Search results

The purpose of this study is to investigate the prediction and optimization of multiple performance characteristics in the face milling process of tool steel ASSAB XW-42.

The face milling parameters (cutting speed, feed rate and axial depth of cut) and flow rate (FR) of cryogenic cooling were optimized with consideration of multiple performance characteristics, i.e. surface roughness (SR), cutting force (F_c) and metal removal rate (MRR). FR of cryogenic cooling has two levels, whereas the three face milling parameters each have three levels. Using Taguchi method, an L₁₈ mixed-orthogonal array was selected as the design of experiments. The rough estimation of the optimum face milling parameters was determined by using grey fuzzy analysis. The global optimum face milling parameters were searched by applying the backpropagation neural network-based genetic algorithm (BPNN-GA) method.

The optimum SR, cutting force (F_c) and MRR could be obtained by setting FR, cutting speed, feed rate and axial depth of cut at 0.5 l/min, 280 m/min, 90 mm/min and 0.2 mm, respectively. The experimental confirmation results showed that BPNN-based GA optimization method could accurately predict and significantly improve all of the multiple performance characteristics.

To the best of the authors’ knowledge, there were no publications available regarding multi-response optimization using the combination of grey fuzzy analysis and BPNN-based GA methods during cryogenically face milling process.

The quality and properties of Inconel 718 (IN718) from selective laser melting (SLM), a major additive manufacturing (AM) process, have been studied extensively. Among all aspects of quality, relative density (RD) is most widely investigated, and it significantly affects the mechanical properties of SLM-ed materials. This study aims to develop robust RD prediction models based on the data accumulated in literature using machining learning approaches.

By mining the literature of SLM-ed IN718, a comprehensive data set is created, which consists of the four major process parameters of laser power, scan speed, hatch spacing, layer thickness and RD data. A back propagation neural network (BPNN) model, along with its two optimized models: genetic algorithm (GA) optimized BPNN (GA-BPNN) and adaptive GA optimized BPNN (AGA-BPNN) models are created for predicting the RD of SLM-ed IN718, and their prediction performances are compared.

Overall, satisfactory prediction accuracies are obtained – the R² values of the built BPNN, GA-BPNN and AGA-BPNN models are 73.5%, 75.3% and 79.9%, respectively. This also shows that by incorporating the optimization technique, the prediction accuracy of BPNN is improved and AGA-BPNN has the highest accuracy. Moreover, SLM experiments are conducted to test the model predictability. It is found that the predictions generally agree well with the experiment data, and the order of the model prediction accuracies is consistent with that based on the literature data.

This research highlights that by mining literature data, prediction models of RD of SLM-ed IN718 can be obtained with satisfactory performance, which consider more process parameters and cover wider parameter ranges than any individual studies, in a cost-effective manner.

The purpose of this paper is to investigate prediction and optimization of multiple performance characteristics in the wire electrical discharge machining (wire-EDM) process of SKD 61 (AISI H13) tool steel.

The experimental studies were conducted under varying wire-EDM process parameters, which were arc on time, on time, open voltage, off time and servo voltage. The optimized responses were recast layer thickness (RLT), surface roughness (SR) and surface crack density (SCD). Arc on time was set at two different levels, whereas the other four parameters were set at three different levels. Based on Taguchi method, an L18 mixed-orthogonal array was selected for the experiments. Further, three methods, namely grey relational analysis (GRA), backpropagation neural network (BPNN) and genetic algorithm (GA), were applied separately. GRA was performed to obtain a rough estimation of optimum drilling parameters. The influences of drilling parameters on multiple performance characteristics were determined by using percentage contributions. BPNN architecture was determined to predict the multiple performance characteristics. GA method was then applied to determine the optimum wire-EDM parameters.

The minimum RLT, SR and SCD could be obtained by setting arc on time, on time, open voltage, off time and servo voltage at 2 ms, 3 ms, 90 volt, 10 ms and 38 volt, respectively. The experimental confirmation results showed that BPNN-based GA optimization method could accurately predict and significantly improve all of the responses.

