Developing Methods and Algorithms for Evaluating Vehicle Reliability

“Minsk Automobile Plant” OJSC (MAZ) asked to create methods and algorithms that use censored samples of limited size to evaluate reliability of vehicles.Develop similar

The customer

“Minsk Automobile Plant” OJSC (MAZ) is the largest state-owned enterprise in Belarus that produces heavy vehicles, including buses, trolleybuses, and trailers.

MAZ is a well-known brand inside as well as outside the CIS. The vehicles produced by MAZ have a good reputation for reliability and have been successfully employed in the Far North of Russia, deserts, including the Karakum desert and the Sahara, the tropics of Africa, Southeast Asia, and America, at Near East and Middle East, as well as in the mountains of Latin America. In total, vehicles produced by MAZ have been imported to more than 45 countries around the globe.

The need

Initially, we had no data on failures and damage to vehicles produced by MAZ throughout the warranty period. This made it difficult to ensure an adequate supply of replacement parts, plan service center workloads, etc. This is why it was necessary to create methods and algorithms that use censored samples of limited size to evaluate reliability of vehicles throughout warranty periods and after their expiration.

The solution

We developed statistical methods and algorithms for:

  • statistical analysis of the database that contains data on failures and censorings of replacement parts;
  • forecasting standard consumption of replacement parts during warranty periods and after their expiration;
  • statistical analysis and forecasting of technical and economic performance based on trend models.

Our team of engineers developed a software system that uses censored samplings of limited size to evaluate and forecast vehicle reliability. We also made a research on reliability of the new vehicle models produced by MAZ in real-life conditions.

The outcome

The new software system developed by our team was used to monitor reliability of the control sampling of vehicles produced by MAZ, assess reliability of standard models, and determine standard consumption levels for replacement parts that limit reliability. As the result, the customer was able to decrease the number of spare parts that had to be stored at warehouses throughout and after the expiry of warranty periods and forecast the time during which spare parts had to be supplied for models which are no longer in production. This improved the quality of services and customer satisfaction with the products by MAZ, which in its turn increased the amount of sales and the overall revenue of the company.

Contact us

Siarhei Sukhadolski

Siarhei Sukhadolski

Artificial Intelligence practice head


location icon830 Stewart Dr., Suite 119Sunnyvale, CA 94085
First Name*
Last Name*
Your company name*
How can we help you?