Solution Of Numerical And Statistical Data And Construction Of Prediction Model By Means Of Computer- Aided Learning Program Light gbm For Solar Heat Supply «Flat Solar Collector +Heat Pump» System

Yedilkhan Amirgaliyev, Murat Kunelbayev, Aliya Kalizhanova, Ainur Kozbakova, Astanayeva Aigerim

In the work herein we have carried out numerical solutions of statistical data and construction of the prediction model by means of machine learning program LightGBM for the solar heat supply system. Gradient Boosting Decision Tree (GBDT) is a popular algorithm of computer-aided learning and it has quite a few effective implementations, such as XGBoost and pGBRT. For solving the problem solution definite algorithm there have been conventionally selected the thresholds, in order the right-sside purpose oriented variable drops down, entry variables grow. We have developed four algorithms, which show, how there is established the possibility of constructing the functional dependences between inlet and outlet parameters by means of machine learning program LightGBM. From conducted numerical experiments we can draw a conclusion, that the LightGBMis considerably well trained, it accelerates training process and further, we can say, the LightGBM package has been well enough trained for the solar heat supply systems.

Volume 11 | 08-Special Issue

Pages: 2733-2745