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

Bustillo, Andrés; Díez-Pastor, José-Francisco; Quintana, Guillem; García‐Osorio, César

doi:10.1007/s00170-011-3300-z

Cited by 36 publications

(33 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These cutting conditions were selected following the cutting tool provider specifications and the machine-tool capabilities. Although cutting conditions are regularly changed in experiments that focus on roughness quality prediction, in order to generate extensive datasets (Bustillo et al 2011a), the number of different cutting conditions is strongly reduced (Prasad et al 2011) or just limited to one cutting condition, as in this research (Rivero et al 2008), in the case of including tool wear. The main drive power was established in accordance with the online readings made using a Mori Seiki NMV 5000 during machining.…”

Section: Methodsmentioning

confidence: 99%

“…The high accuracy of ensemble predictions has been demonstrated in many milling processes: Bustillo et al (2011a) proposed the use of ensembles to predict surface roughness in ball-end milling operations. Maudes et al (2017) used random forest (RF) ensembles for the prediction of dimensional parameters in laser micromanufacturing of stents, and Ferreiro and Sierra (2012) used different kinds of ensembles for burr detection in a drydrilling process on aluminum Al 7075-T6.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Artificial intelligence for automatic prediction of required surface roughness by monitoring wear on face mill teeth

2017

View full text Add to dashboard Cite

Nowadays, face milling is one of the most widely used machining processes for the generation of flat surfaces. Following international standards, the quality of a machined surface is measured in terms of surface roughness, Ra, a parameter that will decrease with increased tool wear. So, cutting inserts of the milling tool have to be changed before a given surface quality threshold is exceeded. The use of artificial intelligence methods is suggested in this paper for real-time prediction of surface roughness deviations, depending on the main drive power, and taking tool wear, V B into account. This method ensures comprehensive use of the potential of modern CNC machines that are able to monitor the main drive power, N , in real-time. It can likewise estimate the three parameters -maximum tool wear, machining time, and cutting power-that are required to generate a given surface roughness, thereby making the most efficient use of the cutting tool. A series of artificial intelligence methods are tested: random forest (RF), standard Multilayer perceptrons (MLP), Regression Trees, and radial-based functions. Random forest was shown to have the highest model accuracy, followed by regression trees, displaying higher accuracy than the standard MLP and the radial-basis function. Moreover, RF techniques are easily tuned and generate visual information for direct use by the process engineer, such as the linear relationships between process parameters and roughness, and thresholds for avoiding rapid tool wear. All of this information can be directly extracted from the tree structure or by drawing 3D charts plotting two process inputs and the predicted roughness depending on workshop requirements. Keywords

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Artificial intelligence for automatic prediction of required surface roughness by monitoring wear on face mill teeth

2017

View full text Add to dashboard Cite

show abstract

“…Artificial Neural Networks (ANN) are the most widely used AI technique Grzenda et al 2012;Díez-Pastor et al 2012;Benardos and Vosniakos 2002;Samanta et al 2008;Correa et al 2008), although other techniques like neuro-fuzzy inference systems (Samanta et al 2008), Bayesian networks (Correa et al 2008), genetic algorithms (Brezocnik et al 2004), swarm optimization techniques (Zainal et al 2016), and support vector machines (Prakasvudhisarn et al 2008) have also been tested for the same industrial task. Unfortunately, ANN models are highly dependent on the parameters of the neural networks (Bustillo et al 2011) and the process of fine-tuning these parameters is a highly time-consuming task that frequently requires expertise for good results. Moreover, studies on surface-roughness prediction in face milling are scarce, compared with the large amount of studies focused on this prediction task for other milling operations, as emphasized in reviews of this domain (Chandrasekaran et al 2010;Abellan-Nebot 2010).…”

Section: Introductionmentioning

confidence: 99%

Semi-supervised roughness prediction with partly unlabeled vibration data streams

