Skip to content

Disaggregation Properties File

Jump to bottom Edit New page
Anthony Chrysopoulos edited this page Jul 25, 2014 · 14 revisions

General Description

In the Training Module folder, there is a Disaggregation Properties File which enables adjusting some of the configuration properties of the Disaggregation procedure. There are several options available for the user to change.

  1. Clean Dataset: This option enables the cleaning pre-process procedure of the data set. This option may fill missing values, remove days that meters where stuck etc. Default Value = true

  2. Normalize Dataset: This option enables the normalization pre-process procedure of the data set. This option may detect and handle outliers, such as negative or extremely high values of active power. Default Value = true

  3. Appliance Type: This option is responsible for the characterization of the appliances type produced by the Disaggregation procedure. There are the following choices available:

  • Activity: In this case, the appliances are categorized to a certain activity and (in some cases) a certain appliance type. The result may vary from reality, since they are based on heuristic rules.
  • Generic: In this case, the appliances are not categorized on activities and appliance types. They are are all under the Generic activity and numbered as different appliances.
  • List: In this case, the user must provide the list of appliances that the installation is equipped with, as well as their nominal active and reactive power as given by their manufacturer. The loads found by the disaggregation procedure will be then distributed to the appliances provided from the user. Default Value = Activity

  1. Loose Coupling: This option provides a less strict coupling for the appliances. This can be very helpful in case the measurements are a bit noisy and/or there is much uncertainty as far as the appliances' consumption is concerned. This usually leads to less appliances that are more genetic in nature and can be described more loosely. Default Value = true

  2. Fridge Loose Coupling: This option provides a less strict coupling for the refrigerator detection procedure. This can be very helpful in case the measurements are a bit noisy and/or there is much uncertainty as far as the refrigerator's consumption is concerned. Default Value = true

  3. Washing Machine Detection: The disaggregation procedure implemented in the Training Module includes a very complex and time consuming washing machine detection and extraction process. In case the installation under examination doesn't contain a washing machine, this process can be omitted, making the procedure more quick. Default Value = false

  4. Cleaning Point Of Interest: Before identifying the appliances from the measurements, there is a pre-process procedure to clean the events detected from small increases or decreases that may be attributed to noise and not in real appliances consumption. This procedure can be done in an Automatic or Manual way. Default Value = Automatic

  5. Cleaning Threshold: In case the user wants to proceed with Manual cleaning of the events, the threshold under which the fluctuations will be omitted must be provided. Default Value = 50

  6. Median Threshold: The event detection threshold, meaning the minimum value that the active power must have to surpass in order to detect a new event, is estimated by using the median of the minimum values appearing each day of measurements. If this option is false, then the mean value of the minimum values is used instead. Default Value = true

  7. Points Per Cluster: When analyzing large events, the points of interest are seperated in clusters based in their active power increase or decrease, in order to make the procedure more efficient. This option is adjusting the number of points allowed per cluster. Default Value = 10

  8. Combination Points: This option defines the number of rising points that can be combined with a single reduction point or the other way round. This is another way to reduce the complexity of some calculations and make the process more efficient and quick. Default Value = 6

  9. Time Threshold Complexity: This option is responsible for the characterization of the appliances type produced by the Disaggregation procedure. There are the following choices available:

  • Simple: In this case, the threshold for the distance between the points of interest connected (rising points with their appropriate reduction points) is set by the user (in this property file). The rest of the more distant points are removed by the procedure of matching.
  • Complex: In this case, the point selection is weighted according to the distance between the rising and the reduction points. Default Value = Simple

  1. Time Threshold: This option sets the time distance limit, in case of the Time Threshold Complexity is set to Simple. Default Value = 180

  2. Remove Large Events: This option is used when the user doesn't want to allow the events that have duration larger than a preset value. This may help to have a more clean and precise procedure, if for example you separate each day. Default Value = true

  3. Large Events Threshold: This option is used in combination with the option above, setting the duration threshold of the large events. Default Value = 1440

Add a custom footer
Add a custom sidebar
Clone this wiki locally