Pseudo-Value Generation for Low Voltage State Estimation With Poor Input Data From Smart Meter Using Gradient Descent Method

This paper presents a novel method for pseudo-value generation of household profiles based on poor input data, i.e. low number of measurements. In compliance with the current German law, for privacy reasons only voltage magnitudes from smart meters and measurements at the substation are considered. The gradient descent method is used to optimize the load power pseudo-values so that the power flow calculation results are more accurate. The results show a clear improvement over other methods such as e.g., probabilistic load profiles. The generated pseudo-values can be used to improve the low voltage state estimation accuracy compared to other approaches. This shall support the acceptance of state estimation-based monitoring and control systems in the future.