Non-intrusive Load Monitoring Based On Transient Characteristics

Load monitoring is very important for power system. In conventional loadmonitoring systems, the hardware circuits are installed on each load to be monitored.This method takes a lot of time and money in the installation and maintenance andcan not meet the challenges of the future power systems. Non-intrusive loadmonitoring (NILM) system only needs to install monitoring devices on the electricpower entrance and can identify the load class and operation conditions of theindividual load with the measured data. NILM is convenient and will save time andbudget in installation and maintenance. NILM has attracted much more attention inrecent years. With NILM, the utility can obtained load compositions and improve theaccuracy for load forecasting and modeling in plan stage and dynamic simulation.For the customers, NILM can help them to reduce the energy costs and also reducethe demand for fossil fuels.Non-intrusive load monitoring (NILM) system has a simple hardware andcomplex software. The non-intrusive load monitoring and analysis methods fall intotwo categories, namely the steady signatures based methods and transient signaturesbased methods. The transient response of the load is closely related with its physicaltasks. Different classes of loads have unique signatures during the turn on/off event.With these signatures, NILM can obtain the compositions of the cluster loads in thestation and building and it overcome the limits of the algorithms with steadysignatures. The NILM should be able to detect the transients automatically andtrigger the related program to save the transient data for further analysis. In thisthesis, a sliding window version two-sided CUSUM based transient event detectionalgorithm is proposed. With the measured signals such as voltage, current or activepower, the algorithm can detect the transients and trig the related program to recordthe transient events. Then these data can be used for further analysis for loadidentification.In addition, the NILM is extended to three phase power grid with high voltage level and a cross correlation function base load identification algorithm is proposed.The transient signatures of the loads are obtained by synchronous dq transformation.And then the signatures are compared with the load library to decide the load classand operation conditions.Digital simulation has been used to demonstrate the feasibility of the proposedalgorithms and the results show they can be applied in non-intrusive load monitoringsystem.