they are benign or malignant. Upon completion of the study, a classification process is performed by determining whether the targeted person has cancerous or non-cancerous cells. In the study carried out in this context; Data mining analyzes were performed using WEKA and Orange programs, SVM (Support Vector Machine), Random Forest algorithms. Along with the analysis results, a comparison was made on the data set, taking into account the previous studies. It is aimed that the conclusions obtained at the end of the study will guide medical professionals working in this field in the diagnosis of breast cancer.

Swing JIB rack stacker crane

Introduction. Rack stacker cranes are the main hoisting and transport machine for maintenance of multi-level rack warehouses. The efficiency and profitability of the warehouse depends on their efficiency. The traditional design of a rack stacker crane includes a trolley that moves along the rails on the floor of a warehouse, a column mounted vertically on the trolley, along which a carriage with a load gripping device and a load rises. The disadvantages of the traditional crane design are the relatively large mass of the trolley, the need to move it along with the column in a horizontal direction when moving around the warehouse, the relatively high energy consumption and the need to strengthen the floor of the warehouse. The crane rail tracks on the floor of the warehouse must be cleared of objects falling from above, which is associated with stopping the operation of the crane and reducing productivity.Materials and Methods. The design of a stacker crane is related to the shape of its working space. In order to improve the stacker crane, its design was developed on the basis of a jib with a counterweight, which does not require rail crane tracks. The crane of the proposed design is capable of storage in the hangars of a semicircular shape. For two warehouses: traditional rectangular and semicircular shapes, a comparative analysis of the sums of geometric distances was performed when the load gripping device moved from the loading point to randomly selected target cells. As a comparison criterion, the sum of the Cartesian distances of the load gripping device movements between the warehouse loading point and the target points, randomly selected according to the law of uniform distribution, was used. Three types of crane operation cycles were studied – simple single, double and mixed.Results. Comparative diagrams of criterion values are given. It has been established that for all types of cycles studied, to the greatest extent when working on a single cycle, the use of a crane of the proposed design with the same storage capacity significantly reduces the total geometric distance that the load gripping device must pass.Discussion and conclusions. Reducing the sum of the geometric distances of movement of the load handling device of the new crane allows to conclude that the time spent on the movements that determine the productivity of the work performed by the crane, as well as the energy costs associated with them, can be significantly reduced. Reducing energy costs is possible by replacing the translational movement of a massive undercarriage with the rotational movement of a balanced jib of a relatively small mass at the crane of the proposed design. Time is not wasted on regular cleaning of the crane tracks, associated with stopping the crane. The number of friction pairs in the proposed design of the crane is reduced compared with the traditional one.

Impact of Training Set Size and Lead Time on Early Tomato Crop Mapping Accuracy

Estimating key crop parameters (e.g., phenology, yield prediction) is a prerequisite for optimizing agrifood supply chains through the use of satellite imagery, but requires timely and accurate crop mapping. The moment in the season and the number of training sites used are two main drivers of crop classification performance. The combined effect of these two parameters was analysed for tomato crop classification, through 125 experiments, using the three main machine learning (ML) classifiers (neural network, random forest, and support vector machine) using a response surface methodology (RSM). Crop classification performance between minority (tomato) and majority (‘other crops’) classes was assessed through two evaluation metrics: Overall Accuracy (OA) and G-Mean (GM), which were calculated on large independent test sets (over 400,000 fields). RSM results demonstrated that lead time and the interaction between the number of majority and minority classes were the two most important drivers for crop classification performance for all three ML classifiers. The results demonstrate the feasibility of preharvest classification of tomato with high performance, and that an RSM-based approach enables the identification of simultaneous effects of several factors on classification performance. SVM achieved the best grading performances across the three ML classifiers, according to both evaluation metrics. SVM reached highest accuracy (0.95 of OA and 0.97 of GM) earlier in the season (low lead time) and with less training sites than the other two classifiers, permitting a reduction in cost and time for ground truth collection through field campaigns.

Experimental investigation of WEDM process parameters on properties of bronze particles using the Taguchi method

Electrical discharge erosion in liquid media is a simple, effective technique to generate multi-scale metallic particles. An experimental investigation was conducted to determine the main machining parameters which contribute to particle size (PS) and size distribution (SD) in WEDM of bronze. Five factors, including three-level built-in processing patterns, wire speed and dielectric medium, are selected as control parameters using an orthogonal array L16(45) Taguchi method, PS and SD of produced powders are the measure of quality. The goal is to obtain micron-scale spherical metal powder with uniform particle size. Results show that cutting-section pattern is the most significant factor, and the electrolyte is the least. The PS and SD performance of the optimized powders is significantly improved by using Taguchi method. Spherical bronze powder with particle size of d10 11.05 μm, d50 31.13 μm and d90 67.48 μm are prepared using the optimized WEDM process parameter, which is A3–B4–C3–D1–E1 including dielectric medium water, cutting class 4, cutting division 3, cutting Sect. 1 and wire speed 200 μm/s. The research confirms that WEDM can be used to produce micron-scale metallic powders, providing cheap and high-quality metallic powders for the metal additive manufacturing industry.

