Temperature Prediction of Heating Furnace Based on Deep Transfer Learning

Heating temperature is very important in the process of billet production, and it directly affects the quality of billet. However, there is no direct method to measure billet temperature, so we need to accurately predict the temperature of each heating zone in the furnace in order to approximate the billet temperature. Due to the complexity of the heating process, it is difficult to accurately predict the temperature of each heating zone and each heating zone sensor datum to establish a model, which will increase the cost of calculation. To solve these two problems, a two-layer transfer learning framework based on a temporal convolution network (TL-TCN) is proposed for the first time, which transfers the knowledge learned from the source heating zone to the target heating zone. In the first layer, the TCN model is built for the source domain data, and the self-transfer learning method is used to optimize the TCN model to obtain the basic model, which improves the prediction accuracy of the source domain. In the second layer, we propose two frameworks: one is to generate the target model directly by using fine-tuning, and the other is to generate the target model by using generative adversarial networks (GAN) for domain adaption. Case studies demonstrated that the proposed TL-TCN framework achieves state-of-the-art prediction results on each dataset, and the prediction errors are significantly reduced. Consistent results applied to each dataset indicate that this framework is the most advanced method to solve the above problem under the condition of limited samples.

Prediction Model of Temperature of Cast Billet Based on Its Heating Retrospection Using Boosting “Random Forest” Structure

The scope of this research is the prediction of a cast billet surface temperature, which it will have in the rolling mill after the heating process. The main problem is that such a prediction is needed before the cast billet will really leave the furnace. In many cases, the boundary value problem of the heat transfer, particularly the differential equations of the transient heat conduction, is used to solve this problem. But in this research an alternative data-driven approach is proposed, which is based on a model of the dependence of the billet temperature on the retrospection of its heating in the continuous furnace. Such a model is developed as a result of the analysis of the data from the furnace control system. Such data from the real furnace were collected and stored in the data warehouse. Their exploratory analysis was conducted. All data were splitted into training, testing and validation subsets. As a part of this research, the regression model previously developed by the authors was also validated. It seemed to be overfitted (the error on the test set was significantly higher than the one on the training set). To overcome this disadvantage, an alternative method to develop the required data-based model is proposed by authors on the basis of the Boosting and Bagging algorithms. They belong to the machine learning field. As a result of the experiments with the bagging and boosting, the required model structure was chosen as a “Random Forest” with special class of the regression trees known as DART (Dropout Adaptive Regression Trees). Based on a significant number of experiments with that model, the two confidence intervals of the temperature prediction were found: 68 % and 95 % ones. The mean value of the temperature prediction error was estimated as ~ 9 °C for both the test and validation sets.

METHOD OF THE RESEARCH OF OXIDIZABLE METAL HEATING BY SOFTWARE SUITES OF ENGINEERING ANALYSIS

The article is devoted to the development of a method for modeling the heating of oxidized metal billets, in which the dimensions and thickness of the scale layer vary with time. The approach used in this development facilitates the appliance of modern software packages for the analysis of objects with varying geometry; and due to this the complexity of developing mathematical models of several metallurgical processes can be dramatically reduced. To simulate the process of metal oxidation, the method of equivalent thermal conductivity was used. The experimental verification of the method is performed and the possibility of its use for improving the methods of controlling the processes of industrial heating is shown. This method was worked out during experiments on the furnace №3 with walking beam of the mill 150 at Nizhne-Serginsk Hardware and Metallurgical Plant. Calculations were made to determine the thickness of the scale layer, which varies with time; the corresponding dependencies were constructed. The problem was solved by ANSYS Multiphysics software package as a problem of non-stationary heat conduction with boundary conditions of the first kind. During modeling, a finite-element grid was constructed, sufficiently detailed to obtain reliable results and, at the same time, allowing to solve the problem on low-power computers. In the course of solution, a number of simplifications were applied, in particular, simplification of the computational algorithm, in which the thickness of the scale layer is uniquely determined by surface temperature of the billet. Temperature distribution along the billet’s thickness was determined. Graphs and isotherms were constructed to compare values of the temperatures in metal and in scale layer. Also, a comparison of the temperature differences in the scale layer determined by the calculation method was made for the furnace and experimental conditions. In this study, the problem is considered as nonstationary, with varying boundaries. The research object is preparation of the metal (real solid) with scale layer, increasing with time. When solving a problem, this real solid was replaced by a conditional one with constant averaged dimensions. According to the equality of thermophysical processes, properties of the conditional solid were determined, whose change is equivalent to the dimensions of the real solid.

Thixoforming of Hypereutectic AlSi12Cu2NiMg Automotive Pistons

The microstructure and mechanical properties of thixoformed AlSi12Cu2NiMg (AЛ25) aluminium alloy were investigated. Cooling slope method was employed in order to produce non-dendritic billets. Thixoforming process parameters were determined as follows: die temperature of 250 °C, billet temperature of 555 - 560 °C, punch velocity of 7 mm/s. Mechanical properties of automotive piston with ultimate strength of 309 MPa, yield strength of 274 MPa and elongation of 6.8 % in the T6 condition were obtained successfully, implying success of advantages of cooling slope method.

Process Parameter Optimization of Hydrostatic Extrusion Using Metaheuristic

Friction at die work piece interface can be reduced by the presence of lubricant in the entry zone. As the metal–metal contact in the deformation zone is avoided, the deformation force is reduced and at the same time the product quality and die life are improved. So the prediction of minimum film thickness and its control through process variables is gaining interest in the industry. Two multi-objective nonlinear optimization problems are formulated in this study, one using the minimum thermal thickness and billet temperature at the work zone. And the other optimization problem is formulated by including the minimum isothermal thickness at entry zone and the billet temperature at the work zone. These two multi-objective optimization problems have been solved using particle swarm optimization algorithm (PSO). The key process variables are identified by coefficient of variance (COV) analysis. A sensitivity analysis on the process variables is performed to prioritize the process variables. A design procedure is explored to demonstrate the industrial application of this analysis.