Avoiding neural network fine tuning by using ensemble learning: application to ball-end milling operations

Surface roughness plays a key role in the performance of machined components??specially dies and moulds??manufactured for the aerospace and automotive industries, among others. However, roughness can only be measured off-line after the part has been machined, when cutting conditions may no longer be adjusted to surface roughness requirements. A reliable surface roughness prediction application is presented in this paper. It is based on ensemble learning for vertical high-speed milling operations with ball-end mills for finishing operations on quenched steel 1.2344 (AISI H13) that are widely used in the manufacture of moulds and dies. The new approach was validated with an experimental dataset that includes geometrical tool factors, cutting conditions, dynamic factors and lubricant type. An intensive comparison with an artificial neural network approach for the same dataset is included, to reveal the improvements of the new technique over other well-established ones for this industrial problem. This comparison shows that ensemble learning can by-pass the time-consuming task of tuning neural network parameters and can also improve prediction model accuracy, both of which are features that could lead to greater use of these kinds of prediction models in real workshops. Finally, a methodology, based on this new approach, is presented, in order to illustrate how the prediction model can be used in workshops to optimize cutting conditions in terms of their surface quality and productivity.     

On-line 3-D system for the inspection of deformable parts

In spite of the development of automated tolerance inspection systems for manufactured parts over the years, there are still processes that inevitably require manual intervention making full automation impossible in most cases; in particular when dealing with deformable parts. In most current industrial inspection systems, a deformable part under inspection must first be mechanically constrained on a rigid support or jig so as to be able to compare it with its nominal shape. This paper presents a new system to perform real-time surface inspection of deformable parts that does not require fixturing. Instead, the proposed system applies virtual forces to the part??s CAD model as if the part was installed in the fixturing device. Normally, a precise finite element method (FEM) simulation should be used to approximate the deformation that appends when the part is installed in the device. Even with a fast parallel computer, FEM is far from being real-time and cannot be used for on-line inspection. In the proposed system, a radial basis function approximation of the FEM simulation is trained off-line and used to speed-up the simulation by an order of magnitude. Experimental evaluation of the proposed system is presented for three plastic parts. Using the proposed scheme, an approximation of 0.25?mm compared with the real deformation was obtained. In this paper, statistical results are presented such as the average deviation, standard deviation, and processing time between the approximations obtained with the proposed method and with the finite element method applied to the full CAD model.     

Experimental study and modelling of tool temperature distribution in orthogonal cutting of AISI 316L and AISI 3115 steels

Cutting tool temperature distribution was mapped using the IR-CCD technique during machining of carbon steel AISI 3115 and stainless steel AISI 316L under orthogonal cutting conditions using flat-face geometry inserts. The effect of work material treatment on tool temperature was investigated, and the results showed that AISI 3115 in heat-treated state displayed higher tool temperature than the as-rolled state. Stainless steel 316L with high sulphur content (0.027?wt.%) and calcium treatment displayed lower cutting tool temperature than the variant with low sulphur (0.009?wt.%). The experimental results were compared with theoretical tool temperature distributions based on a modified version of Komanduri and Hou??s analytical model. In particular, variable frictional heat source and secondary shear were introduced and modelling of the tool stress distribution on rake surface was also considered.     

Effect of very high cutting speeds on shearing, cutting forces and roughness in dry turning of austenitic stainless steels

Behavior of austenitic stainless steels has been studied at very high cutting speeds. Turning tests were carried out using the AISI 303 austenitic stainless steel. In particular, the influence of cutting speed on tool wear, surface quality, cutting forces and chip geometry has been investigated. These parameters have been compared when performing machining at traditional cutting speeds (lower than 350?m/min) versus high cutting speeds. The analysis of results shows that the material undergoes a significant change in its behavior when machining at cutting speeds above 450?m/min, that favors the machining operation. The main component of cutting forces reaches a minimum value at this cutting speed. The SEM micrographs of the machined surfaces show how at the traditional cutting speeds the machined surfaces contain cavities, metal debris and feed marks with smeared material particles. Surfaces machined at high cutting speeds show evidence of material side flow, which is more evident at cutting speeds above 600?m/min. Tool wear is located at the tool nose radius for lower cutting speeds, whereas it slides toward the secondary edge when cutting speed increases. An analysis of chips indicates also an important decrement in chip thickness for cutting speeds above 450?m/min. This study concludes that there is an unexplored range of cutting speeds very interesting for high-performance machining. In this range, the behavior of stainless steels is very favorable although tool wear rate is also significant. Nevertheless, nowadays the cost of tool inserts can be considered as secondary when comparing to other operation costs, for instance the machine hourly cost for high-end multitasking machines.     

Simulation of low rigidity part machining applied to thin-walled structures

The aim of this study is to evaluate the modelling of machining vibrations of thin-walled aluminium workpieces at high productivity rate. The use of numerical simulation is generally aimed at giving optimal cutting conditions for the precision and the surface finish needed. The proposed modelling includes all the ingredients needed for real productive machining of thin-walled parts. It has been tested with a specially designed machining test with high cutting engagement and taking into account all the phenomena involved in the dynamics of cutting. The system has been modelled using several simulation techniques. On the one hand, the milling process was modelled using a dynamic mechanistic model, with time domain simulation. On the other hand, the dynamic parameters of the system were obtained step by step by finite element analysis; thus the variation due to metal removal and the cutting edge position has been accurately taken into account. The results of the simulations were compared to those of the experiments; the discussion is based on the analysis of the cutting forces, the amplitude and the frequency of the vibrations evaluating the presence of chatter. The specific difficulties to perfect simulation of thin-walled workpiece chatter have been finely analysed.     