Harvester Maintenance Resource Scheduling Optimization, Based on the Combine Harvester Operation and Maintenance Platform

The combine harvester is the main machine for fieldwork during the harvest season. When the harvester fails and cannot continue to work, this indirectly affects the harvest time and the yield in the field. The emergency maintenance service of agricultural machinery can be optimized through the dynamic planning of harvester maintenance tasks, using the operation and maintenance platform. According to the scene, a priority scheme for the operation and maintenance tasks, based on the improved Q-learning algorithm, was proposed. The continuous approximation capability of the model was improved by using the BP neural network algorithm and the Q function value, in iterations, was updated continuously. At the same time, the improved TOPSIS method, based on Mahalanobis distance, was used to calculate the closeness of each harvester maintenance task, so as to determine the priority of the equipment maintenance tasks. An operation and maintenance service platform for combine harvesters was developed based on the B/S architecture, with the goal of minimizing the operation and maintenance costs and improving the tasks’ complete efficiency. In this research process, dynamic scheduling rules were formulated. Operation and maintenance resources were optimized and rationally allocated through dynamic optimization scheduling methods, and feasible solution information was generated from the operation and maintenance service platform. Finally, the actual data from the enterprise were used for verification and analysis. The verification showed the following: through a comparison of algorithm performance, it was seen that the improved BP-Q-Learning algorithm can quickly find the operation and maintenance scheduling scheme in the maintenance scheduling; the priority rules can improve the efficiency of task execution, to a certain extent; the cost of the tasks’ execution can be significantly reduced; and the maintenance distance can be shortened. This research has reference significance for the formulation and optimization of agricultural machinery maintenance for cross-regional operations.

Comparison of Machine Learning Algorithms for Sand Production Prediction: An Example for a Gas-Hydrate-Bearing Sand Case

This paper demonstrates the applicability of machine learning algorithms in sand production problems with natural gas hydrate (NGH)-bearing sands, which have been regarded as a grave concern for commercialization. The sanding problem hinders the commercial exploration of NGH reservoirs. The common sand production prediction methods need assumptions for complicated mathematical derivations. The main contribution of this paper was to introduce machine learning into the prediction sand production by using data from laboratory experiments. Four main machine learning algorithms were selected, namely, K-Nearest Neighbor, Support Vector Regression, Boosting Tree, and Multi-Layer Perceptron. Training datasets for machine learning were collected from a sand production experiment. The experiment considered both the geological parameters and the sand control effect. The machine learning algorithms were mainly evaluated according to their mean absolute error and coefficient of determination. The evaluation results showed that the most accurate results under the given conditions were from the Boosting Tree algorithm, while the K-Nearest Neighbor had the worst prediction performance. Considering an ensemble prediction model, the Support Vector Regression and Multi-Layer Perceptron could also be applied for the prediction of sand production. The tuning process revealed that the Gaussian kernel was the proper kernel function for improving the prediction performance of SVR. In addition, the best parameters for both the Boosting Tree and Multi-Layer Perceptron were recommended for the accurate prediction of sand production. This paper also involved one case study to compare the prediction results of the machine learning models and classic numerical simulation, which showed the capability of machine learning of accurately predicting sand production, especially under stable pressure conditions.

A New Concept of Jig Rotary Holder System for 3-Axis CNC Milling Machine Operated by the Main Machine Control

Abstract This study aims to develop a new jig holding system that is able to be controlled by a Computer Numeric Control (CNC) installed on three-axis machining centers, which can drastically improve the productivity in machining operations, enabling the machining of unparallel plans in the same setup. An action research methodology was adopted for this work, which, through a practical approach, intends to generate transferrable knowledge to other organizations whose situations are like those in this study. Together, the practical actions and the knowledge acquired create the changes needed for improving these processes. By conducting a case study, it was observed that savings of about 40 % can be easily achieved for parts with low geometric complexity. If the complexity of the parts increases, it is expected that these savings can be even higher. The return of investment is less than 2 years, which is usually affordable for enterprises. Through this study, it was possible to develop a new jig holding system that can be attached to a three-axis CNC machining center and clearly expands its functions and productivity. With this system, it is possible to work in different planes of the part in sequence, as well as use a double-sided table for the jigs, doubling the production batch each time the machine is loaded. Moreover, a list of key settings has been created with the main requirements and recommendations to adopt this kind of production system, which can be highlighted as the main research output.

