Omsk State Agrarian University scientists develop an AI system for identifying agricultural machinery breakdowns

On the basis of the International scientific and methodological center (ISMC) of Tomsk State University, there was defense of the projects developed by the faculty staff members of Russian HEIs using end-to-end technology. Out of 100 projects, 26 made it into the final. According to international experts, the prototype of the digital predictive analytics platform for agromachines, Hector Al, created by scientists of Omsk SAU and the University's industrial partners, is in the Top-3 initiatives. This new tool can significantly reduce the economic losses that Russian farmers incur during the harvest period due to the breakdowns of agricultural machinery.м

The projects development was parallel to a four-month training course organized by the TSU ISMC as part of the national project "Digital Economy". It was attended by 1.200 faculty staff members who obtained new skills and knowledge in the field of artificial intelligence, big data analysis, use of mixed and augmented reality technologies, etc. The acquired knowledge and pieces of expert advice helped many participating teams transform promising ideas to real products.

The project, that was recognized by the jury as one of the most successful, was developed by the team of Omsk State Agrarian University. Its members were Head of IT Department Pavel Revyakin, IT Department engineers Anna Basakina, Vyacheslav Kling, Vitaly Chernopolsky, and Associate Professor of the Department of Mathematical and Natural Sciences Pavel Kiyko.  The developers decided to solve one of the most crucial problems faced by farmers in many countries – breakdowns of agricultural machinery during the harvesting campaign. They decided to put emphasis on combine harvesters, which are the most in-demand machines in fields. 

Pavel Revyakin, Head of the IT Department of the Omsk State Agrarian University: 

The combine harvester works two or three weeks a year, but it is a highly intensive work that challenges all the elements of the machine. The quality of grain is affected by night humidity and dew, so combine harvesters work constantly.

The failure analysis shows that combine harvesters have components that wear out faster than others - elements of the grain transportation system: augers and their drive, conveyor blades and the conveyor belt itself, bearings and belts.

A platform for predicting breakdowns will help reduce the need for emergency repairs. Its main tools are big data analysis and the use of machine learning technologies – artificial intelligence methods that allow teaching a neural network to identify problems before they occur and develop solutions.

The project team worked with a data sample containing telemetry data of 5 test combine harvesters for the year of operation, data on breakdowns, and operational and passport characteristics of the machines.

Using statistical research methods, the developers identified trends in the frequency of failures in the operation of combine harvester units. After that, a predictive analysis model was created. It allowed detecting the rotation speeds of the working units of the combine harvester that sufficiently differed from the nominal parameters.

Using this data, a computer model can predict a malfunction in a particular unit in advance. For example, the AI can predict the malfunctions caused by breakage and stretching of belts that set in motion all the working units of the combine. Emergency diagnostics and replacement of belts leads to daily downtime of the combine harvester (taking into account the time for delivery of spare parts and the arrival of the service team to the broken vehicle).

The platform makes it possible to assess the wear and remaining lifetime of machines and mechanisms, and monitor compliance with technological operations. Finally, this can significantly increase the economic efficiency of agricultural machines.

Alexander Zamyatin, Director of the TSU ISMC

The Omsk SAU team had a good potential when joining our project. During the training, the colleagues obtained those competencies that they needed to transform their prototype into the final product. The Omsk team is ready to improve it, possibly, together with TSU staff.

We believe that this project is very promising in terms of commercialization. I think that agricultural machinery manufacturers will definitely appreciate this development. It can increase the reliability of their products, and, therefore, its competitiveness in the Russian and foreign markets.

Mr. Zamyatin noted that the digitalization of industry is a trend spread worldwide. Such well-known companies as Siemens, Yokogawa, and Schneider Electric are working on creating predictive analytics systems. Supported by the State, Russian scientists are also working on similar systems. 

One of the priorities of the national project "Digital Economy" is the development of the "Industry 4.0" concept. This will help Russia very significantly increase the number of solutions to improve the efficiency of the domestic industry.