Material removal rate optimization with bayesian optimized differential evolution based on deep learning in robotic polishing

Electrochemical machining is a promising non-traditional manufacturing process to make high-quality parts. The benefits of minimal thermally and mechanically induced stresses, free of burr, and a low surface roughness are appealing for industry and research institutes. However, the combined chemical reaction, electric field, fluid mechanics, and material properties involve a significant number of independent parameters which are difficult to analyze in order to draw comprehensive conclusions. To our current knowledge, process responses such as the material removal rate, optimal feed rate, and cutting profile cannot be represented accurately by analytical solutions. In recent years, deep learning has had tremendous success in analyzing sophisticated systems. The improved computation efficiency and reduced size of the training dataset required for deep learning have enabled various prediction models in the manufacturing industry. In this paper, a new approach is developed using the deep convolutional network with the Bayesian optimization algorithm to predict the diameters of the drilled hole from an electrochemical machining process. The Keras application programming interface (API) was used to build the deep convolutional network; the feed rate, pulse-on time, and voltage were used as input parameters to provide a fair comparison with a neural network from previous research. Random dropout layers were added to prevent overfitting of the network. Instead of tuning the network parameter by trial and error, the Bayesian parameter optimization algorithm was implemented to find the optimal set of parameters of the deep convolutional network that yields the minimum mean square error. The proposed algorithm was compared with a previously developed neural network with partially embedded physical knowledge. Improved training speed and accuracy were observed in comparison with the traditional neural network. The prediction model using the proposed deep learning algorithm demonstrated better prediction accuracy and provided a more systematic way to select the hyperparameter for the deep convolutional network.

In this work, a new aluminum gate chemical mechanical planarization (CMP) model for material removal rate (MRR) is proposed in high-k metal gate (HKMG) process. Using the basic principles of steady-state oxidation reaction and mechanical abrasion mechanism, the combinational interaction of chemical and mechanical coupled effects on MRR was systematically described by process parameters, pad properties and concentration of oxidizer, and then balanced to construct an overall polishing rate. Because of the great significance of the slurry pH on MRR in CMP process, the effects of surface forces, including the influences of van der Waals (vdW) and double-layer (DL) forces simultaneously acted between the wafer and the particles were investigated. Meanwhile, the influence of particle sizes, abrasive loadings and zeta potentials of the wafer and particles on MRR was also analyzed. It is found that the attractive vdW forces strengthen the MRR, while the DL forces, calculated to be repulsive, lower the MRR. The magnitude of surface forces increases with a smaller particle size compared with the pad-particle force. When zeta potentials of the wafer and particles are considered as a function of slurry pH, the experimental trends for the MRR with slurry pH, applied pressure and abrasive loading were predicted well by the present model. Therefore, the governing equation of aluminum removal reveals some insights into the HKMG CMP process and can be utilized for optimizing and controlling the polishing rate of aluminum gate structures and performing sensitivity analyses of operating parameters.

Material removal rate model for chemical–mechanical polishing of single-crystal SiC substrates using agglomerated diamond abrasive

Maximization of material removal rate may yield impractical results if the cost for generation of removal rate is not included in the optimization. Presently, there are many studies which have applied different linear, nonlinear, and metaheuristics optimization techniques to maximize material removal rate but none of the approaches have the capability of self‐adaptation. In none of these studies impact of cost on removal rate was considered. That is why, in the present study, the Group Method of Data Handling which is a technique to develop polynomial neural network (PNN) models was utilized for the first time to maximize material removal rate with the help of the on‐ and off‐stage current as the design variables constrained by the operating costs. The PNN architecture is widely used in many fields of technology and science. However, application of this architecture is scarce in case of optimization of material removal rate. According to the results the accuracy level and time of convergence in case of PNN was found to be better compared to that from artificial neural network and linear models both of which were utilized for estimation of the output variable.

