In this paper, focusing on the slow time-varying characteristics, a series of works have been conducted to implement an accurate quality prediction for batch processes. To deal with the time-varying characteristics along the batch direction, sliding windows can be constructed. Then, the start-up process is identified and the whole process is divided into two modes according to the steady-state identification. In the most important mode, the process data matrix, used to establish the regression model of the current batch, is expanded to involve the process data of previous batches, which is called batch augmentation. Thus, the process data of previous batches, which have an important influence on the quality of the current batch, will be identified and form a new batch augmentation matrix for modeling using the partial least squares (PLS) method. Moreover, considering the multiphase characteristic, batch augmentation analysis and modeling is conducted within each phase. Finally, the proposed method is applied to a typical batch process, the injection molding process. The quality prediction results are compared with those of the traditional quality prediction method based on PLS and the ridge regression method under the proposed batch augmentation analysis framework. The conclusion is obtained that the proposed method based on the batch augmentation analysis is superior.

In batch processing, not only the characteristics of different phases are different, but also there may be different characteristics between batches. These characteristics of different phases and batches will have different effects on the final product quality. In order to enhance the safety of batch processes, it is necessary to establish an appropriate monitoring system to monitor the production process based on quality-related information. In this work, based on multi-phase and multi-mode quality prediction, a new quality-analysis-based process-monitoring strategy is developed for batch processes. Firstly, the time-slice models are established to determine the critical-to-quality phases. Secondly, a multi-phase residual recursive model is established using each quality residual of the phase mean models. Subsequently, a new process-monitoring strategy based on quality analysis is proposed for a single mode. After that, multi-mode quality analysis is carried out to judge the relevance between the historical modes and the new mode. Further, online quality prediction is achieved applying the selected model based on multi-mode quality analysis, and an according process-monitoring strategy is developed. The simulation results show the availability of this method for multi-phase multi-mode batch processes.

This paper proposed a quality-related online monitoring strategy based on time and batch two-dimensional evolution information for batch processes. In the direction of time, considering the difference between each phase and the steady part and the transition part in the same phase, the change trend of the regression coefficient of the PLS model is used to divide each batch into phases, and each phase into parts. The phases, the steady parts, and the transition parts are finally distinguished and dealt with separately in the subsequent modeling process. In the batch direction, considering the slow time-varying characteristics of batch evolution, sliding windows are used to perform mode division by analyzing the evolution trend of the score matrix T in the PLS model on the base of phase division and within-phase part division. Finally, an online monitoring model that comprehensively considers the evolution information of time and batch is obtained. In a typical batch operation process, injection molding is used as an example for experimental analysis. The results show that the proposed algorithm takes advantage of mixing the time-batch two-dimensional evolution information. Compared with the traditional methods, the proposed method can overcome the shortcomings caused by the single dimension analysis and has better monitoring results.

In wind tunnels, the Mach number needs to be precisely monitored and controlled. It is difficult to obtain the Mach number directly online, especially when the wind tunnel system is operating in multiple modes. To deal with this problem, a Mach number prediction algorithm based on the kernel partial least squares method is proposed for multi-mode wind tunnel systems. First, in order to reflect real-time changes, the time-slice partial least squares regression method is adopted. Then, in order to enable the model to represent information about the whole working mode divided by key process variables, the mean-value partial least squares model is established and is compared with the time-slice model. Then, considering that wind tunnel systems exhibit strong nonlinear characteristics, the kernel partial least squares method, which is suitable for nonlinear systems, is used to predict the Mach number. The results show that the mean-value model is better than the time-slice model, the models for single modes show better prediction abilities than those for multiple modes, and the kernel partial least square method is more suitable for wind tunnel systems than the partial least square method.

Mach number, as an important index to judge the system performance in wind tunnel tests, its stability determines the quality of the flow field of this wind tunnel and needs to be controlled precisely. Due to the complex process in the wind tunnel test, it is difficult to ensure the smooth operation of the Mach number by the traditional control strategy, therefore, a Mach number prediction model based on a nonlinear autoregressive exogenous model (NARX)-Elman is proposed in this paper. Firstly, the NARX model is adopted as the basic framework of this model, the false nearest neighbor (FNN) is used for model order solving, and the Elman network is used for dynamic nonlinear fitting of the model. Secondly, considering the wind tunnel system with multiple conditions and the high cost of conducting complete data experiments, a new model migration method, the input-output slope/bias correction-genetic algorithm (IOSBC-GA), is proposed, which uses IOSBC method as the basic framework to construct the migration model based on the historical condition model and a small amount of data of the new condition, and uses GA to find the slope of deviation in the new model and correct the input-output relationship between the old and new conditions, so as to establish the model for the new condition. By comparing the model respectively with the traditional algorithm and the model built without migration, the root mean square error (RMSE) and the maximum deviation (MD) of this model are less than 0.001, indicating that the model has high prediction accuracy.