Journal of Process Control最新文献

Two-dimensional control has been considered as an effective strategy to accomplish high-accuracy tracking for batch processes because of its excellent learning ability and time-domain stability. However, being a model-based control method, the performance of the two-dimensional control system will inevitably decrease due to unknown uncertainties or unmodeled dynamics. In addition, the high computational cost and complex design process of the control system severely limit its application in batch processes. For this reason, this paper proposes a new data-driven two-dimensional integrated control (DDTDIC) method for nonlinear batch processes. In the presented control scheme, the P-type iterative learning control (ILC) is adopted along the batch-axis to ensure the convergence of the system, and the proportional-integral-differential (PID) control is used in the time-axis to reject the influence of real-time disturbance. The parameters of the PID controller are obtained by utilizing the virtual reference feedback tuning (VRFT) method. The entire design process of the control system only requires the input and output (I/O) data of the batch processes and does not depend on any explicit model information. The simulation results show that compared with the ILC and the two-dimensional control, the presented control method not only has faster convergence speed and smaller tracking error, but also the computational efficiency is improved by more than 40% and 50% respectively.

Industrial data-driven models may require frequent reconstruction to maintain model performance due to the dynamics, uncertainty, and complexity of industrial processes. The infrastructure of the industrial processes is usually distributed control systems (DCS) with energy-sensitive and resource-constrained. In this context, this article proposes a geometric constructive network with block increments (BI-GCN) to reduce the modeling consumption while achieving comparable accuracy. First, this article proposes a geometric control strategy with block increments, which is capable of adding multiple nodes to the BI-GCN simultaneously. Second, this article demonstrates the universal approximation property of BI-GCN, which in turn guarantees the potential high performance of BI-GCN for modeling tasks. Finally, experiments on benchmark datasets and the grinding process show that BI-GCN can effectively reduce the number of iterations in the modeling process while maintaining comparable accuracy.

For chain interconnected systems with state and input constraints, a partial output consensus (POC) optimization problem is studied when the set-points are infeasible. In this case, outputs with and without consensus requirements cannot converge to the set-points achieved from real-time optimization. For this case, a novel set-point optimization method is developed, which is called distributed partial output consensus optimization. Based on this method, the set-points for two-part outputs i.e. having a part that must achieve consensus and a part that has a set-point, can be recalculated simultaneously and their feasibility can be ensured by using a distributed projection operator. The convergence of the strategy is then analyzed. From the results of both simulation and experimental testing, the effectiveness of the proposed method is validated.

During coal-fired power generation, uniform combustion temperature in the boiler is desired which will benefit both economical efficiency and pollution reduction. To this end, a model predictive control (MPC) algorithm based on the Nonlinear Auto-Regressive Exogenous Inputs (NARX) neural network and KS-function is proposed, and the uniform combustion in the boiler is realized by controlling the opening travel of secondary windgates. In the modeling process, a multi-input and multi-output(MIMO) NARX neural network is developed using the historical data of the real system The NARX neural network is then used to predict the state variables, and the optimal control input is achieved by applying sequential quadratic programming (SQP), comparing with linear MPC the mean temperature difference is reduced by 64.2%. In addition, this paper proposes a new method to reduce the computational time of the online optimization process based on KS-function, which greatly accelerates the searching speed of SQP by 67.3%. The proposed MPC algorithm is applied to a 660 MW power generating unit. The results show that by applying the proposed algorithm, the temperature difference in the boiler is kept within 100 °C, the average coal consumption of the power plant is reduced by 5.71 g/kWh, and the NOx emission is reduced to 23.84 mg/m₃. It can be concluded that the proposed algorithm greatly improves the economical efficiency of the power plant and reduces the emission of pollutants.

Typically, failure time is modeled using continuous distributions such as the Weibull or Gamma distributions. In many practical scenarios, data is recorded in terms of discrete counts, such as the number of days or cycles, therefore the Discrete Weibull distribution is employed to model such cases. In this paper, we propose the use of a Shewhart $\overline{X}$ control chart to monitor the mean of a Discrete Weibull process. While the distribution of the sum of Discrete Weibull random variables does not have a closed-form expression, it can be determined through a Markov Chain procedure, which enables the calculation of precise control limits. The Average Run Length ($ARL$) is the metric used to assess the performance of the control chart. Two numerical examples are provided to illustrate its practical application.

