Power Generation

Data Analysis

Operations research has used different analytical models for decision making for long. Advancements in machine learning techniques have made the implementation (development and maintenance) of many of these models easier and predictions more accurate.

For example, instead of developing a static model that is based on manual data analysis, machine learning models can now use historical data to automatically establish correlations between events. Using reinforcement learning, we can build adaptive systems, where algorithms learn while in action.

Historical Data Analysis

We can develop models to layout a plan for operating the power generation plant based on the readily available information:

• Historical electricity consumption per unit time per zone.
• Historical weather conditions in different zones and weather forecasts.
• Price of electricity forward and future contracts in the market.
• Commitments in terms of minimum and maximum electricity delivery.
• Operating parameters of the power generation plant.

Using the above, we can establish,

• Correlation between power consumption and weather – temperature and humidity.
• Correlation between power consumption over time – one unit time vs. the other.

Advancements in predictive analytics have now made it possible to easily deal with seemingly intractable problems.