Machine Learning-Driven Regression Modelling and RSM-Based Experimental Analysis

In this paper, an inverse heat conduction method is applied to estimate the amount of the energy (Fc) transferred to the workpiece during electric discharge machining (EDM) process. Embedded thermocouples which were connected to a four channel data logger were utilized to measure the temperature of a specific location on a rectangular workpiece during the EDM process. After temperature measurements were done, the 2-D heat conduction model of the workpiece and the Levenberg-Marquardt (LM) scheme were used to determine the energy transferred to the workpiece. This inverse procedure facilitates the determination of the heat energy at discharge-workpiece interface in EDM processes, which yet is a challenge for existing numerical models. The obtained results showed that the energy transferred to the workpiece varies with the discharge current and pulse duration from 5 % up to 45 %, which shows that the value of Fc is a function of discharge current and pulse duration and that the fixed value of energy assumed in majority of the previous researches is not in accordance with real EDM conditions. Furthermore, the effects of machining parameters such as discharge current and pulse duration on Fc were studied. It was evident that the Fc has a direct but non-linear relationship with both discharge current and pulse duration, while discharge current has a higher impact on Fc.

Examining the electrical discharge machining (EDM) process is challenging in manufacturing technology due to the complexity of the physical events that take place in the gaps between electrodes. In this study, we examined the EDM process in detail and developed multiple machine learning (ML) models to describe the relationship between the EDM independent (process parameters) and dependent (responses) variables. The collected experimental data was used to train the machine learning models. According to the results, the GPR model outperformed other ML models across different materials, with average RMSE values of 0.9234 and 3.0216 for the material removal rate (MRR) and surface roughness (Sa), respectively, highlighting the effectiveness of ML tools at modeling complex machining processes, such as EDM. In addition, as a practical implication, this study opens the door to employing the developed ML models to predict the EDM process performance.

Modelling Electrical Discharge Machining (EDM) processes based on physical laws has been the subject of much debate in the literature. Therefore empirical regression modelling is frequently used in practice. In this study, a polynomial regression model was developed to correlate the electrode wear with four EDM process parameters (current, pulse on-time, pulse off-time and capacitance) while machining the cobalt-bonded tungsten carbide ceramic. As many factor interactions were found significant, lambda plot was used to find an appropriate response transformation that can simplify the model and improve its interpretability. This was attained when adopting the inverse transformation as only three parameters (current, pulse on-time and pulse off-time) were significant with an adjusted-R2 of 0.939. Due to the possibility of rendering illogical predicted values when detransforming the results of the fitted model, an alternative Generalized Linear Model (GLM) with gamma distribution and a reciprocal link function was developed. Being associated with shorter mean response confidence intervals, the GLM was more reliable for estimating and predicting the response. This paper is the first to report the use of GLM in the context of modelling EDM processes.

Surface treatment by electric discharge machining of Ti–6Al–4V alloy for potential application in orthopaedics

This study presents a machine learning (ML)based approach to predict the performance parameters (Gain and S11) of a Y-shaped microstrip patch antenna (MPA). The dataset used for training and testing the models was generated using HFSS simulations for the antenna design, which includes six key geometrical parameters: lu2, scale1, scale2, scale3, ud1, and uy1. These parameters were systematically varied to evaluate their impact on the antenna’s performance over a frequency range of 3.25 GHz to 4.25 GHz, with resonance achieved at 3.6 GHz. The substrate material, Rogers $\mathbf{5 8 8 0}$, has a dielectric constant of 2.33 and a loss tangent of 0.0023. Several machine learning regression models, including Linear Regression, Random Forest, Support Vector Regression (SVR), Neural Networks, and ensemble methods such as XGBoost, LightGBM, and CatBoost, were applied to predict the Gain and S11 parameters. The predictive performance of these models was assessed using Mean Absolute Error (MAE) and Mean Squared Error (MSE). The results demonstrated that ensemble models, particularly CatBoost, outperform traditional methods, achieving the lowest error in Gain and S11 predictions. This work showcases the potential of machine learning techniques as efficient surrogates for computationally expensive EM simulations, thus enhancing the efficiency of antenna design workflows in modern wireless communication systems.

