Online Intelligent Recognition of Load Based on Multiorder Active Bandpass Filter and Online Correlation Coefficient

In this paper, a mathematical model of load online recognition problem is established in time domain, and the recognition task is shared by hardware and algorithm. A multiorder active band-pass filter is used for hardware preprocessing to convert the load signals with large power difference into voltage signals that can be collected under the same condition. The online correlation coefficient is defined from the Pearson correlation coefficient as the only time domain characteristic parameter, and the judgment criterion of "online correlation coefficient gradually increases and eventually equals 1" is proposed. The maximum first alignment method and the mean method are used for data processing, and the successive comparison method and the screening method are combined for recognition. The experiment results show that based on a fifth-order active band-pass filter with the bandwidth of 340 Hz to 900 Hz and online correlation algorithm in STM32 processor the system can well complete the intelligent recognition of any online combination of a power difference of more than 1500 times (i. e. induction cooker, rice cooker, laptop, and small light bulb).