There were no publications regarding multi-response optimization using a combination of GRA and BPNN-based GA methods during wire-EDM process available.

Three-axis accelerometers play a vital role in monitoring the vibrations in aircraft machinery, especially in variable flight temperature environments. The sensitivity of a three-axis accelerometer under different temperature conditions needs to be calibrated before the flight test. Hence, the authors investigated the efficiency and sensitivity calibration of three-axis accelerometers under different conditions. This paper aims to propose the novel calibration algorithm for the three-axis accelerometers or the similar accelerometers.

The authors propose a hybrid genetic algorithm–particle swarm optimisation–back-propagation neural network (GA–PSO–BPNN) algorithm. This method has high global search ability, fast convergence speed and strong non-linear fitting capability; it follows the rules of natural selection and survival of the fittest. The authors describe the experimental setup for the calibration of the three-axis accelerometer using a three-comprehensive electrodynamic vibration test box, which provides different temperatures. Furthermore, to evaluate the performance of the hybrid GA–PSO–BPNN algorithm for sensitivity calibration, the authors performed a detailed comparative experimental analysis of the BPNN, GA–BPNN, PSO–BPNN and GA–PSO–BPNN algorithms under different temperatures (−55, 0 , 25 and 70 °C).

It has been showed that the prediction error of three-axis accelerometer under the hybrid GA–PSO–BPNN algorithm is the least (approximately ±0.1), which proved that the proposed GA–PSO–BPNN algorithm performed well on the sensitivity calibration of the three-axis accelerometer under different temperatures conditions.

The designed GA–PSO–BPNN algorithm with high global search ability, fast convergence speed and strong non-linear fitting capability has been proposed to decrease the sensitivity calibration error of three-axis accelerometer, and the hybrid algorithm could reach the global optimal solution rapidly and accurately.

The amount, type and addition conditions of additives of lubricants should be continuously adjusted to obtain appealing performance. To obtain the optimal pretreatment parameters and reduce the cost of time-consuming experiments, the purpose of this paper is to establish an optimal back propagation neural network (BPNN) model combined with genetic algorithm (GA) in this work.

Using trimethylolpropane trioleate as the base oil and three types of phosphorus compounds as additives, 25 sets of lubricant formulas were designed regarding lubricant performances of average friction coefficient, average spot diameter, disk wear volume and extreme pressure. The data set was used for training and learning of BPNN and then combined with GA to optimize BPNN with continuously optimization by adjusting various parameters.

Comparing prediction data of BPNN with actual test data, correlation coefficients were above 90%, indicating that the model could accurately predict the performance of lubricants. When combined with GA, all performance errors were less than 5%, indicating that BPNN could be optimized by GA to obtain an accurate combined model for prediction of lubricant performance. The best additive formula with excellent performances was obtained from the BPNN–GA model.

This work developed a new method to study lubricant compounding. The combined model was expected to provide a theoretical basis and guidance for the compounding optimization of lubricant additives with high efficiency and low cost and to expand the scope to practical applications.

The peer review history for this article is available at: https://publons.com/publon/10.1108/ILT-05-2020-0165/

The purpose of this paper is to develop a transform method and a deep learning model to identify the inner surface shape based on the measurement temperature at the outer boundary of the pipe.

The training process is assisted by the finite element method (FEM) simulation which solves the direct problem for the data preparation. To avoid re-meshing the domain when the inner surface shape varies, a new transform method is proposed to transform the shape identification problem into the effective thermal conductivity identification problem. The deep learning model is established to set up the relationship between the measurement temperature and the effective thermal conductivity. Then the unknown geometry shape is acquired by the mapping between the inner shape and the effective thermal conductivity through the inverse transform method.

The new method is successfully applied to identify the internal boundary of a pipe with eccentric circle, ellipse and nephroid inner geometries. The results show that as the measurement points increased and the measurement error decreased, the results became more accurate. The position of the measurement point and mesh density of the FEM model have less effect on the results.

The deep learning model and the transform method are developed to identify the pipe inner surface shape. There is no need to re-mesh the domain during the computation progress. The results show that the proposed method is a fast and an accurate tool for identifying the pipe inner surface.

