Optimization of cutting process by GA approach

The paper proposes a new optimization technique based on genetic algorithms (GA) for the determination of the cutting parameters in machining operations. In metal cutting processes, cutting conditions have an influence on reducing the production cost and time and deciding the quality of a final product. This paper presents a new methodology for continual improvement of cutting conditions with GA. It performs the following: the modification of recommended cutting conditions obtained from a machining data, learning of obtained cutting conditions using neural networks and the substitution of better cutting conditions for those learned previously by a proposed GA. Experimental results show that the proposed genetic algorithm-based procedure for solving the optimization problem is both effective and efficient, and can be integrated into an intelligent manufacturing system for solving complex machining optimization problems.     

A neural network approach to multiple-objective cutting parameter optimization based on fuzzy preference information

This paper presents a neural network approach to multiple-objective cutting parameter optimization for planning turning operations. Productivity, operation cost, and cutting quality are considered as criteria for optimizing machining operations. A feedforward neural network and a dynamic training procedure are proposed for modeling manufacturers' preferences using sampled fuzzy preferential data. Optimum cutting parameters are determined based on neural network representations of manufacturers' fuzzy preference structures.     

A review of optimization techniques in metal cutting processes

In today's rapidly changing scenario in manufacturing industries, applications of optimization techniques in metal cutting processes is essential for a manufacturing unit to respond effectively to severe competitiveness and increasing demand of quality product in the market. Optimization methods in metal cutting processes, considered to be a vital tool for continual improvement of output quality in products and processes include modelling of input–output and in-process parameters relationship and determination of optimal cutting conditions. However, determination of optimal cutting conditions through cost-effective mathematical models is a complex research endeavour, and over the years, the techniques of modelling and optimization have undergone substantial development and expansion. In this paper, the application potential of several modelling and optimization techniques in metal cutting processes, classified under several criteria, has been critically appraised, and a generic framework for parameter optimization in metal cutting processes is suggested for the benefits of selection of an appropriate approach.     

Genetic algorithm-based multi-objective optimization of cutting parameters in turning processes

Determination of optimal cutting parameters is one of the most important elements in any process planning of metal parts. This paper presents a multi-objective optimization technique, based on genetic algorithms, to optimize the cutting parameters in turning processes: cutting depth, feed and speed. Two conflicting objectives, tool life and operation time, are simultaneously optimized. The proposed model uses a microgenetic algorithm in order to obtain the non-dominated points and build the Pareto front graph. An application sample is developed and its results are analysed for several different production conditions. This paper also remarks the advantages of multi-objective optimization approach over the single-objective one.     

人工神经网络预测刀具磨损和切削力

刀具磨损和切削力预测与控制是切削加工过程中需要考虑的重要问题.本文介绍了利用人工神经网络模型预测刀具磨损和切削力的步骤并且针对产生误差的因素进行分析.首先将切削速度、切削深度、切削时间、主轴转速和不同频带的能量值通过归一化法处理,作为输入特征值,对改进的神经网络模型进行训练.然后利用训练完成的神经网络模型预测刀具磨损和切削力.结果表明:神经网络模型能够综合考虑加工过程中更多的影响因素,与经验公式结果对比,具有更高的预测精度.研究结果表明神经网络模型预测刀具磨损和切削力具有可行性和准确性,为刀具结构的优化及加工参数的选择提供了依据.     

Prediction of stable cutting depths in turning operation using soft computing methods

This article suggests soft computing methods to predict stable cutting depths in turning operations without chatter vibrations. Chatter vibrations cause poor surface finish. Therefore, preventing these vibrations is an important area of research. Predicting stable cutting depths is vital to determine the stable cutting region. In this study, a set of cutting experiments has been used and the stable cutting depths are predicted as a function of cutting, modal and tool-working material parameters. Regression analyses, artificial neural networks (ANN) decision trees and heuristic optimization models are used to develop the generalization models. The purpose of the models is to estimate stable cutting depths with minimum error. ANN produces better results compared to the other models. This study helps operators and engineers to perform turning operations in an appropriate cutting region without chatter vibrations. It also helps to take precautions against chatter.     

Tool wear monitoring and selection of optimum cutting conditions with progressive tool wear effect and input uncertainties

One of the big challenges in machining is replacing the cutting tool at the right time. Carrying on the process with a dull tool may degrade the product quality. However, it may be unnecessary to change the cutting tool if it is still capable of continuing the cutting operation. Both of these cases could increase the production cost. Therefore, an effective tool condition monitoring system may reduce production cost and increase productivity. This paper presents a neural network based sensor fusion model for a tool wear monitoring system in turning operations. A wavelet packet tree approach was used for the analysis of the acquired signals, namely cutting strains in tool holder and motor current, and the extraction of wear-sensitive features. Once a list of possible features had been extracted, the dimension of the input feature space was reduced using principal component analysis. Novel strategies, such as the robustness of the developed ANN models against uncertainty in the input data, and the integration of the monitoring information to an optimization system in order to utilize the progressive tool wear information for selecting the optimum cutting conditions, are proposed and validated in manual turning operations. The approach is simple and flexible enough for online implementation.     

