Electrical load curve reconstruction required for demand response using compressed sensing techniques

This work presents techniques for obtaining a reliable electrical load-curve based on comparative analysis between the different compressed sensing algorithms. Therefore, the goal is implementing compressed sensing (CS) when a wireless heterogeneous network, that exchanges information between electrical enterprise and smart meters, has a fault. Then, the data cannot be sent totally, and we would have the data only of some smart meters; thus, using the adequate technique of compressed sensing is possible to the reconstruction of load-curve required for generating demand response (DR) with the minimum error. In the advanced metering infrastructure (AMI) there may be communication faults; then, it is necessary to have other forms for estimating the demand response using few measurements. In addition, using a dictionary based on the DCT transform does not mean that the sea is the best option for the representation of a signal. For example, among other results, in this work we obtain an average of percent root mean square difference nearest to the 5% in relation with a Gaussian function or Wavelet basis with values between 1.4 and 1.7% average PRD.