A non-intrusive load monitoring and identification system based on cloud platform

In order to monitor the load′s running condition and identify the load′s type in real time and long distance, a non-intrusive load monitoring system and a load identification algorithm based on PCA and kNN are designed and developed. On the side of power supply inlet, operating current is sampled through putting a series constantan resistor in the load circuit and operating voltage is sampled through a resistive subdivision network. Thus the load′s real-time active power is calculated and uploaded to the cloud server on the frequency of 1 Hz. On the side of cloud server, features extraction and dimensionality reduction are processed by PCA. The running load is classified by kNN. Users can visit the load monitoring interface by terminal devices. In the experiments, the system is installed in the wall socket to monitor and identify eight types of household appliances. Multiple experimental results show that the average rate of identification is above 98%, which verifies that the method proposed is accurate and