Optimization free neural network approach for solving ordinary and partial differential equations

Current work introduces a fast converging neural network-based approach for solution of ordinary and partial differential equations. Proposed technique eliminates the need of time-consuming optimization procedure for training of neural network. Rather, it uses the extreme learning machine algorithm for calculating the neural network parameters so as to make it satisfy the differential equation and associated boundary conditions. Various ordinary and partial differential equations are treated using this technique, and accuracy and convergence aspects of the procedure are discussed.

     

Prediction of workpiece elastic deflections under cutting forces in turning

One of the problems faced in turning processes is the elastic deformation of the workpiece due to the cutting forces resulting in the actual depth of cut being different than the desirable one. In this paper, a cutting mechanism is described suggesting that the above problem results in an over-dimensioned part. Consequently, the problem of determining the workpiece elastic deflection is addressed from two different points of view. The first approach is based on solving the analytical equations of the elastic line, in discretized segments of the workpiece, by considering a stored modal energy formulation due to the cutting forces. Given the mechanical properties of the workpiece material, the geometry of the final part and the cutting force values, this numerical method can predict the elastic deflection. The whole approach is implemented through a Microsoft Excel^© workbook. The second approach involves the use of artificial neural networks (ANNs) in order to develop a model that can predict the dimensional deviation of the final part by correlating the cutting parameters and certain workpiece geometrical characteristics with the deviations of the depth of cut. These deviations are calculated with reference to final diameter values measured with precision micrometers or on a CMM. The verification of the numerical method and the development of the ANN model were based on data gathered from turning experiments conducted on a CNC lathe. The results support the proposed cutting mechanism. The numerical method qualitatively agrees with the experimental data while the ANN model is accurate and consistent in its predictions.     

A neuro-genetic approach for selection of process parameters in abrasive waterjet cutting considering variation in diameter of focusing nozzle

This paper presents a neuro-genetic approach proposed to suggest the process parameters for maintaining the desired depth of cut in abrasive waterjet (AWJ) cutting by considering the change in diameter of focusing nozzle, i.e. for adaptive control of AWJ cutting process. An artificial neural network (ANN) based model is developed for prediction of depth of cut by considering the diameter of focusing nozzle along with the controllable process parameters such as water pressure, abrasive flow rate, jet traverse rate. ANN model combined with genetic algorithm (GA), i.e. neuro-genetic approach, is proposed to suggest the process parameters. Further, the merits of the proposed approach is shown by comparing the results obtained with the proposed approach to the results obtained with fuzzy-genetic approach [P.S. Chakravarthy, N. Ramesh Babu, A hybrid approach for selection of optimal process parameters in abrasive water jet cutting, Proceedings of the Institution of Mechanical Engineers, Part B: J. Eng. Manuf. 214 (2000) 781–791]. Finally, the effectiveness of the proposed approach is assessed by conducting the experiments with the suggested process parameters and comparing them with the desired results.     

Transfer learning enabled convolutional neural networks for estimating health state of cutting tools

Effective Prognostics and Health Management (PHM) for cutting tools during Computerized Numerical Control (CNC) processes can significantly reduce downtime and decrease losses throughout manufacturing processes. In recent years, deep learning algorithms have demonstrated great potentials for PHM. However, the algorithms are still hindered by the challenge of the limited amount data available in practical manufacturing situations for effective algorithm training. To address this issue, in this research, a transfer learning enabled Convolutional Neural Networks (CNNs) approach is developed to predict the health state of cutting tools. With the integration of a transfer learning strategy, CNNs can effectively perform tool health state prediction based on a modest number of the relevant images of cutting tools. Quantitative benchmarks and analyses on the performance of the developed approach based on six typical CNNs models using several optimization techniques were conducted. The results indicated the suitability of the developed approach, particularly using ResNet-18, for estimating the wear width of cutting tools. Therefore, by exploiting the integrated design of CNNs and transfer learning, viable PHM strategies for cutting tools can be established to support practical CNC machining applications.     

