Research On Non-intrusive Power Load Decomposition Method Based On Multiple Load Characteristics

With the economy growing rapidly,people are becoming more likely caring for the issues of ecological environment.Nowadays,how to make effective use of energy has become a topic which people are concerned on.While the energy monitoring is the basis for energy effective utilization.With the technology of Internet of things coming out,energy consumption monitoring is also developing towards the Internet of things.Based on this,a non-invasive method of load decomposition for embedded devices of ARM platform and a multi-load characteristic parameters which were applied to decomposition problems were studied.The selection of load characteristics were studied in this paper,according to the requirements of the load characteristics nature in representativeness,additivity and stability.The current harmonic amplitude value,the active power value,the reactive power value and the instantaneous current value referred in voltage peak were studied as the feasibility of the load characteristics.And then,the extraction method of four load characteristic values mentioned above were given.A method of pretreatment in load characteristics extracting were given through the studies of the influence of voltage fluctuation of power grid and harmonic fluctuations of power grid to the load parameters.The load characteristic decomposition model s were established in this paper,according to the voltage and current data collected by the sensors.The characteristic decomposition models which were consisting of the current fundamental and harmonic amplitude value,the active power value and the current second and higher harmonic amplitude value was discussed.To fully exploit each load characteristic benefits,the methods of evaluating function based on multi-load characteristic were proposed.The algorithm of improved particle swarm optimization(pso)which was suitable for the decomposition model in this paper was studied.The disaggregation result was obtained by solving the decomposition model while processing with the algorithm of improved pso.The decomposition models based on active power,the fundamental and harmonic amplitude and the combine form of them were compared in accuracy in this paper.The problems that the algorithm of improved pso was suitable for the noninvasive electricity load decomposition problem and the multi-load characteristic evaluation function had a better performance in accuracy were verified by some examples.The enumeration method which is a kind of deterministic algorithm and the algorithm of improved pso which is a kind of heuristic algorithm were compared in the performance of solving the decomposition mo del.The experimental comparison showed that the algorithm of improved pos had more advantages in load decomposition business while the enumeration method had obvious disadvantages.For low sampling frequency,a multi-factor decision method based on the pr ocess of disaggregation and verification was proposed according to the experiment.The accuracy of disaggregation result was improved by the multi-factor decision method under the conditions of low sampling frequency.