Operational Data Analysis
System Operation Optimization Through Data Mining
Data mining is a database research process for identifying hidden correlations and new information. Data are submitted to a mathematical processing line (statistical analyses, artificial intelligence, decision trees, etc.) in order to extract additional knowledge. This knowledge (rules, models and guides) can then be implemented as online software solutions to:
- Understand process variability causes
- Develop key performance indicators
- Monitor process performance trend and keep it optimal
- Perform simulation scenarios
- Plan operations and manage requests
- Better manage abnormal situations (fault detection and diagnosis)
Analyzing Process Data to Reduce Operating Costs
Data mining techniques allow reducing:
Improvement of product quality, increased process stability and reduced interventions and non-scheduled downtime are additional benefits that can also be generated through data mining techniques.
CanmetENERGY is currently developing software that will allow its users to:
- Clean, prepare and analyze its data to gain knowledge
- Understand variability and develop key performance indicators (KPI) by using principal component analysis (PCA)
- Perform process modeling and develop predictive model through neural networks and linear regression (PLS)
- Monitor and detect faults by using a early warning system dashboard
A kiln drying planning tool for the lumber sector has also been developed, in collaboration with Laval University’s FORAC Consortium.
CanmetENERGY, in collaboration with various partners, including the Bureau de l’efficacité et de l’innovation énergétiques, Hydro-Québec and BC Hydro, is in the process of developing case studies to demonstrate the viability and impacts of data mining-based solutions in several industrial sectors.
- Date modified: