Quality prediction method for electric porcelain under sample expansion

Porcelain insulators are widely used in the power industry, and with the introduction of the ’Industry 4.0’ plan and the development of smart grids, higher requirements are being placed on the production of porcelain. In the actual manufacturing process of electric porcelain, the quality of the product is not high due to the complexity of the process, and the quality data of the production process is multidimensional and multi-characteristic, while the existing data samples are not enough, which makes it more difficult to accurately predict the quality of electric porcelain. For the above problems, there is not a complete solution. To this end, this paper proposes a method for predicting the quality of electric porcelain under the condition of sample expansion, which consists of two parts, first using Auxiliary Classifier GAN(ACGAN) to expand the dataset, then using an Attention mechanism and Bidirectional Long Short-Term Memory(BiLSTM) method to predict the qualification rate of electric porcelain products. After comparing multiple regression methods, our proposed method performs better.