Polynomial approximation of dynamic signals of single platform railway scales

Abstract

Purpose. Obtaining an approximating function (or a system of approximating equations), which, with a minimum error, will make approximations to the available data on a train of railway objects through 1 platform scales. Methodology. To solve this problem, numerical methods are used, namely, the approximation by polynomial functions of the nth order. The experimental data on the basis of which the experiments were carried out were obtained from the weighing and identification system of wagon in motion on a single platform scale. The approximation process is automated using a program written in the Python programming language in which the polyPit and polyid functions of the numPy library are used to obtain the polynomial coefficients. Findings. Due to the use of polynomial approximation in data processing from tensometric railroad weighing systems, it was possible to obtain a system of linear equations that, with minimal error, restored the experimental data that were obtained from the existing system of the Severny GOK: Metinvest enterprise. When normalizing the readings of the sensors from conventional units, obtained from the summing box to the range of values [0; 1] it became possible, in percentage terms, to describe a railway object. This makes it possible to avoid the dependence of the final results on the travel speed of the carriage or locomotive, which leads to an increase in the accuracy of the identification of cars in the rolling stock due to the use of the percentage of the axles staying on the weighing platform (approach / exit). It became possible to determine the type of carriage with the same number of axles, but different characteristics of the center space and the base of the rolling stock. Originality. The novelty is to obtain a general method of approximation of experimental data of the passage of wagons through a single-platform scales, which can be used to train intelligent systems and generate close to real data of the passage of a car (due to the imposition of noise, etc.). Practical value. Improving the accuracy and speed of the carriage identification as a whole, which reduces the plant downtime, contributes to an increase in the number of weighed and identified moving objects, as well as the ability to identify the type of carriage with the same number of axles in the train. The methods presented in the work can be used both for identification and for tasks, the end result of which is the classification of input data (neural networks, etc.).