Using FEM simulations of cutting for evaluating the performance of different johnson cook parameter sets acquired with inverse methods

Material model parameters are the primary source of error in the finite element analysis (FEM) of cutting processes. Expensive and time consuming material testing is required in order to describe the material's behavior in high temperature and high strain rate conditions during cutting. An alternative approach has been suggested in research papers; inverse analysis using cutting experiments together with FE analysis or analytical models. The latest approach is to combine an analytical model together with a material model capable of describing flow stress in terms of strain, strain rate and temperature, and using cutting experiments to acquire input parameters for inverse analysis, from which the material model parameters can be solved. In this paper, performance evaluation is done for five different sets of Johnson Cook parameters for AISI 1045, acquired with materials testing, inverse analysis with FEM, and the proposed combined inverse analysis with an analytical model and cutting experiments. The performance is evaluated by running simulations with a wide range of cutting parameters and comparing the simulated results of cutting forces and temperature to known experimental results found in literature. It was found that the proposed inverse method produces better performing model parameters than those found in literature.     

Energy-efficient multi-pass cutting parameters optimisation for aviation parts in flank milling with deep reinforcement learning

Cutting parameters play a major role in improving the energy efficiency of the manufacturing industry. As the main processing method for aviation parts, flank milling usually adopts multi-pass constant and conservative cutting parameters to prevent workpiece deformation but degrades energy efficiency. To address the issue, this paper proposes a novel multi-pass parametric optimisation based on deep reinforcement learning (DRL), allowing parameters to vary to boost energy efficiency under the changing deformation limits in each pass. Firstly, it designs a variable workpiece deformation const.raint on the principle of stiffness decreasing along the passes, based on which it constructs an energy-efficient parametric optimisation model, giving suitable decisions that respond to the varying cutting conditions. Secondly, it transforms the model into a Markov Decision Process and Soft Actor Critic is applied as the DRL agent to cope with the dynamics in multi-pass machining. Among them, an artificial neural network-enabled surrogate model is applied to approximate the real-world machining, facilitating enough explorations of DRL. Experimental results show that, compared with the conventional method, the proposed method improves 45.71% of material removal rate and 32.27% of specific cutting energy while meeting deformation tolerance, which substantiates the benefits of the energy-efficient parametric optimisation, significantly contributing to sustainable manufacturing.     

A data and knowledge-driven cutting parameter adaptive optimization method considering dynamic tool wear

Tool wear prediction is of significance to reduce energy consumption through cutting parameter optimization. However, the current studies ignore the effect of machine aging on the tool wear prediction model, and their cutting parameter optimization methods cannot cope with the dynamic change of tool wear in the machining process. Thus, a reinforcement learning-enabled integrated method of tool wear prediction and cutting parameter optimization is proposed for minimizing energy consumption and production time. Specifically, the multi-source heterogeneous data fusion-based (MHDF) tool wear prediction model considering machine aging is first proposed to obtain the tool wear of the cutting tool. Then, a Markov Decision Process is designed to model the cutting parameter optimization process, which can be utilized to determine the proper cutting parameters adapted to the dynamic change of tool wear. Finally, the proposed method is demonstrated by extensive comparative experiments, and the results show that: 1) The proposed tool wear prediction model eliminates the influence of machine aging on prediction accuracy and has better generalizability for the machining data under different machine aging conditions, and its testing accuracy reaches 96.09%. 2) The proposed optimization method can adapt to the dynamic change of tool wear and further reduce the energy consumption and production time by 6.72% and 8.60% compared to that of not considering tool wear. The computation time of the proposed method is reduced by an average of 71.80%.     

Estimation of cutting forces and surface roughness for hard turning using neural networks

Metal cutting mechanics is quite complicated and it is very difficult to develop a comprehensive model which involves all cutting parameters affecting machining variables. In this study, machining variables such as cutting forces and surface roughness are measured during turning at different cutting parameters such as approaching angle, speed, feed and depth of cut. The data obtained by experimentation is analyzed and used to construct model using neural networks. The model obtained is then tested with the experimental data and results are indicated.     

ForceNet: An offline cutting force prediction model based on neuro-physical learning approach

Accurate cutting force prediction serves as an important reference to the optimization of numerically controlled machining process. Traditional cutting force modeling via theoretical cutting mechanism hampers accurate prediction for actual machining process due to its highly suppressed modeling flexibility. On the other hand, machine learning based modeling approaches demand large amount of diversified labeled samples to achieve comparable prediction results, while collecting these samples can be tedious and costly because the cutter workpiece engagement (CWE) keeps changing during actual process. This paper presents a cutting force prediction model, named ForceNet, which incorporates elementary physical priori into structured neural networks to predict cutting force for end-milling process of complex CWE. The main idea is to use grayscale images to represent CWE geometry, providing a universal input to the ForceNet. Unlike traditional deep neural networks served as an unexplainable black box, the core of the ForceNet is constructed by the vector summation of directional primitive cutting force elements, which are approximated using elementary neural networks. Preliminary results indicate that ForceNet outperformed existing methods not only with greater prediction accuracy in unseen cutting situations, but also with less training data needed thanks to its inherent neuro-physical structure.     