Globally, wind energy if properly harnessed, could serve as a source of energy generation in Africa. This study compared the performance of two Machine Learning (ML) algorithms (Linear regression and Random Forest) in predicting wind speed in five major cities in Africa (Yaoundé, Pretoria, Nairobi, Cairo and Abuja). Wind data were collected between January 1, 2000, and December 31, 2022, using the Solar Radiation Data Archive. The data preprocessing was carried out with 80% of the data used for training and 20% for validation. The performance of these ML algorithms was evaluated using Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and coefficient of determination (R2). The result shows that Nairobi (3.814795 m/s) closely followed by Cairo (3.606453 m/s) has the highest mean wind speed while Yaoundé (1.090512 m/s) has the lowest. Based on the performance metrics used, the two Machine Learning algorithms were competitive. Still, the Linear Regression (LR) algorithm outperformed the Random Forest Algorithm in predicting wind speed in all the selected major African cities. In Yaoundé (RMSE = 0.3892, MAE= 0.3001, MAPE =0.5030), Pretoria (RMSE=1.2339, MAE=0.9480, MAPE=0.7450) Nairobi (RMSE= 0.4223, MAE =0.6499, MAPE =0.1872), Nairobi (RMSE=0.6499, MAE=0.5171, MAPE =0.1872), Cairo (RMSE =1.0909, MAE =0.8544, MAPE =0.3541) and Abuja (RMSE = 0.70245, MAE =0.5441, MAPE= 0.4515) the Linear regression algorithms was found to outperformed Random Forest Regression. Therefore, the Linear regression algorithm is more reliable in predicting wind speed compared with the Random Forest regression.

A model of tool wear in electrical discharge machining process based on electromagnetic theory

A review on various methods to improve process capabilities of electrical discharge machining process

This paper presents the experimental investigation of the machining characteristics of AISI D2 tool steel on the electrical discharge machining (EDM) process by replacing the copper electrode tool with carbon nanotube (CNT) fused electrode, and checking for improvement in machinability factors like material removal rate (MRR), electrode wear rate (EWR) and surface finish (SR). The problems of high electrode wear in EDM machining, such as poor dimensional and shape accuracy of the machined part, need to be addressed. The work material is spark machined with pure copper and CNT-infused copper electrodes by varying parameters on an EDM. The results of the machining characteristics with and without the CNT are compared and analysed using Taguchi analysis. This study investigates the optimisation of EDM parameters for AISI D2 tool steel using TOPSIS method. Analysis of variance (ANOVA) and F-test are used to check the validity of regression model and determine the significant parameter affecting the surface roughness.

Although it is time-consuming to perform multi-pass electrical discharge machining (EDM) processes, EDM is exceedingly feasible for processing the partially unmelt profile of Powder-Bed-Fusion additive manufacture mold steel. This study is based on the principle of removing three times the thickness of the recast layer and reducing the EDM processes depth. And then, the surface quality and the processing time with EDM passes are decreased by adjusting the EDM passes suggested by the machine builder. Secondly, reduction of four more passes and the predetermined depth of processing so that the total processing time can be reduced further. Finally, the efficiency is increased by more than 77% compared to machine builder’s processes, and the final surface roughness reaches the specification of 0.17 μm Ra. The HAZ and recast layer thickness are also reduced drastically and distributed uniformly. Therefore, a new version of the EDM process planning suitable for AM CORRAX mold surface treatment is proposed, and its processing efficiency and surface integrity are investigated.

Electrical discharge machining (EDM) is an unconventional machining process used for machining of hard-to-cut materials. Both EDM and micro-EDM processes are extensively used for producing dies and molds, complex cavities, and 3D structures. In recent years, researchers have intensively focused on improving the performance of both micro-EDM and EDM processes. This paper reviews the research work carried out by the researchers on vibration-assisted EDM, micro-EDM, and wire EDM. The consolidated review of this research work enables better understanding of the vibration-assisted EDM process. This study also discusses the influence of vibration parameters such as vibration frequency and amplitude on the material removal rate (MRR), electrode wear rate (EWR), and surface roughness (SR). The important issues and research gaps in the respective area of research are also presented in this paper.

