A grid state estimation typically includes real measured values ​​and, if appropriate, supplemental pseudo-measured values ​​from models. In distribution networks, the estimation is based on underdetermined systems, there are fewer measurements available than nodes and edges in the network.

Figure 1 – Basic principle of grid state estimator

Grid state estimation and modeling takes place according to different approaches:

  • Extended profile creation for end customers via their own load profiles:
    • Here, measurement data with socio-economic data are blended in such a way that privacy compliant load profiles are generated based on the characteristics of customer groups. The models of customer groups can in turn be transferred back to grid connection points. Only a subset of real measurements is needed here.
  • Extended load profiles for network nodes and cable routes:
    • Powerful models are generated by simple statistical evaluations of measured grid nodes and cable routes. In the simplest case, these are similar in the format of VDE load profiles.
  • Statistical or machine-learning models of measured grid nodes or items such as cable routes (see Figure 2 – Predictive models):
    • More extensive modeling takes into account not only pure time-series information and load measurements for training (conventional load profiles), but also other input data such as external environmental data (e.g. temperature) or near-term measurements. The accuracy of the results of such forecasts is well within the range of a few percent.
    • Model updates (review and recalibration based on recent history) are performed on average every 6 to 12 months.

Figure 2 – Predictive models – basic structure


By continuing to use the site, you agree to the use of cookies. Further informationen

The cookie settings on this website are set to "Allow cookies" to provide the best browsing experience. If you use this website without changing the cookie settings or clicking "Accept", you agree.