In this paper, a comparative experimental analysis of die-sinking electric discharge machining (EDM) to two most exhaustively used aluminum metal matrix composites (AMMCs) has been performed using Copper and Tungsten as tool electrodes. AMMCs containing silicon carbide (SiC) and alumina (Al2O[Formula: see text] as reinforcement (10[Formula: see text]wt%) were fabricated by stir casting method. The Box–Behnken Design (BBD) approach of response surface methodology was used to develop experimental models for material removal rate (MRR) and surface roughness (SR). Effect of input parameters such as current ([Formula: see text]–12[Formula: see text]A), gap voltage ([Formula: see text][Formula: see text]V), pulse-on time ([Formula: see text]s), and duty factor ([Formula: see text]–6) on the output responses has been investigated with response surface plots. Effectiveness of design of experiment (DoE) and evolutionary algorithm-based multi-objective optimization (MOO) technique have been compared to find the best feasible optimal solution. ANOVA analysis reveals that for alumina reinforced AMMC interaction between [Formula: see text] has significant effect on both MRR and SR using Cu electrode. But for tungsten, electrode interaction between [Formula: see text], [Formula: see text], and [Formula: see text] have major role on MRR whereas SR is mostly influenced by interaction between [Formula: see text] and [Formula: see text]. The parametric analysis reveals that an increase of current from 4[Formula: see text]A to 12[Formula: see text]A at a higher pulse-ontime increases the MRR more significantly, and higher MRR occurs in cases of alumina-reinforced AMMC. Increase of pulse-ontime at low current (4[Formula: see text]A) reduces the MRR in AMMC/Al2O3. Good surface finish can be obtained by combining high voltage (60[Formula: see text]V) with either small current (4[Formula: see text]A) or small duty factor (4) for both AMMCs. Both DoE and metaheuristic-based MOO technique reveals that copper electrode should be preferred for die-sinking EDM of AMMC/SiC. Metaheuristic approach should be preferred for optimization of die-sinking EDM of AMMCs using different electrodes because it requires low current for effective machining of different AMMCs.

Modeling of material removal rate and surface roughness in finishing of bevel gears by electrochemical honing process

Stir casting process is well known for reinforcement of ceramic particle in aluminium. Due to improved properties of Al6061-T6/SiC particulate metal matrix composites have attracted by industry as well as received wider attention of material expertise. The purpose of this study is to evaluate the turning process parameters through algebraic matrix which is unique approach. Also studied effects of spindle speed, feed, SiC silicon carbide-3% and 6%, depth of cut and nose radius on modifier element for instance surface roughness Ra and material removal rate MRR in both dry and wet condition. In addition to optimize the turning process parameters are on Al6061-T6/SiC 3% and 6% with a coated tungsten tool. The experimental runs are designed using the taguchi based multiple factorial design DOE and their outcomes are analyzed using Analysis of Variance ANOVA. Mathematical models are established using algebraic matrix to represent the relationship among turning process parameters as independent variables, surface roughness and material removal rate as dependent variables. For every experimental run, a same cutting insert was used to encourage accurate reporting of the surface roughness and material removal rate. The statistical variations revealed that the main effect of spindle speed, feed, SiC silicon carbide-3% and 6%, depth of cut and nose radius are influenced the surface roughness and material removal rate. Moreover, Built-up-edge BUE formation was observed at every combinations of machining parameters such as spindle speed, feed, SiC silicon carbide-3% and 6%, depth of cut and nose radius which affected the surface quality negatively. The proximity of predicted results and experimental results provide proof that the algebraic matrix – DOE method has successfully established the predictive models. It is strongly suggested SiC silicon carbide 3% and 6% are successfully reinforced in Al6061 as well as optimize turning process parameters in wet condition than dry condition. Mathematical models for surface roughness and material removal rate are found to be statistically significant in wet condition.

This study employs a multi-objective particle swarm optimization (MOPSO) algorithm to address the dual-objective challenge in the robotic polishing of Ti-6Al-4V. The aim is to determine optimal parameters that minimize surface roughness while maximizing the material removal rate (MRR), thereby improving both surface quality and processing efficiency. First, a material removal depth model for end-face polishing is established based on Preston's equation and theoretical analysis, from which the MRR model is derived. Subsequently, orthogonal experiments are conducted to investigate the influence of process parameters and their interactions on surface roughness, followed by the development of a quadratic polynomial roughness prediction model. Analysis of variance (ANOVA) and model validation confirm the model's reliability. Finally, the MOPSO algorithm is applied to obtain the Pareto optimal solution set, yielding the optimal parameter combination. Experimental results demonstrate that at a normal contact force of 7.58 N, a feed rate of 4.52 mm/s, and a spindle speed of 5851 rpm, the achieved MRR and Ra values are 0.2197 mm3/s and 0.291 μm, respectively. These results exhibit errors of only 5.64% and 2.65% compared to model predictions, validating the proposed method's effectiveness.

