Load Identification Method Based on ISMA-GRNN

The noninvasive load monitoring method carries out load identification after event detection and feature extraction of load data. At present, nonload intrusive load monitoring faces the problems of low load identification accuracy and long load identification time. In order to solve these problems, a load identification method based on the improved slime mould algorithm-generalized regression neural network (ISMA-GRNN) is proposed. Firstly, by adding mutation operation in slime mould algorithm (SMA) position update, the global optimization ability of SMA is improved. Then, the improved slime mould algorithm (ISMA) is used to optimize the smoothing factor of GRNN and find the best smoothing factor. Finally, the best smoothing factor is input into GRNN for load identification, and the load identification results are output. To measure the effect of load identification, load identification precision, load identification accuracy, and load identification time are used as evaluation indicators. The simulation results show that compared with HHO-GRNN and WOA-GRNN, the load identification time of SMA-GRNN is greatly shortened, but the results are not satisfactory. On the basis of SMA-GRNN, ISMA-GRNN has significantly improved the accuracy and precision of load identification. In conclusion, ISMA-GRNN can better adapt to the load identification of multiple electrical equipment scenes.