The application of neural networks for modeling of the complex dependences of steel compositions characteristics from the relatively big number of alloying elements considered demonstrated their power as non-linear MIMO functions approximations. They allow us to carry out further optimization experiments by simulations in order to spent expensive laboratory synthesis and analysis of different steel compositions.
The results of the solved above optimization tasks are meaningful and will help further creation of high strength steel compositions for machine-building industry.
Our intensions for our future work are to include some other optimization criteria as well as to test the obtained steel “recipes” in practical experimental investigations in order to prove and refine their quality.