Hybrid MLP-CNN (Neural Network)

History

2025: Successful application of new algorithm for automatic diagnostics of bearings using neural network

Scientists of the South Ural State University (SUSU) for the first time in Russia successfully applied a new algorithm for automatic diagnostics of bearings using a neural network. The innovative approach made it possible to increase the speed of determining the state of bearings by 15 times compared to traditional methods. This was announced by the research laboratory of Technical self-diagnostics and self-control of devices and systems at the end of April 2025.

According to TASS, the key scientific and technical problem was that a hybrid model of the Hybrid MLP-CNN neural network was previously used for automatic diagnostics of bearings. This model required significant computational and time, as well as large amounts of data for training. If there is not enough training data, the system worked with errors.

𝓲

Фото: Freepik

Deputy Head of the Research Laboratory of Technical Self-Diagnostics and Self-Control of Devices and Systems Vladimir Sinitsyn said that to solve this problem, Russian scientists for the first time successfully used the LPC (Linear Predictive Coding) algorithm, which requires 15 times less time to train the neural network. He explained that the LPC algorithm is widely known, but was used mainly for recognizing human speech and synthesizing it.

The effectiveness of the new algorithm was tested on test datasets, and the results of the study are published in the highly rated international journal Algorithms. The introduction of this technology is of high practical importance for various industries.

The simplification and acceleration of the bearing diagnostics system is especially important for industrial enterprises, motor transport and railway companies. Timely troubleshooting of bearing assemblies prevents costly equipment from failing, reduces downtime, and reduces operating costs.^[1]

Notes