With the recent advancement of products pipelines digitization, a large number of sensors have been installed in pumping stations for real-time flow parameters measurement. In these asynchronous multi-sensor systems, data missing and false data attacks are likely to occur when performing online operation monitoring of the oil pipeline system. In this paper, a hybrid state estimation method is proposed to process both the missing and fault measurement, considering the dynamic operation process of the whole system. Combing frequency-domain analysis method with model-free adaptive control algorithm, the state estimation model with adaptive deviation compensation is established to characterize the nonlinear transient flow process of the pumping station. And the Kalman Filter method is adopted to overcome the interference of sensor noise. In terms of multi-rate observation data processing, this study innovatively proposes an algorithm based on the first principle and generalized predictive control theory to improve the accuracy of traditional missing data processing methods based on statistical analysis. Moreover, non-obvious abnormal observations are identified by introducing long short-term memory network characterized by deviations between sensor measurements and multi-rate state estimation results. To verify the effectiveness of proposed method, it is adopted to the unsteady state estimation of a refined oil pumping station system under the attack of noise, nonuniform asynchronous sampling and insignificant abnormal data.

Following the recent advent of Process Analytical Technologies, dataset production has undergone significant leverage. In this new abundance of data, isolating meaningful, informative content is critical for process dynamic modeling. This paper proposes a data-driven algorithm based on low-rank matrix approximation, the so-called successive projection algorithm, to retrieve a minimal set of macroscopic reactions, the corresponding stoichiometry, and a consistent kinetic model structure from the measurements of the trajectories of the species concentrations during cultures in a bioreactor. The proposed method is successfully validated in simulation, considering a case study related to monoclonal antibody (MAb) production with hybridoma cell cultures.

High-rate thickeners are used in the mining industry to improve water recovery from slurries and increase their solids ratio. High-rate thickeners operate under strict constraints and several disturbances. To control this process, a constrained model predictive control (MPC) is developed in this paper. For process identification, a historical data-driven methodology is used and a vector autoregressive with exogenous variables (VARX) model structure is considered. The model takes underflow slurry density as both a state variable and the process output, along with turbidity, bed level, rake torque, and cone pressure as additional state variables. It takes feed slurry and flocculant flow rates as manipulated inputs and considers inlet slurry density, slurry circulation flow rate, and underflow slurry flow rate as disturbances. The VARX model structural parameters (orders and delays) and coefficients are estimated using a bilevel optimization method. From the model obtained, a discrete state-space representation is derived. This latter is augmented to obtain a standard formulation without delays. The MPC is then formulated considering the process constraints. To evaluate the control performance, simulations are conducted and a baseline comparison is established using proportional integral (PI) control. Simulation results demonstrate that the proposed control method outperforms the baseline method by providing reduced settling times (−32%), minimized peak errors (−20%), and constraints handling ability. Accordingly, the proposed MPC is implemented in an industrial environment and compared to existing manual control based on an object linking and embedding (OLE) for process control (OPC) architecture. Finally, the industrial results show that the proposed control method effectively stabilizes the underflow slurry density and handles process constraints, resulting in a minimized average error (−90%) and a reduced standard deviation (−50%) compared to existing manual control.

Nonferrous metals are indispensable raw materials for modern industry. The price forecasting of nonferrous metals is vital for business operators and investors. Based on the decomposition-integration framework, we propose a signal decomposition model combining variational mode decomposition (VMD) and an improved long-short time memory (LSTM) network. Using the MAE metric as a benchmark, the improved LSTM model (Mogrifier LSTM) obtained an average accuracy improvement of 5.99%. VMD is an efficient decomposition algorithm. However, it needs to set hyperparameters in advance. Unreasonable parameters will lead to poor decomposition results. Therefore, a method based on subseries complexity and reconstruction error (CAE) is proposed to reasonably decompose signals, improving 21.13% accuracy and reducing 37.56% computational overhead than other strategies. The structural model is introduced as a complement to the signal decomposition model, which learns different features by incorporating theoretical analyses into the choice of explanatory variables. The combining of two models achieves effective complementarity, obtaining an average accuracy improvement of 7.43%. Comparative tests on three datasets demonstrate the superiority of the proposed prediction framework. On the one hand, a reasonable decomposition strategy can play an essential role in the signal decomposition model. On the other hand, improving the prediction model and integrating different models is also an effective strategy to enhance accuracy.

This paper focuses on the problem of crude oil operations scheduling carried out in a system composed of a refinery and a marine terminal, considering uncertainty in the arrival date of the ships that supply the crudes. To tackle this problem, we develop a two-stage stochastic mixed-integer nonlinear programming (MINLP) model based on continuous-time representation. Furthermore, we extend the proposed model to include risk management by considering the Conditional Value-at-Risk (CVaR) measure as the objective function, and we analyze the solutions obtained for different risk levels. Finally, to evaluate the solution obtained, we calculate the Expected Value of Perfect Information (EVPI) and the Value of the Stochastic Solution (VSS) to assess whether two-stage stochastic programming model offers any advantage over simpler deterministic approaches.