基于多阶有源带通滤波器和在线相关系数的负载在线智能识别

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

从时域建立了负载在线识别的数学模型,识别任务由硬件和算法共同承担。采用多阶有源带通滤波器进行硬件预处理,从而将功率相差很大的负载信号转换成可以在同一个条件下采集的电压信号。从皮尔逊相关系数出发定义了在线相关系数,将其作为唯一的时域特征参量,并提出“在线相关系数逐渐增大并最终等于1”的判决准则。利用最大值首位对齐法和均值法进行数据处理,采用逐次比较法和筛选法相结合进行识别。实验数据表明:基于带宽为340~900 Hz的五阶有源带通滤波器和STM32处理器中的在线相关算法很好完成了功率相差1500多倍的电磁炉、电饭煲、笔记本电脑、小灯泡等4种负载的任意在线组合的智能识别。

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).