Abstract
To achieve the optimal performance of an object to be heat treated, it is necessary to know the value of the Heat Transfer Coefficient (HTC) describing the amount of heat exchange between the work piece and the cooling medium. The prediction of the HTC is a typical Inverse Heat Transfer Problem (IHCP), which cannot be solved by direct numerical methods. Numerous techniques are used to solve the IHCP based on heuristic search algorithms having very high computational demand. As another approach, it would be possible to use machine-learning methods for the same purpose, which are capable of giving prompt estimations about the main characteristics of the HTC function. As known, a key requirement for all successful machine-learning projects is the availability of high quality training data. In this case, the amount of real-world measurements is far from satisfactory because of the high cost of these tests. As an alternative, it is possible to generate the necessary databases using simulations. This paper presents a novel model for random HTC function generation based on control points and additional parameters defining the shape of curve segments. As an additional step, a GPU accelerated finite-element method was used to simulate the cooling process resulting in the required temporary data records. These datasets make it possible for researchers to develop and test their IHCP solver algorithms.
Highlights
Quenching processes are typically operated in different liquid or gaseous media to achieve the desired material properties by following the adequate heat transfer rates
The heat transfer phenomena occurring during the quenching process may pass three different boiling regimes [1]
The component is surrounded by a vapor film, which insulates the work piece
Summary
Heat treatment operations are used to change the material properties of components. The controlled input or extraction of heat is essential during the whole process. The prediction of the HTC is a typical ill-posed task, which cannot be solved by direct numerical methods In recent years, this Inverse Heat Transfer Problem (IHCP) has been studied extensively [2,3,4,5,6], presenting various heuristic solutions based on Genetic Algorithms (GA) [7,8] or Particle Swarm Optimization (PSO) [7,9]. The back-propagation algorithm starts with random W and b values, feeds the network with the input data and measures the difference between the prediction of the network (Y) and the known valid output (Y ). The error is propagated back to the previous layers recursively and the weights of edges are adjusted according to this This process is repeated until the loss (difference between the desired and actual output) is satisfactory.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