The purpose of this paper is to provide reference for researchers by reviewing the research advances and trend of agricultural product price forecasting methods in recent years.

This paper reviews the main research methods and their application of forecasting of agricultural product prices, summarizes the application examples of common forecasting methods, and prospects the future research directions.

1) It is the trend to use hybrid models to predict agricultural products prices in the future research; 2) the application of the prediction model based on price influencing factors should be further expanded in the future research; 3) the performance of the model should be evaluated based on DS rather than just error-based metrics in the future research; 4) seasonal adjustment models can be applied to the difficult seasonal forecasting tasks in the agriculture product prices in the future research; 5) hybrid optimization algorithm can be used to improve the prediction performance of the model in the future research.

The methods from this paper can provide reference for researchers, and the research trends proposed at the end of this paper can provide solutions or new research directions for relevant researchers.

This study presents finite element (FE) and generalized regression neural network (GRNN) approaches for modeling multiple crack growth problems and predicting crack-growth directions under the influence of multiple crack parameters.

To determine the crack-growth direction in aluminum specimens, multiple crack parameters representing some degree of crack propagation complexity, including crack length, inclination angle, offset and distance, were examined. FE method models were developed for multiple crack growth simulations. To capture the complex relationships among multiple crack-growth variables, GRNN models were developed as nonlinear regression models. Six input variables and one output variable comprising 65 training and 20 test datasets were established.

The FE model could conveniently simulate the crack-growth directions. However, several multiple crack parameters could affect the simulation accuracy. The GRNN offers a reliable method for modeling the growth of multiple cracks. Using 76% of the total dataset, the NN model attained an R2 value of 0.985.

The models are presented for static multiple crack growth problems. No material anisotropy is observed.

In practical crack-growth analyses, the NN approach provides significant benefits and savings.

The proposed GRNN model is simple to develop and accurate. Its performance was superior to that of other NN models. This model is also suitable for modeling multiple crack growths with arbitrary geometries. The proposed GRNN model demonstrates its prediction capability with a simpler learning process, thus producing efficient multiple crack growth predictions and assessments.

Here, the authors use step angles, stage ratios, feed rates and spindle speeds as predictors to develop a Gaussian process regression for predicting thrust force during composite laminates drilling with step drills.

Use of machine learning methods could benefit machining process optimizations. Accurate, stable and robust performance is one of major criteria in choosing among different models. For industrial applications, it is also important to consider model applicability, ease of implementations and cost effectiveness.

This model turns out to be simple, accurate and stable, which helps fast estimates of thrust force. Through combining the Taguchi method's optimization results and the Gaussian process regression, more data could be expected to be extracted through fewer experiments.

Through combining the Taguchi method's optimization results and the Gaussian process regression, more data could be expected to be extracted through fewer experiments.

This paper aims to discuss the utilization of artificial neural networks (ANNs) and multiple regression method for estimating surface roughness in milling medium density fiberboard (MDF) material with a parallel robot.

In ANN modeling, performance parameters such as root mean square error, mean error percentage, mean square error and correlation coefficients (R²) for the experimental data were determined based on conjugate gradient back propagation, Levenberg–Marquardt (LM), resilient back propagation, scaled conjugate gradient and quasi-Newton back propagation feed forward back propagation training algorithm with logistic transfer function.

In the ANN architecture established for the surface roughness (Ra), three neurons [cutting speed (V), feed rate (f) and depth of cut (a)] were contained in the input layer, five neurons were included in its hidden layer and one neuron was contained in the output layer (3-5-1).Trials showed that LM learning algorithm was the best learning algorithm for the surface roughness. The ANN model obtained with the LM learning algorithm yielded estimation training values R² (97.5 per cent) and testing values R² (99 per cent). The R² for multiple regressions was obtained as 96.1 per cent.

The result of the surface roughness estimation model showed that the equation obtained from the multiple regressions with quadratic model had an acceptable estimation capacity. The ANN model showed a more dependable estimation when compared with the multiple regression models. Hereby, these models can be used to effectively control the milling process to reach a satisfactory surface quality.