Analytical models of composite material drilling

Drilling of composite material structures is widely used for aeronautical assemblies. When drilling, damage to the composite laminate is directly related to the cutter geometry and the cutting conditions. Delamination of the composite materials at the hole exit as directly related to the axial force (F _Z) of the cutter is considered to be the major such defect. To address this issue, an orthotropic analytical model is developed in order to calculate the critical force of delamination during drilling and a number of hypotheses for loading are proposed. This critical axial load is related to the delamination conditions (propagation of cracks in the last layers) and the mechanical characteristics of the composite material machined. A numerical model is also drawn up to allow for numerical validation of the analytical approach. A comparison between these analytical and numerical modellings and experimental results from quasi-static punch tests led to the choice of the loading hypothesis closest to the experimental conditions. The selection of corresponding load permits to model the drilling critical thrust force on delamination and then to optimise the cutting conditions. The dimensions and geometrical shape of the cutter are of considerable importance when it comes to choosing this load. The present article focuses on the case of the twist drill, which is commonly used to drill thick plates. However, this work can be adapted to different cutter geometries.     

Effect on the journal impact factor of the number and document type of citing records: a wide-scale study

We studied the effect on journal impact factors (JIF) of citations from documents labeled as articles and reviews (usually peer reviewed) versus citations coming from other documents. In addition, we studied the effect on JIF of the number of citing records. This number is usually different from the number of citations. We selected a set of 700 journals indexed in the SCI section of JCR that receive a low number of citations. The reason for this choice is that in these instances some citations may have a greater impact on the JIF than in more highly-cited journals. After excluding some journals for different reasons, our sample consisted of 674 journals. We obtained data on citations that contributed to the JIF for the years 1998?C2006. In general, we found that most journals obtained citations that contribute to the impact factor from documents labeled as articles and reviews. In addition, in most of journals the ratio between citations that contributed to the impact factor and citing records was greater than 80% in all years. Thus, in general, we did not find evidence that citations that contributed to the impact factor were dependent on non-peer reviewed documents or only a few citing records.     

Design of a Calibration System for Heat Flux Meters

Accurate heat flux measurements are needed to gain a better knowledge of the thermal performance of buildings and to evaluate the heat exchange among various parts of a building envelope. Heat flux meters (HFMs) are commonly used both in laboratory applications and in situ for measuring one-dimensional heat fluxes and, thus, estimating the thermal transmittance of material samples and existing buildings components. Building applications often requires heat flux measurements below 100 W · m^?2. However, a standard reference system generating such a low heat flux is available only in a few national metrology institutes (NMIs). In this work, a numerical study aimed at designing an HFM calibration apparatus operating in the heat flux range from 5 W·m^?2 to 100 W · m^?2 is presented. Predictions about the metrological performance of such a calibration system were estimated by numerical modeling exploiting a commercial FEM code (COMSOL^®). On the basis of the modeling results, an engineered design of such an apparatus was developed and discussed in detail. The system was designed for two different purposes: (i) for measuring the thermal conductivity of insulators and (ii) for calibrating an HFM with an absolute method (i.e., by measuring the applied power from the heater and its active cross section) or by a relative method (i.e., by measuring the temperature drop across a reference material of known thickness and thermal conductivity). The numerical investigations show that in order to minimize the uncertainty of the generated heat flux, a fine temperature control on the thermal guard is needed. The predicted standard uncertainty is within 2% at 10W·m^?2 and within 0.5% at 100 W · m^?2.     

A study of the use of concept selection methods from inside a company

Design methods have been studied by researchers for decades. Academia considers their impact on industry to be insufficient. The objective of this research is to understand the use and impact of design methods in the context of a specific company, Volvo Car Corporation (VCC), by describing the behaviour of engineers in relation to methods, to assist in the future development of design methods and tools. We mainly concentrate on concept selection methods because of their relevance in this company. The data presented is the result of qualitative research carried out during 4 years at VCC, where the authors were located as researchers. The research shows that many methods are employed besides those with an academic name, that some in-company methods used contain improvements to methods researched by academia, that some modifications to academic methods lead to unreliable results, and that there is a lack of objectivity in method modification. For these reasons, the authors suggest further research on understanding the principles of successful and unreliable modification of concept selection methods.     

Envelope computation in the plane by approximate implicitization

Given a rational family of planar rational curves in a certain region of interest, we are interested in computing an implicit representation of the envelope. The points of the envelope correspond to the zero set of a function (which represents the envelope condition) in the parameter space combining the curve parameter and the motion parameter. We analyze the connection of this function to the implicit equation of the envelope. This connection enables us to use approximate implicitization for computing the (exact or approximate) implicit representation of the envelope. Based on these results, we formulate an algorithm for computing a piecewise algebraic approximation of low degree and illustrate its performance by several examples.