Design of a system for diagnosing the technical condition of a gas pumping unit based on the analysis of existing methods of processing vibroacoustic signals

Abstract

The gas transportation system of Ukraine provides transportation of natural gas from deposits to consumers located both on the territory of Ukraine and outside its borders. One of the key elements of the gas transportation system of Ukraine are compressor stations, the task of which is to maintain the specified pressure in the system for uninterrupted transportation of natural gas. The main element of the compressor station is the gas pumping unit (GPU). It is a complex mechanical system consisting of many nodes. In the process of functioning of the gas pumping unit, in its nodes and units, the process of wear of parts occurs, which, subsequently, can lead to an accident. To avoid such situations, current and capital repairs of the system are periodically carried out. A number of non-destructive technical condition control methods have been developed to monitor the technical condition of the GPU between repairs, to optimize the periodicity of their carrying out and to provide early warning of the GPU leaving the nominal state. Such methods are divided into parametric methods, based on the analysis of the operating parameters of the system (temperature and pressure at various GPU nodes, turbine rotation frequency, chemical composition of combustion products), and vibroacoustic, based on the analysis of vibration and acoustic oscillations generated by the GPU nodes. The current state of vibroacoustic signal processing methods, such as Fourier transform, wavelet transform, artificial neural networks, discrete cosine transform, and autocorrelation functions are considered in the work. Based on the analysis, a system for diagnosing the technical state of the GPU based on the Fourier transform in combination with the wavelet transformation and an artificial neural network with two hidden layers is proposed, the task of which is to recognize the technical condition of the GPU, based on the result of the wavelet transformation, divided into separate images in primary colors.