مدل‌سازي و پيش‌بيني تغييرات ضخامت نسوز پاتيل‌هاي مذاب با استفاده از اسكن ليزري

The measurement of refractory-coating thickness is recognized as a key factor for monitoring the condition an‎d supervising electric-arc furnaces (EAFs). In this study, the measurement of the refractory lining of electric arc furnaces hearths is performed by means of laser scanning an‎d subsequent point cloud analysis, which is presented an‎d developed as a novel an‎d efficient method for assessing the refractory condition of EAFs. The research workflow comprises the following steps: (i) designing the furnace hearth, (ii) applying corrosion equations to the inner surface of the hearth, an‎d (iii) virtually measuring the refractory-coating thickness using laser-scan simulation. The resulting thickness measured by laser scan generates a three-dimensional representation of the refractory lining, providing a view of the interior zones an‎d the corresponding corrosion values of the EAF hearth. To predict corrosion depth over successive melting cycles an‎d to generalize this prediction to various corrosion scenarios (different regions of the hearth’s inner surface), ConvLSTM network as a deep model is employed. The predictive network estimates the corrosion amount for each melting cycle an‎d for each spatial zone with an error lower than 5 percent. Subsequently, to estimate corrosion for hearths built with different bricklaying patterns an‎d other design variations, the corrosion equations are updat‎ed with new parameters reflecting the altered corrosion rates on the hearth surface. For more accurate predictions under these new conditions, the pre-trained network is fine-tuned (re-trained) using a small set of corrosion data from the initial melting cycles of the new furnace. The fine-tuned model predicts the corrosion of the newly simulated EAF hearth with an error below 10 percent. Finally, a recursive-estimation technique is applied: using only the corrosion values from the early melting cycles, the method forecasts the corrosion in the successive melting cycles an‎d consequently estimates the remaining service life of the refractory lining. Overall, the aim of this research is to measure an‎d predict the remaining life of the refractory lining in electric-arc furnaces by integrating laser-scanning technology with deep-learning models, offering a more advanced an‎d reliable approach compared with conventional methods.

حامد جلالي بيدگلي , ايمان ايزدي نجف آبادي

جعفر قيصري , مرضيه كمالي