Value chain optimization control method based on deep learning

The invention provides a value chain optimization control method and system based on deep learning, and the method comprises the steps: obtaining a first input sample; determining a second input sample according to the first input sample based on a VAE model; constructing an enterprise value chain risk prediction model based on a deep learning technology and a space-time attention mechanism; inputting the second input sample into the enterprise value chain risk prediction model to obtain a probability distribution result of the enterprise value chain risk level; and performing optimization control according to a probability distribution result. According to the value chain optimization management and control method and system based on deep learning, the VAE model is introduced to perform data enhancement on the obtained first input sample, the second input sample is obtained, and the sample data is more balanced. A deep learning technology and a space-time attention mechanism are introduced, risk prediction i