Development of a device-specific energy management for industry, trade, trade & service (IGHD) using NILM
Project Period: 2 Years (October 2015 – October 2017)


The disaggregation of the overall energy consumption of a single measuring point onto single facilities and appliances is an elegant alternative to submetering. The core idea of the NILM technology is based on the assumption, that every facility and every device influences current and voltage in a characteristic way and therefore leaving a “fingerprint” in the electricity grid. This “fingerprint” is recorded in form of the total aggregated energy consumption by measurement procedures such as Smart Metering. The individual characteristics are itemized by machine learning algorithms (NILM algorithms). This process is called disaggregation of the load. This method can be further classified in eventless and event-based approaches. As part of the event-based recognition a device is switched on/off and the algorithm learns its switching patterns. In eventless approaches the algorithm tries to disaggregate the load without the switching information by trying to find the best fit of the sum of different device fingerprints. As a result, detailed insight is gained by disaggregation, how high the individual energy consumption of devises is and which costs they cause. This way, energy-efficiency-potentials can be identified and device-specific cost-cutting measures can be defined – a really smart energy management.


Project Goals

The total result of the project is an energy management system, which breaks down the energy consumption into individual appliances and facilities for the energy manager, so that the device-specific energy consumption can be analyzed. A meta-study comprising 36 individual studies from North America and Europe has shown that saving effects of up to 12 % can be achieved solely in the private sector. In the industrial, trade and service sector even higher saving effects are expected. Within the project, the three branches from the industrial, trade and service sector with a high energy-saving-potential will be identified initially. The appliances and facilities which are deployed in these branches will be defined and measured. High-frequent, electrical characteristics of the appliances will be examined for their possibility of load-disaggregation subsequently. Reliable algorithms are developed ensuing thereon, which separate the individual device’s energy consumption from the total consumption. The developed measuring technique is being tested in a field test as a prototype and the actual savings potential is estimated.

The advantages of the project

  • Measurement and analysis of high frequency data
    • Analysis of different electric parameter besides the effective power
    • Only 1 measuring point instead of several submeter
  • Creation of new analysis possibilities
  • Identification of device-specific efficiency potentials


Has the air-conditioning been switched off? Has the fridge been left open?

This situation is known to most of us: You leave a building as the last person believing that everything has been switched off. But did you really remember everything? With the help of device-specific consumption analysis, deviations of typical consumption patterns are identified for every device – without elaborate assessment of causes. In combination with incident and time-based alarms, an alarm is set off immediately after the occurrence of a deviation, so that corrective actions can be taken in time.

Does my facility operate as it should or is a breakdown imminent?

Whether company-owned facilities or appliances are operating within normal parameters or a breakdown is imminent, that is of enormous economic importance. A malfunction shows up first in the device-specific consumption pattern, subsequently in changes of the total energy consumption.  For example, does the frequency band of a compressor change in normal operation; an impending loss of production can be avoided by taking counter measures in time. The alarmed employees can examine causes for the malfunction at an early stage. Is a handling error given? Is there maladjustment? Is the compressor defect? Furthermore, the appearance of cost-intensive load peaks can be avoided by timely intervention.

Which process causes the highest costs?

Individual appliances can, due to their utilization or installation location, be assigned to certain processes and applications. For example, a compressor and two engines could be part of a refinement process, five compressors and one engine for the process of assembly in an industrial enterprise. In gastronomy, the coffee machine and cash register can be associated with the sales area, whereas the dishwasher can be associated with the kitchen. With device-specific consumption analysis, appliances can be classified and assigned to processes and end-uses. Hence costs can be source-specifically apportioned, compared and processes and applications analyzed holistically. Efficiency measurements do not only focus on the exchange between individual, efficient appliances, but also on the energetic optimization of a whole process or a field of application with the related equipment park.

application areas

Energy Efficiency

Condition Monitoring

Process Optimization

This post is also available in: German