The Load Identification Of Current Measurement System For Smarter Home Based On AMR Sensors

With the development of the construction of energy-saving society, people pay more and more attention to energy conservation and environmental protection. As a smarter home energy management system in energy-saving society which plays an important role. As the basement of energy management systems, the hardware development of current measurement system and the corresponding load identifications become more and more important.We design two type hardwares of non-invasive current measurement based on AMR sensor. One is a new sensor unit structure of current measurement and corresponding location algorithms, it only needs two axial anisotropic magnetoresistive sensors (AMR). Estimated location accurately by calculating the angle between the magnetic field strength and the wire position, using Kaiser window function to preprocess the sensor output voltage signal, Kalman filter is design for preprocessing signal to reduce the influence of the external magnetic field interference and noise, finally get the current waveform.which error is less than 1%. Through measuring the current flowing through the wire which located different positions on the sensor unit, and verifing the no calibration of location algorithm, also it can be applied in the occasion which wires location needed to be estimated. One is Non-invasive branch current measurement system based on ADSC of Singapore. The hardware can be very easy to paste on the switch box at home, it solves the problem of difficult and dangerous installation. We have optimized the hardware design at the same time so that reduce data processing circuit and reduced the difficulty of communication to ensure high-speed sampling. Finally it reach the purpose of each branch current can be estimated and measured. When AMR sensor measuring magnetic field which produced by the corresponded branch current, the AMR sensor is also influenced by the near magnetic field produced by the closed branch current. ADSC proposed a solution of switching the switch by worker to distinguish magnetic interference, but it is not convient for woker and user. Based on ADSC, we proposed automatic magnetic interference algorithm to solve the unconvient swiching behavior, the drawbacks of distinguishing magnetic interference is solved at last.We use the developed current measurement system to mesure the actual current value of the common electrical appliances and office appliances. The normalized amplitude spectrum and phase angle of harmonic current as a steady-state signal recognition feature, the detail coefficients power spectrum of normalized wavelet transform current as a transient signal recognition feature.The approximate coefficients power spectrum concepts is proposed based on detail coefficients power spectrum of wavelet transform, and the approximate coefficients power spectrum of wavelet transform current as stable signal recognition feature. Based on the above recognition feature, we use Matlab software programming to build BP neural network model and BP neural network structure is determined according to the error rate simulation. Under transient current of turnning on and turning off switches and stable current of appliance running situation, the load identification simulation is completed to verify the correction of identification feature and BP neural network.