International Scientific Journals
of Scientific Technical Union of Mechanical Engineering "Industry 4.0"

High-voltage electrical discharge (HVED) treatment of powder mixtures is a modern, efficient, and economically advantageous method for both particle size reduction and modification of the material’s phase composition. The primary mechanisms of particle destruction within the discharge zone include shock waves, microcavitation, ablation, collisions with chamber components, and mutual abrasion between particles.
The application of machine learning methods to model HVED processes for titanium—a promising material for composite applications— enables more accurate predictions and optimization of the technological workflow.
The data used for modeling were obtained between 2013 and 2021 and include results from the treatment of the initial titanium powder (with an average diameter of d₀ = 60 μm) in ethanol. This setup enabled the formation of a volume-distributed multi-spark discharge (VMD) within the ethanol–powder dispersed system. The dataset includes information on the number of treatment pulses, discharge gap, pressure in the discharge channels, pressure on the chamber walls, and the amount of titanium carbide formed during the treatment process.
It was shown that the concentration of TiC gradually rises with the increase of specific treatment energy, regardless of the interelectrode gap. Specifically, at a specific energy (Ws) of 5 to 15 MJ/kg, the amount of titanium carbide reaches 10%; at 15 to 30 MJ/kg, it increases to 20%; and at energy levels above 30 MJ/kg, the TiC content reaches 30%.
Keywords: Ethanol, Titanium, Titanium Carbide, High-Voltage Electrical Discharge, Volume-Distributed Multispark Discharge, Electric Discharge Dispersion, Plasma Technologies, Machine Learning, Logistic Regression, Random Forest

The development of Industry 4.0 and Predictive Maintenance Systems allows for effective utilization of equipment by incorporating ML methods for identifying tool condition. However, including large number of Condition Indicators for machinery monitoring increases computational complexity, hence the response of the system elongates. Therefore, it is important to check the utility of indicators and reduce them. In this paper, we investigate different feature selection methods: AIC (Akaike Information Criterion), BIC (Bayesian Information Criterion), Random Forest, Lasso Regression (L1 Regularization) for NASA Gearbox Fault Detection Dataset, PHM 2009. We processed the raw data and calculated CI from time domain, frequency domain and envelope. An SVM Classifier model was trained on full collection of indicator and reduced, then performances were compared. The obtained results highlight the advantage of feature selection, proving that effective PdM systems can be based on diminished number of Health Indicators.

The possibility of using machine learning methods to predict the results of high-voltage electric discharge treatment of titanium powder in a hydrocarbon liquid is studied. As a result of the work, distribution surfaces for the average particle diameter of Titanium powder. the amount of Titanium carbide formed during processing, and the number of spherical particles of titanium powder depending on the interelectrode gap and the number of pulses, when using spark discharge and with Titanium powder concentration in kerosene of 0.07 kg / dm3, pulse repetition frequency 0.3 Hz and the energy of single discharge of 1 kJ, were obtained.