Dynamic planning model for determining cutting parameters using neural networks in feature-based process planning

Although feature-based computer-aided process planning plays a vital role in automating and integrating design and manufacturing for efficient production, its off-line properties prohibit the shop floor controllers from rapidly coping with unexpected production errors. The objective of the paper is to suggest a neural network-based dynamic planning model, by which the shop floor controllers determine cutting parameters in real-time based on shop floor status. At off-line is the dynamic planning model constructed as a neural network form, and then embedded into each removal feature. The dynamic planning model will be executed by the shop floor controllers to determine the cutting parameters. A prototype system is constructed to validate whether the dynamic planning model is capable of determining dynamically and efficiently the cutting parameters for a particular set of shop operating factors. Owing to the dynamic planning model, the shop floor controller will increase flexibility and robustness by rapidly and adaptively determining the cutting parameters in unexpected errors occurring.     

Multi-objective evolutionary optimization of polynomial neural networks for modelling and prediction of explosive cutting process

In this paper, evolutionary algorithms (EAs) are deployed for multi-objective Pareto optimal design of group method of data handling (GMDH)-type neural networks which have been used for modelling an explosive cutting process using some input–output experimental data. In this way, multi-objective EAs (non-dominated sorting genetic algorithm, NSGA-II) with a new diversity-preserving mechanism are used for Pareto optimization of such GMDH-type neural networks. The important conflicting objectives of GMDH-type neural networks that are considered in this work are, namely, training error (TE), prediction error (PE), and number of neurons (N) of such neural networks. Different pairs of theses objective functions are selected for 2-objective optimization processes. Therefore, optimal Pareto fronts of such models are obtained in each case which exhibit the trade-off between the corresponding pair of conflicting objectives and, thus, provide different non-dominated optimal choices of GMDH-type neural networks models for explosive cutting process. Moreover, all the three objectives are considered in a 3-objective optimization process, which consequently leads to some more non-dominated choices of GMDH-type models representing the trade-offs among the training error, prediction error, and number of neurons (complexity of network), simultaneously. The overlay graphs of these Pareto fronts also reveal that the 3-objective results include those of the 2-objective results and, thus, provide more optimal choices for the multi-objective design of GMDH-type neural networks in terms of minimum training error, minimum prediction error, and minimum complexity.     

Selection of cutting conditions for power constrained parallel machine scheduling

Most previous studies on machining optimization focused on aspects related to machining efficiency and economics, without accounting for environmental considerations. Higher cutting speed is usually desirable to maximize machining productivity, but this requires a high power load peak. In Taiwan, electricity prices rise sharply if instantaneous power demand exceeds contract capacity. Many studies over the previous decades have examined production scheduling problems. However, most such studies focused on well-defined jobs with known processing times. In addition, traditional sequencing and scheduling models focus primarily on economic objectives and largely disregard environmental issues raised by production scheduling problems. This study investigates a parallel machine scheduling problem for a manufacturing system with a bounded power demand peak. The challenge is to simultaneously determine proper cutting conditions for various jobs and assign them to machines for processing under the condition that power consumption never exceed the electricity load limit. A two-stage heuristic approach is proposed to solve the parallel machine scheduling problem with the goal of minimizing makespan. The heuristic performance is tested by distributing 20 jobs over 3 machines with four possible cutting parameter settings.     

An innovative approach for automatic generation,verification and optimization of part programs in turning

Continuous innovation of products and optimization of manufacturing processes are of fundamental importance for preserving competitiveness. In the last decades, several approaches based on analytic models for optimization of basic machining operations such as cylindrical turning and face milling have been developed. However, the analytic approaches may not be adequate for real industrial applications, since they are based on average cutting parameters and thus they are not capable of taking into account the effect of complex geometries and instantaneous cutting conditions. In this paper, an innovative integrated system for automatic generation of optimized part programs in turning based on realistic machining simulation is proposed. The system components are described in detail and the machining simulator is validated by comparison with the results of real cutting tests. Then, the optimization approach is applied to a simple case study. The results show that the behavior of the cost function is rather complex, even for simple workpieces. Moreover, the simulator can detect unfeasible combinations of cutting parameters and thus reduce inline part program refinement and optimization. The optimal combination of cutting parameters determined by the new system was competitive with the solutions derived from tool specifications or proposed by a machining expert.