Nitinol-shape memory alloys (SMAs) are widely preferred for applications of automobile, biomedical, aerospace, robotics, and other industrial area. Therefore, precise machining of Nitinol SMA plays a vital role in achieving better surface roughness, higher productivity and geometrical accuracy for the manufacturing of devices. Wire electric discharge machining (WEDM) has proven to be an appropriate technique for machining nitinol shape memory alloy (SMA). The present study investigated the influence of near-dry WEDM technique to reduce the environmental impact from wet WEDM. A parametric optimization was carried out with the consideration of design variables of current, pulse-on-time (Ton), and pulse-off-time (Toff) and their effect were studied on output characteristics of material removal rate (MRR), and surface roughness (SR) for near-dry WEDM of nitinol SMA. ANOVA was carried out for MRR, and SR using statistical analysis to investigate the impact of design variables on response measures. ANOVA results depicted the significance of the developed quadratic model for both MRR and SR. Current, and Ton were found to be major contributors on the response value of MRR, and SR, respectively. A teaching–learning-based optimization (TLBO) algorithm was employed to find the optimal combination of process parameters. Single-response optimization has yielded a maximum MRR of 1.114 mm3/s at Ton of 95 µs, Toff of 9 µs, current of 6 A. Least SR was obtained at Ton of 35 µs, Toff of 27 µs, current of 2 A with a predicted value of 2.81 µm. Near-dry WEDM process yielded an 8.94% reduction in MRR in comparison with wet-WEDM, while the performance of SR has been substantially improved by 41.56%. As per the obtained results from SEM micrographs, low viscosity, reduced thermal energy at IEG, and improved flushing of eroded material for air-mist mixture during NDWEDM has provided better surface morphology over the wet-WEDM process in terms of reduction in surface defects and better surface quality of nitinol SMA. Thus, for obtaining the better surface quality with reduced surface defects, near-dry WEDM process is largely suitable.

Recently, a roll-type linear chemical mechanical polishing (Roll-CMP) process was developed for fabricating large flexible substrates such as flexible printed circuit boards (FPCBs). The major difference between the Roll-CMP and the conventional CMP is the type of contact between the polishing pad and the substrate. Roll-CMP uses line contact for material removal of the Cu film on FPCBs. Many researchers have studied mathematical models to understand the conventional CMP process. In this paper, a mathematical model on the material removal rate (MRR) of Roll-CMP is proposed based on Hertzian contact theory and previously studied models on conventional CMP to understand the effect of the polishing pad. Two kinds of polishing pads were prepared to investigate their effect on the material removal of copper clad laminate (CCL). The increase in the average radius of pad asperities over the standard deviation of pad asperities increases MRR. The slurry loading capacity of the polishing pad impacts the MRR of Cu as well. The proposed model may offer a theoretical understanding for the Roll-CMP process.

<title>Abstract</title> The high-efficiency precision grinding of Zerodur still faces challenges in improving the material removal rate while ensuring the surface quality. In this study, the forming mechanism of glass-ceramic grinding surface under the coupling of multiple process parameters was systematically explored through theoretical modeling and experimental verification. Based on the assumption of rigid conical abrasive particles, a theoretical model of surface roughness (Ra) and material removal rate (MRR) was established. The influence of grinding wheel particle size, normal pressure and feed speed on Ra and MRR was revealed and verified by orthogonal experiments. The results show that the grinding wheel particle size has the most significant effect on Ra, and the use of 1000 # grinding wheel can reduce Ra by about 66 percent compared with 600 # grinding wheel. MRR increases significantly with the increase of grinding wheel particle size and normal pressure, but the surface quality decreases. The feed rate and feed speed have little effect on Ra, but can effectively improve MRR; increasing the rotational grinding wheel speed can improve the surface roughness to a certain extent, but the improvement effect on MRR is limited. Studies have confirmed that by optimizing the combination of grinding wheel particle size, normal pressure and motion parameters, high-efficiency grinding of Zerodur can be achieved under the condition of ensuring surface quality.

