Double-layer-order product quality prediction method based on residual error correction

In traditional production management, the hysteresis quality of quality prediction may cause a large number of unqualified products. Therefore, the invention provides a double-layer-order product quality prediction method based on residual error correction, and the method comprises the steps: firstly predicting the processing parameters through a random forest algorithm, and guaranteeing the parameter integrity; secondly, analyzing parameters by using a regression model constructed by combining a genetic algorithm and a fully connected neural network (NSGA-FCNN), and predicting quality features and residual errors; in order to solve the problem of low prediction precision, residual error correction is carried out by adopting residual error analysis to train an NSGA-FCNN model. Finally, a product quality prediction result and a residual error correction result are combined to form a double-layer-order product quality prediction method. The quality prediction value obtained through the method is compared with t