Grzenda

Bustillo

2018

J Intell Manuf

Self Cite

View full text Add to dashboard Cite

Experimental data sets that include tool settings, tool and machine-tool behavior, and surface roughness data for milling processes are usually of limited size, due mainly to the high costs of machining tests. This fact restricts the application of machine-learning techniques for surface roughness prediction in industrial settings. The primary objective of this work is to investigate the way data streams that are missing product features (i.e. unlabeled data streams) can contribute to the development of prediction models. The investigation is followed by a proposal for a semi-supervised approach to the development of roughness prediction models that can use partly unlabeled data to improve the accuracy of roughness prediction. Following this strategy, records collected during the milling process, which miss roughness measurements, but contain vibration data are used to increase the accuracy of the prediction models. The method proposed in this work is based on the selective use of such unlabelled instances, collected at tool settings that are not represented in the labeled data. This strategy, when applied properly, yields both extended training data sets and higher accuracy in the roughness prediction models that are derived from them. The scale of accuracy improvement and its statistical significance are shown in the study case of high-torque face milling of F114 steel. The semi-supervised approach proposed in this work has been used in combination with supervised k Nearest Neighbours and random forest techniques. Furthermore, the study of both continuous and discretized roughness prediction, showed higher gains in accuracy in the second.

show abstract

“…There are four broad lines of research into surface roughness prediction (Benardos and Vosniakos 2003) based on: (1) machining theory (Montgomery and Altintas 1991;Ismail et al 1993;Quintana et al 2010;Arizmendi et al 2009); (2) experimental investigations (Beggan et al 1999;Vivancos et al 2005); (3) designed experiments (Choudhury and Bartarya 2003) and (4) artificial intelligence (Correa et al 2008;Benardos and Vosniakos 2002;Brezocnik and Kovacic 2003;Dhokia et al 2008;Quintana et al 2009;Lo 2003;Ho et al 2009;Iqbal et al 2007;Samanta et al 2008;Correa et al 2009;Brezocnik et al 2004;Prakasvudhisarn et al 2009;Bustillo et al 2011b). Nevertheless, the classifications in much of the literature, where no one single methodology is followed, are not always straightforward.…”

Section: Introductionmentioning

confidence: 99%

“…Different alternative approaches to the industrial task of roughness prediction in milling operations have been tested with good results: neuro-fuzzy inference system (Lo 2003;Ho et al 2009;Iqbal et al 2007;Samanta et al 2008), Bayesian networks (Correa et al 2008(Correa et al , 2009, genetic algorithms (Brezocnik et al 2004;Brezocnik and Kovacic 2003) and support vector machines (Prakasvudhisarn et al 2009). However, the neural networks approach (Quintana et al 2009;Choudhury and Bartarya 2003;Benardos and Vosniakos 2002;Correa et al 2009;Bustillo et al 2011b) is the most widely used solution; Its most common configuration is a multilayer perceptron (MLP) with a single hidden layer (Groove 2006). Unfortunately, the results obtained are highly dependent on the parameters of the neural networks (Bustillo et al 2011b).…”

Section: Introductionmentioning

confidence: 99%

Boosting Projections to improve surface roughness prediction in high-torque milling operations

et al. 2012

Self Cite

View full text Add to dashboard Cite

Industrial solutions for surface roughness prediction are in great demand, especially in high-torque milling operations, owing to the exponential expansion of wind power energy generation over the past decade. In this paper, we use Boosting Projections to predict surface roughness in high-torque, high-power face milling operations. A data set is generated from experiments performed under industrial conditions, using a milling machine with a high working volume, to train and validate the new algorithm. The experimental data comprise a very extensive set of parameters that influence surface roughness: cutting tool properties, machining parameters and cutting phenomena. The proposed method is based on non-linear boosting projections (although it uses linear projections to speed up the training process). To the best of our knowledge this is the first time it has been used in an industrial context. It demonstrates a higher prediction accuracy when compared with single multilayer perceptrons, decision trees and classical ensemble methods.

show abstract

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

Cited by 36 publications

References 35 publications

Artificial intelligence for automatic prediction of required surface roughness by monitoring wear on face mill teeth

Artificial intelligence for automatic prediction of required surface roughness by monitoring wear on face mill teeth

Semi-supervised roughness prediction with partly unlabeled vibration data streams

Boosting Projections to improve surface roughness prediction in high-torque milling operations

Contact Info

Product

Resources

About