Clover PMM
Lightweight Analytics for Heavyweight Machinery

Predict equipment failures with 100% accuracy. Carry out repairs based on actual technical condition. Reduce downtime and increase efficiency.
Get a pilot project
Monitoring and Forecasting in the Heavyweight-Machinery Industry
Clover PMM employs telemetry sensors to collect streams of data in an enterprise. Holistic mathematical models analyze both the operation of the equipment and the external parameters, while machine learning predicts possible breakdowns and their causes.
1
Collects data
3-5 weeks of project development
2
Analyzes the work
2 months after integration
3
Predicts failures and stoppages
for an interval of up to 3 months
Clover PMM Capabilities
Predicts equipment failure
Clover PMM analyzes large data streams, tracks anomalies and predicts breakdowns with a lead time from 2 weeks
Reduces downtime in an enterprise
Clover PMM clients in the engineering industry reduce their enterprise downtimes by 35-45%
Cuts on the expenses on unplanned repairs
Using Clover PMM cuts down the expenses spent on unplanned mechanical engineering repairs by more than 20%
The Problem of Mechanical Engineering
Scheduled preventive repairs in mechanical engineering do not insure against unplanned equipment stops. This raises the costs spent on components and the remuneration of labor, maintenance crews, as well as independent contractors. Stops often lead to punitive penalties and loss of funds.
Unforseen expenses
Preventive maintenances in mechanical engineering do not insure against unplanned equipment stops.
Clover PMM Solution
Clover PMM incorporates two modules: anomaly detection and predictive analytics. Prediction is based on unique MX-models (Math eXperience). These are developed using physical and mathematical modeling, as well as machine learning and engineer-expert rule packages.
Instead of just singular units, mathematical models describe a set of interrelated equipment. For example, in transport engineering, the Clover MX-Model is not built for individual installations, but for the entire locomotive.
Get a pilot project
Hybrid Math Models for "Raw" Data
Conventional math models are developed under "ideal" conditions, and cannot be used to monitor actual production – any changes in parameters require modifications. Also, disparate models of various components do not run together.

At the same time, production-related data contains large errors, both in quality and quantity.
The Clover Team arrived at the solution to the math models problem from two angles. The scientific part of the team investigated the operation of the equipment in all modes, climatic zones, as well as regulated and non-regulated scenarios. IT professionals developed a convenient multi-component platform on which the models operate.
Get a pilot project
Clover PMM: Real-World Testing
Mathematicians and engineers, production experts and programmers summarized the theory, general methods of forecasting and probabilistic estimations, added packages of expert rules and made the mathematical modeling tool operative, accurate and convenient.

The testing of Clover PMM in real-world conditions was also facilitated by the wide distribution of climate zones in Russia, as well as different series of equipment. The models were time-tested in industrial plants, on power plants, locomotives and other objects.
Clover works with "raw" data, complements it and sorts it. Machine learning finds patterns that are not obvious to humans efficiently.
Advantages of Clover PMM
Hybrid math models
Describe a set of interrelated equipment
Machine learning
Identifies interconnections in big data streams that are non-obvious to humans
Predictive Analytics
Predicts breakdowns within a two-week interval, unplanned stops within 30 minutes
Integration into IT-systems
Easily integrates with any enterprise management software
User-friendly appearance
The customer is able to view the results of the diagnostics and forecasting in charts, tables, and convenient reports
Sprint implementation
Implementation takes 2-3 months, and the effect of the system is already observed after 3-4 months
Clover PMM Post-Implementation Results
Production
  • Recognizes the actual technical condition of the equipment
  • Plans repairs and maintenance effectively
  • Introduces design changes promptly
Operation
  • Tracks the residual value of equipment
  • Enhances business continuity
  • Plans the procurement of equipment based on actual requirements
Service
  • Plans the optimal time and place of repair for each unit of equipment
  • Develops a repair plan, based on the actual status
  • Reduces fines and costs
Analyzed equipment
Locomotive
Electric Locomotive
Electric Train
Subway Train
Case
Service-Company for Servicing Locomotives
Task
The largest outsourcer of the Russian Railways services more than 6,200 sections of locomotives. Information about possible breakdowns and failures of locomotives with a precise indication of equipment units should reduce the number of unplanned stops and downtime, the workload on the company employees, as well as the financial expenses for these items.
6200
sections of locomotives
Decision
Pre-project research was complicated by the different data formats used in the client's work. Works on the project also affected business processes. This increased the lead time for effective implementation.

However, thanks to the mathematical models developed and infrastructure prepared, Clover PMM implemented a module for anomaly detection and predictive analytics in the enterprise.
Result
The first results began arriving 2 months after the start of operations — the number of unplanned stops had dropped, the client has prior knowledge of possible breakdowns. Expenses on unplanned repairs cut by 30%. Maintenance costs also decreased.

The system automatically detects more than 50 types of malfunctions in the operation of equipment and operating modes of the locomotives. Decisions on repairs are made in advance, in a planned mode, which reduces the load on the staff.
–30%
on unplanned repairs
–50
types of malfunctions
The Clover Team
27 mathematicians and engineers
develop physical and mathematical models for each type of equipment, study the boundary processes
28 programmers
create software that integrates into the IT-infrastructure of the production company
13 industry experts
study the equipment and submit their expert advice to mathematicians, business analysts in the building stages of the MX-models
53 scientific institutions
advise on sophisticated equipment of the oil industry, including the machine learning process
Your Benefits with Clover PMM
–25%
Reduced cost of technical equipment
–35%
Reduced downtime
–70%
Elimination of failures and unplanned stops
+20%
Increased production volumes
+HR
Improved staff utilization
+$
Enhanced competitiveness