Use of artificial neural networks for solving the problem of residual resource estimation of hoisting cranes

The work is devoted to the problems of practical application of artificial neural networks (ANN) for estimation of residual resource of hoisting machines on the example of estimation of residual resource of bridge-type cranes. The work describes in detail the programming process of ANN which allows to determine the residual resource of metal structures of a hoisting crane depending on degradation of hardness of the metal structures material. The purpose of the work was to create new software for solving the problem of estimating the residual resource of the hoisting crane which appointed service life has ended. The basis for training of ANN was the existing knowledge of hoisting cranes operation, namely the statistical database of typical crane damages, expert assessment of the technical condition of hoisting cranes, and recommendations of the existing standards to identify defect indicators of hoisting cranes. The software is written in Python programming language. The work lists different ways of programming a neural network for an estimation of residual resource of hoisting cranes. The first way is more complex and painstaking and consists of writing the programme code of the neural network manually, the second way involves using a free interactive library Skicit-learn. A new software has been created to estimate the residual resource of hoisting cranes depending on degradation of hardness of metal structures material. The programme can be used by specialists and experts as an intellectual decision-making support system for determining the residual resource of hoisting cranes. The detailed algorithm of neural network programming is presented for possible use of this information in other areas of technical diagnostics of various mechanical engineering objects.