Validation of different tungsten carbide-cobalt grades as tool electrode material for sinking EDM

The prediction of wind energy generation is important to enhance the performance and dependability of renewable energy systems due to the rising demand for wind-generated electricity and advancements in wind energy technology competitiveness. This study leverages advanced machine learning (ML) and some other statistical and deep learning based time series forecasting models to enhance the accuracy of wind energy predictions. This comprehensive analysis includes nine ML models—Linear Regression, Random Forests (RF), Gradient Boosting Machines (GBM), Support Vector Machines (SVM), K-Nearest Neighbors (KNN), AdaBoost, XGBoost, Support Vector Regression (SVR), and Neural Networks—as well as Four time-series forecasting models—ARIMA, Temporal Convolutional Networks (TCNs), Long Short-Term Memory (LSTM) networks and GRU. Each ML model underwent rigorous cross-validation to ensure optimal performance. The assessment criteria utilized here comprised the Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the R² Score. It was found that among the nine ML models, Random Forests, GBM and KNN consistently provided superior accuracy and robustness, making them the top choices for wind energy prediction whereas the performance of linear regression, SVM and SVR were very poor for the considered dataset. From the experiment, Random Forest, GBM, and KNN showed the best performance with low MSE values of 0.77, 1.95, and 1.51 respectively, while other models had MSEs above 7.5, with AdaBoost reaching 30. Their RMSEs (0.88, 1.40, 1.23) and MAEs (0.093, 0.73, 0.10) also indicate strong predictive accuracy compared to the rest.In this paper, time series forecasting, TCNs, LSTM and GRU networks showed strong capabilities in capturing temporal dependencies and trends within the wind energy data. Visualization techniques were employed to compare model performances comprehensively, providing clear insights into their predictive power. Therefore, this present study offers a robust framework for researchers and practitioners aiming to leverage machine learning and time series forecasting in the realm of renewable energy prediction.

This paper reports the development of an intelligent model for the electric discharge machining (EDM) process using finite-element method (FEM) and artificial neural network (ANN). A two-dimensional axisymmetric thermal (FEM) model of single-spark EDM process has been developed based on more realistic assumptions such as Gaussian distribution of heat flux, time- and energy-dependent spark radius, etc. to predict the shape of crater cavity, material removal rate, and tool wear rate. The model is validated using the reported analytical and experimental results. A neural-network-based process model is proposed to establish relation between input process conditions (discharge power, spark on time, and duty factor) and the process responses (crater geometry, material removal rate, and tool wear rate) for various work—tool work materials. The ANN model was trained, tested, and tuned using the data generated from the numerical (FEM) simulations. The ANN model was found to accurately predict EDM process responses for chosen process conditions. It can be used for the selection of optimum process conditions for EDM process.

Electrical Discharge Machining (EDM) is a well-known machining technique since more than fifty years. Its principle is to use the eroding effect on the electrodes of successive electric spark discharges created in a dielectric liquid. EDM is nowadays widely-used in a large number of industrial areas. Nevertheless, few studies have been done on the discharge itself and on the plasma created during this process. Further improvements of EDM, especially for micro-machining, require a better control and understanding of the discharge and of its interaction with the electrodes. In this work, the different phases of the EDM process and the properties of the EDM plasma have been systematically investigated with electrical measurements, with imaging and with time- and spatially-resolved optical emission spectroscopy. The pre-breakdown phase in water is characterized by the generation of numerous small hydrogen bubbles, created by electrolysis. Since streamers propagate more easily in a gaseous medium, these bubbles can facilitate the breakdown process. In oil, no bubbles are observed. Therefore, the breakdown mechanism in oil could be rather enhanced by particles present in the electrode gap. Fast pulses of current and light are simultaneously measured during the pre-breakdown. These pulses are characteristic of the propagation of streamers in the dielectric liquid. The pre-breakdown duration is not constant for given discharge parameters, but distributed following a Weibull distribution. This shows that the breakdown is of stochastic nature. After the breakdown, the plasma develops very rapidly ( 2·1018 cm-3 during the first microsecond). Then it decreases with time, remaining nevertheless above 1016 cm-3 after 50 μs. During the whole discharge, the density is slightly higher in the plasma center. The EDM plasma has such a high density because it is formed from a liquid, and because it is constantly submitted to the pressure imposed by the surrounding liquid. This extreme density produces spectra with strongly-broadened spectral lines, especially the Hα line, and with an important continuum. During the first microsecond when the density is at its maximum, spectral lines are so broadened that they are all merged into a continuum. The low temperature and the high density of the EDM plasma make it weakly non-ideal. Its typical coupling parameter Γ is indeed around 0.3, reaching 0.45 during the first microsecond. In this plasma, the Coulomb interactions between the charged particles are thus of the same order as the mean thermal energy of the particles, which produces coupling phenomena. Spectroscopic results confirm the non-ideality of the EDM plasma. The strong broadening and shift of the Hα line and its asymmetric shape and complex structure, the absence of the Hβ line, and the merging of spectral lines are typical of nonideal plasmas. The EDM plasma has thus extreme physical properties, and the physics involved is astonishingly complex.