Industrial users load pattern extraction method based on multidimensional electrical consumption feature construction

The rapid development of renewable energy generation aggravates the imbalance between supply and demand in power grid, exploring potential side resource can effectively improve such problems. Industrial users (IU) is an important response mining load patterns IU basis studying ability IU, which plays role safe operation lean management grid. Lately, popularity advanced metering infrastructures provides data support for IU. However, high dimensionality complex non-linear relationship IU’s bring difficulties to task clustering. To solve above problems, this paper proposes a pattern extraction method based on multidimensional electrical consumption feature construction. Firstly, industrial characteristic set created with five indexes weighted by improved entropy weight method. In addition, convolutional autoencoder established extract temporal combined build (MFS) finish construction (MECFC). Then, MFS used as input Self-Organization Map network select initial clustering centers K-means algorithm, overcoming problem local optimal solution, complete daily experiment shows that algorithm MECFC solves have better performance stability effect than traditional methods.