Investigation of a novel ultrasonically aided electrochemical magnetic abrasive machining process for SS 316L

Many present day applications require machining and finishing of complex-shaped and tough to machine materials with precision and accuracy by conventional or non-conventional machining and polishing processes. Ultrasonically aided electrochemical magnetic abrasive machining (UAEMAM) process is a novel non-conventional machining technique which can machine and finish the product for better surface quality. In present research work, the development and performance evaluation of UAEMAM process for machining of SS 316L is reported. The UAEMAM process is optimised using response surface methodology (RSM) based on central composite design (CCD) and grey relational analysis. The main machining variables of UAEMAM such as work piece rotational speed, working gap, concentration of NaNO3, percentage of abrasives by weight and pulse on time were considered to determine percentage improvement in surface finish (PISF) and material removal rate (MRR). The significant machining factors and the best combination levels of machining parameters associated with PISF and MRR were determined. The present work shows complete elimination of oxide layer from work piece surface, subsequently increasing the PISF value to 84.73 which is 2.73% higher than value achieved by previous studies employed. The benefits of ultrasonic aided electrochemical magnetic abrasive machining were also confirmed from the micrograph observations of the surface. Experimental results show that UAEMAM process is effective for machining and polishing components for better surface finish.

Fabrication and Test of a 1:4 Scaled HTS Coil for the Defocusing Magnet of 2 GeV FFAG Accelerator

The basic design of a 2 GeV fixed field alternating gradient (FFAG) accelerator was proposed by China Institute of Atomic Energy (CIAE) in 2019. A preliminary study of high temperature superconducting (HTS) magnet for the 2 GeV FFAG accelerator lattice scheme was carried out, considering the beam loss of high energy proton beams and energy efficiency of the main machine. In order to demonstrate the feasibility of manufacturing a spiral-shaped HTS magnet with concave edges based on ReBCO tapes, a 1:4 scaled HTS model magnet is being developed in CIAE. First, the design and fabrication of the 1:4 scaled prototype HTS magnet with concave edges will be described in detail. Then, each double pancake (DP) was tested by liquid nitrogen (LN) immersion. The terminal voltage and magnetic field variation were measured and simulated. Finally, the whole assembled HTS magnet was cooled down to 23 K using helium gas pipeline contact cooling and operated on 300 A for over 24 hours. The central magnetic field reached 719.5 mT, without iron. More detailed electromagnetic characteristics of the magnet during the excitation will be illustrated, which is much helpful for the fabrication of the final full-size HTS magnets.

Short-Term Load Forecasting on Individual Consumers

Maintaining stability and control over the electric system requires increasing information about the consumers’ profiling due to changes in the form of electricity generation and consumption. To overcome this trouble, short-term load forecasting (STLF) on individual consumers gained importance in the last years. Nonetheless, predicting the profile of an individual consumer is a difficult task. The main challenge lies in the uncertainty related to the individual consumption profile, which increases forecasting errors. Thus, this paper aims to implement a load predictive model focused on individual consumers taking into account its randomness. For this purpose, a methodology is proposed to determine and select predictive features for individual STLF. The load forecasting of an individual consumer is simulated based on the four main machine learning techniques used in the literature. A 2.73% reduction in the forecast error is obtained after the correct selection of the predictive features. Compared to the baseline model (persistent forecasting method), the error is reduced by up to 19.8%. Among the techniques analyzed, support vector regression (SVR) showed the smallest errors (8.88% and 9.31%).

Portfolio Optimization using Artificial Intelligence: A Systematic Literature Review

Artificial intelligence (AI) models can help investors find portfolios in which the focus is to optimize the risk-return relationship. There are several algorithms and techniques in the literature that allow the application of tests to a set of historical data for the selection and validation of investment portfolios. Based on this, this research intends to examine the contribution of the main machine learning techniques used in portfolio management through a systematic literature review. By using the Methodi Ordinatio for selection and ranking of articles, we classified papers considering object of study, type of AI used, period of analysis, data frequency, balance and cardinality. In addition, we detail the main contributions and trends conceived until the year 2020. Therefore, our findings reveal gaps and suggest future works on the topic.