Power Enterprise Asset Estimation Algorithm Based on Improved GoogLeNet

In order to promote the reform of power market, it is necessary to accurately evaluate the data assets of power enterprises. Considering the complex and massive power data, a improved GoogLeNet deep learning algorithm for power enterprises asset evaluation is proposed. According to the characteristics of input data, this paper uses two Inception modules as feature extraction network, which are obtained by improving the Inception of GoogLeNet. And then, the global average pooling is replaced by the full connection layer to gets the evaluation value. This paper uses a large number of convolution kernels of 1×2 and 2×1 to lower parameters and establishes more nonlinear transformations, which can enhance the learning ability of the network. Finally, the experiment proves that the algorithm in this paper can realize the accurate evaluation for the assets of power enterprises.