Data-Driven Soft Sensor Model Based on Deep Learning for Quality Prediction of Industrial Processes

Abstract

Fermentation process is a time-varying, nonlinear and multivariable dynamic coupling system. Therefore, it is difficult to directly measure the key biological variables using traditional physical sensors during the process of fermentation, which makes the monitoring and real-time control impossible. To resolve this problem, a data-driven soft sensor modeling method based on deep neural network (DNN) is proposed in this paper. This method is suitable for large amount of data and it enjoys high efficiency and robustness. At the same time, an adaptive moment estimation (Adam) algorithm is used to optimize the hyper-parameters of the DNN model, which is a technique for efficient stochastic optimization that only requires first-order gradients with little memory requirement. The consistent correlation method is used to determine the auxiliary variables of the soft sensor model. The penicillin and l-lysine fermentation processes are taken as the research object, substrate concentration, cell concentration, and product concentration are selected as a target variable. The performance of established soft sensor model is evaluated through the indexes of mean square error (MSE), root-mean-square error (RMSE), and mean absolute error (MAE). The simulation results show that the prediction performance of the soft sensor model based on DNN-Adam is good and compared with model based on stochastic gradient descent (SGD) with momentum optimization algorithm. It is verified that the proposed method can make a more accurate real-time prediction of quality variables in the fermentation process, and it has higher prediction accuracy than DNN-SGD method.

Application of Neural Network Algorithm in Propylene Distillation

“…Soft measurement technology has been widely used in process control systems. With the development of production technology and the increasing complexity of production processes, it is necessary to control important process variables of the system optimally in real time in order to ensure the safer and more efficient operation of production devices [41].…”

Weighted target feature regression neural networks based soft sensing for industrial process

“…[4] For traditional manual inspection methods, cost-effective online measurements are lacking to produce high quality products, and therefore traditional mechanistic models are difficult to accurately describe real industrial processes. [5,6] These problems pose a great challenge for accurate online measurement of key variables in process industries. [7] Based on the above problems, many scholars have searched for effective modelling and prediction methods for industrial processes from the perspective of the complex background of process industries.…”

“…[ 4 ] For traditional manual inspection methods, cost‐effective online measurements are lacking to produce high quality products, and therefore traditional mechanistic models are difficult to accurately describe real industrial processes. [ 5,6 ] These problems pose a great challenge for accurate online measurement of key variables in process industries. [ 7 ]…”

Defect rate prediction and failure‐cause diagnosis in a mass‐production process for precision electric components

“…Based on the predicted values, anomalies can be detected and processes can be controlled. Regression analysis methods include linear methods, such as partial least squares regression 11 and least absolute shrinkage and selection operator, 12 and nonlinear methods include support vector regression, 13 Gaussian process regression, 14 deep neural networks, 15 random forests (RF), 16 the gradient boosting decision tree (GBDT), 17 XGBoost (XGB) 18 and LightGBM (LGB). 19 Adaptive soft sensors 20 have been developed to adapt to the changing plant process conditions.…”