Artificial intelligence

aedifion uses Artificial Intelligence (AI) in several ways to improve the user experience and building operation of our customers and as part of the .analytics product. This allows human engineers to work faster and smarter with their data.

Data point classification¶

This AI application uses the statistical properties of a data stream to identify what kind of data source is producing it. In doing so, it tries to answer the following five questions:

• Is the datapoint an analog or a digital signal?
• Is it a physical measurement, or a virtual data point?
• Is the value an input, or an output?
• What exactly is the type of the data (e.g., temperature, pressure ...)?
• What is the physical unit of the data (e.g., bar, Pascal ...)

As an answer to each question, a label is assigned to each data point. Each label can have a certain amount of values, with all possible choices explained in the Available Classes subsection. The AI assigns a probability to each possible class and then chooses the most probable class as the value for the label.The information is then used to help you.

• Identify the data point you are looking for
• Identify mislabeled data points
• Find installation errors, e.g., swapped values for inflow and return temperatures
• Convert values for further use, e.g., when summing up energy consumption from sources with different units

Available Classes¶

These are the labels that the AI tries to identify, and all possible values it can choose from.