TrendMiner - Self-service Predictive Analytics

by TrendMiner N.V

Self-service Predictive Analytics. Driving operational intelligence through predictive analytics for process data.

Solution Overview

Process manufacturing companies continuously strive to optimize overall equipment effectiveness, performance and profitability, while complying with regulations. Turning real-time data into actionable information is a great way to optimize plant performance, but this often requires a data scientist. Relying on a central group to analyze data and create models can cause an organizational bottle neck as specialists are typically over booked with “big data” projects. This means that new projects need to be proposed, accepted and scheduled, resulting in a delay in receiving results. Process engineers who want to optimize the production process may search for other ways to find root causes of process deviations, for example spreadsheets. However, gathering the right data and handling the large volumes requires highly skilled users to turn the imported data into actual information. Better self-serving tools for the subject matter experts are needed for solving the many uses cases.


Self-service analytics enable process engineers to quickly analyze time series process data captured in the PI System. This ensures that eighty to ninety percent of process behavior assessments can be handled by the subject matter expert. Self-service analytics avoid the need to create analytics models, speeding up the time to insights and identifying influence factors to process anomalies. Monitoring good process behavior improves production consistency and predictability of product quality. Early warnings on process behavior help deliver plant output with optimum energy consumption and reduction of waste, while complying with safety, health and environmental regulations. By monitoring best operating zones for equipment, downtime can be reduced and required maintenance predicted.

Solution Approach

All that is required to implement TrendMiner is a PI System on top of a Distributed Control System (DCS). TrendMiner integrates with the PI System as a business application and reads from the data infrastructure level, so no direct connection with the DCS is required. It does not open another separate data archive; it uses the existing process data in the PI System(s.) TrendMiner can index across multiple PI Systems, increasing interoperability and enabling fast search and analytics. The software is deployed using a virtual machine so there is no impact to the existing infrastructure. End users in turn access TrendMiner via a web-based interface, so no data can "escape". TrendMiner forms a central data analytics platform between the PI System/MES, generic big data toolings and business intelligence systems. Integrations include MES, PLM, CPLM, APM and historian tools, with options to import from SAP and LIMP and export to Excel. TrendMiner is an Original Equipment Manufacturer (OEM) partner of OSIsoft. TrendMiner is a plain HTML5 web-client that needs no local installations or plugins (plug & play). It can be deployed with VMware, Hyper-V and KVM. Downloading and installing the prepackaged virtual appliance takes less than an hour and can be done by your own IT team. As soon as the appliance is connected, users can access TrendMiner and view both retrospective and live data.

At-A-Glance

Features

  • Self-service analytics
  • Time series data visualization
  • Advanced pattern recognition
  • Automated process monitoring

Benefits

  • Plug & play solution
  • Predict process performance
  • Reduce unplanned downtime

PI System Requirements

PI System version 3.4.390 or later, PSA Server

Solution Type

Advanced Analytics, Event/Batch Analysis

Industry

  • Chemical & Petrochemical
  • Mining, Metallurgy & Material
  • Oil & Gas
  • Pharmaceuticals & Life Sciences
  • Power Generation
  • Pulp and Paper
  • Food & Beverages

Business Impacts

Optimize Processes, Evaluate Quality, Increase Asset Health & Uptime, Improve Energy Efficiencies, Manage Risk & Regulatory

Category

Applications for the PI System