ABSTRACTThe current work presents a detailed exploration on real-time wire electric discharge machining (WEDM) experiments and grey relational analysis (GRA)–based multi-criteria optimization of material and machining characteristics for lowered surface roughness (Ra) and improvised material removal rate (MRR) of the newly developed magnesium/boron nitride/cathode ray tube (Mg/BN/CRT) hybrid metal matrix composites (MMCs). The composites were fabricated through powder metallurgy (PM) route by reinforcing silica-rich E-waste CRT panel glass powder crushed for different particle sizes (10, 30, and 50 µm) at various weight percentages (5%, 10%, and 15%) and with 2% boron nitride (BN). Taguchi-based orthogonal array procedure was utilized to formulate the experimental plan for WEDM considering reinforcement level and size, pulse on time (Pon), pulse off time (Poff), and wire feed (Wf) as the input process parameters. ANOVA results reveal that Pon and wt% of reinforcement has more effect on Ra and MRR than any other considered parameters. The developed mathematical model for Ra and MRR predicted values similar to that of experimental results. Multi-criteria optimization was done through GRA technique and the so recommended optimum parameter set furnishes higher MRR (22.34 mm3/min) and reduced Ra (2.87 µm).

The role of non-ionic surfactant concentration in the planarization performance of copper wafer is investigated. The basic components of the slurry contain H2O2, a mixture of benzotriazole (BTA) and ammonium dodecyl sulfate (ADS), FA/O II (macromolecular organic complexing agent), and silica. The static performance of the alkaline slurry including the surface tension, viscosity, particle size distribution and zeta potentials is analyzed. The effect of AEO on material removal rate (MRR) profiles, within-wafer non-uniformity (WIWNU) and surface roughness is also discussed. The AEO in combination with a small amount of BTA and ADS has a large effect on the stability of abrasives, MRR, WIWNU and surface roughness. A new MRR model is proposed to elucidate the polishing mechanism, revealing that the inhibition properties of AEO can be characterized through the corrosion inhibition efficiency and adsorption behavior. An optimal choice of the AEO concentration can achieve a good stability of the slurry, high MRR, low WIWNU and planar surface. The surfactant concentration ranging from 0.1 wt% to 0.5 wt% would contribute to acquiring a good surface planarization performance. Therefore, this work provides a comprehensive understanding of the effect of AEO concentration on the surface planarity of copper CMP in alkaline slurries.

Magnesium is reinforced with three different weight percentages (5%, 10% and 15%) of SiC particles (200 mesh size) by stir casting technique to fabricate Mg/SiCp composites. The Scanning Electron Microscope (SEM) images, micro and macro hardness of three different composites are investigated. The comparison of micro and macro hardness clearly shows that increase in the weight percentage of SiC contributed to increase in hardness. However, uniform dispersion of SiC can be achieved while adding 5% SiC in the composite. Then, the Box Behnken experimental design in response surface methodology is employed for machining 3[Formula: see text]mm diameter hole in the Mg/SiCp samples using EDM. The second-order model for Material Removal Rate (MRR) and Tool Wear Rate (TWR) are developed with the influencing parameters of weight percentage of SiC, current, pulse on time and pulse off time. The parameter optimization yields maximum MRR and minimum TWR.

The wire electrical discharge machining (WEDM) allowed success in the manufacture of the hard, fragile, and materials difficult to cut, especially for electroconductive ceramic materials. In this study, the mathematical models of material removal rate (MRR) and surface roughness (SR) used for the machinability evaluation in the WEDM process of aluminum oxide-based ceramic material (Al2O3 + TiC) have been carried out. The experimental plan adopts the face centered central composite design (CCD). The mathematical models using the response surface methodology (RSM) are developed so as to investigate the influences of four machining parameters, including the peak current, pulse on time, duty factor, and wire speed, on the performance characteristics of MRR and SR. It has been proved that the proposed mathematical models in this study would fit and predict values of the performance characteristics, which would be close to the readings recorded in experiment with a 95% confidence level. The significant parameters that critically affect the performance characteristics are examined.