Search

Though steady-state power features such as real power ($P$), reactive power (Q$), and total voltage/current harmonic distortions (V_{\text{THD}}/I_{\text{THD}}) may contain sufficient information, adopting them directly for nonintrusive aging load monitoring (NIALM) identification process requires higher computational burden and larger memory. To effectively reduce the number of needed steady-state power features without degrading the performance of NIALM, a Hellinger distance (HD) algorithm is proposed and presented in this paper. To minimize the training time and improve recognition accuracy, a particle swarm optimization (PSO) is adopted in this paper to optimize parameters of training algorithm in artificial neural networks (ANNs). The proposed methods are used to analyze and identify the load characteristics of aging loads in residential buildings. The results show that the accuracy can be improved by using the proposed method. [ABSTRACT FROM AUTHOR]

A Non-intrusive load monitoring (NILM) system is an energy demand monitoring and load identification system that only uses voltage and current sensors that are installed at the power service entrance of an electric system. The system is better than traditional intrusive monitoring systems because it is able to reduce the cost of sensors and installations. In this study, a real single-phase three-wire unbalanced 220 V/110 V distribution system model of a residential building is designed and implemented, and some non-intrusive techniques are executed in the Intel Atom Embedded System and a LabView program. To enhance the performance, the paper proposes using Particle Swarm Optimization (PSO) algorithm to optimize the parameters of a Back-propagation Artificial Neural Network (BP-ANN) for training steady-state power signatures such as real and reactive power (PQ). In this paper, the NILM system can identify some major appliances correctly in an unbalanced 220 V/110 V distribution system of a residential building. The real test identification accuracy can reach 100%. [ABSTRACT FROM AUTHOR]

Compared with the traditional load monitoring system, a nonintrusive load monitoring (NILM) system is simple to install and does not need an individual sensor for each load. Accordingly, the NILM system can be applied for wide load monitoring and become a powerful energy management and measurement system. Although several NILM algorithms have been developed during the last two decades, recognition accuracy and computational efficiency remain as challenges. To minimize training time and improve recognition accuracy, particle swarm optimization is adopted in this paper to optimize parameters of training algorithms in artificial neural networks. The proposed algorithm is verified through the combination of Electromagnetic Transients Program simulations and field measurements. The results indicate that the proposed method significantly improves recognition accuracy and computational efficiency under multiple operation conditions. [ABSTRACT FROM AUTHOR]