Search

This paper proposes a tube-based model predictive control strategy for linear systems with bounded disturbances and input delay to ensure input-to-state stability. Firstly, the actual disturbed system is decomposed into a nominal system without disturbances and an error system. For the nominal system, solving an optimization problem, where the delayed control input is set as an optimization variable, yields a nominal control law that enables the nominal state signal to approach to zero. Then, for the error system, the Razumikhin approach is used to identify a robust control invariant set. Using the set invariance theorem, an ancillary control law is developed to confine the error state signal in the invariant set. Combining the two results, we obtain a control law that enables the state signal to remain within a robustly invariant tube. Finally, the effectiveness of the developed control strategy is validated by simulations. • An ancillary control is used to ensure the state errors stay in the invariant set. • A constraint on the delayed input is added to ensure feasibility of the algorithm. • By introducing a state error term, a novel Lyapunov function is proposed. [ABSTRACT FROM AUTHOR]

Anode effect (AE) is influenced by adaptive spatial and temporal factors in the aluminum electrolysis process. Anode current signals (ACS) are the only online distributed signals which provide spatial temporal characteristics of AE. How to extract interpretability adaptive spatial and temporal characteristics of ACS for AE prediction is a challenging problem. In this paper, a multi-generator adversarial dynamic spatial–temporal shapelet network is proposed to capture the explainable spatial–temporal characteristics. Multi-generator adversarial training added into shapelet learning is used to reinforce the interpretability of shapelets. A dynamic distance calculation strategy is proposed to extract dynamic shapelets of multi-variable time series whose dynamic discriminative subsequence is not restricted to dimension. Based on the proposed self-regulative spatial network and diversity regularization, the extracted shapelets will have adaptive spatial networked correlations and diversity. Moreover, the proposed method can guide the domain technicians to locate the abnormal conditions. The results of experiment using real-world aluminum electrolysis plant data show that the proposed method is feasible and quite effective to detect anode effect earlier and enhance the interpretability of shapelets. [ABSTRACT FROM AUTHOR]

Unknown abnormal working condition discovery is the key of refinement industrial production. Clustering industrial time series is an effective way to discover unknown working condition types. However, it is challenge for existing time series cluster methods to discover unknown abnormal working condition from industrial time series. In this study, a novel prior knowledge-augmented unsupervised shapelet learning method is proposed to discover abnormal and meaningful working condition through interpretable subsequences. A prior feature extracting module is proposed to change prior knowledge into a recognizable form for the data model. The prior knowledge contains abnormal working condition information. The knowledge-augmented clustering module can learn informative shapelets which stand for abnormal working condition by combining prior features with data features. Furthermore, the preference of prior knowledge and data are self-adjusted in the learning phase. Numerical test results on the real-world aluminum electrolysis process, simulated Tennessee Eastman process, and continuous stirred tank heater process verify the superior performances of the proposed method. The proposed method provides a new perspective for the fusion of prior knowledge and data model. It also provides a new way to solve the problem of abnormal unknown working condition discovery in industrial process. [ABSTRACT FROM AUTHOR]

Failure Mode and Effect Analysis (FMEA) is a prospective and systematic analytical tool widely used to improve the reliability and safety of various industries. However, the existing FMEA methods have received criticism for their inherent deficiencies that limit their flexibility and effectiveness. To address these issues, we propose a novel FMEA model that combines Probabilistic Free Double Hierarchy Hesitant Linguistic Term Set (PFDHHLTS), Extended Grey Relation Analysis (EGRA), and ORESTE to determine the priorities of Failure Modes (FMs). Firstly, the PFDHHLTSs are proposed to represent experts' assessments, improving the flexibility and accuracy of experts' expressions. Secondly, the PFDHHLTS-EGRA is proposed to determine the evaluation matrix and the weights of risk factors in a large group environment. Furthermore, the ORESTE method is extended to the PFDHHLTS environment. Lastly, we verify the effectiveness of our method through a case study on the aluminum electrolysis process. The results demonstrate the effectiveness and flexibility of our method in expressing experts' assessments and obtaining more accurate and reliable risk priorities. [ABSTRACT FROM AUTHOR]

• An antimony grade prediction model based on CapsNet is proposed for performance monitoring of the flotation process. • An improved DBSCAN algorithm is proposed to denoise the training data set. • The use of CapsNet can solve the problem of the small amount of training data. This paper presents an application of the capsule network to predict the antimony grade of pulp in the roughing cell of an antimony flotation plant in the Hunan Province, China. In this plant, because the chemical testing for analyzing the antimony grade only generated eight data points every day, data could be collected in small amounts and were mixed with some abnormal images. An improved density-based clustering algorithm is introduced to eliminate abnormal images from the training dataset. To use a small amount of data, a capsule network rather than a CNN is adopted to build the recognition model named Froth-CapsNet. Finally, the application of Froth-CapsNet to monitor the working conditions of the antimony flotation process indicates that this model can provide a guide for operators to precisely adjust the dosage of flotation reagents in real-time so that the antimony recovery rate can be improved. [ABSTRACT FROM AUTHOR]

This paper considers an optimal control problem for cascaded irrigation canals. The aim of the optimal control is to guarantee both the minimum water levels for irrigation demands and avoidance of water overflows even dam collapse. Due to the structural complexities involving control gates and interconnected long-distance water delivery reaches that are modeled by the Saint-Venant PDEs with conservation laws, wave superposition effects, coupling effects and strong nonlinearities made the optimal control be a hard task. A nonlinear optimal control method is proposed to deal with the PDE-constrained optimization problem via a control parameterization approach. Control parameterization approximates the time-varying control by a linear combination of basis functions with control parameters. The Hamiltonian function method is used to derive the gradients of the objective function with respect to the control parameters as well as the time scale parameters for providing the search directions of the optimization problem with acceptable amount of computations. Based on the gradient formulas, a gradient-based optimization algorithm is proposed to solve the optimal control problem. The proposed nonlinear optimal control method is validated in two cases: a single reach canal in Yehe Irrigation District in Hebei Province (China) and a cascaded two-reach canal system. [ABSTRACT FROM AUTHOR]

Due to the scarcity of the labeled data, traditional supervised learning methods have a limited application scope, which caused the supervised-based model performance will greatly be decreased. In this paper, we propose a promising model based on self-supervised learning. To update the weight and the contrastive relation in the features, a new self-supervised loss, is introduced. First, the convolution neural network is used in the proposed network to extract the deep feature in the first processing. Second, the self-supervised long–short time memory (LSTM) sequential is constructed for further deal. At last, the teacher net and student net have coordinately fine-tuned the credibility of the temperature prediction. By the experimental comparison, our proposed CNN-SSDLSTM is competitive with other supervised and semi-supervised methods. The evaluation experiments achieve state-of-the-art performance in aluminum electrolysis temperature prediction applications. [Display omitted] • A novel self-supervised architecture for temperature identification is proposed. The model achieves great accuracy with the constraint of the scarcity of labeled data. • A new self-supervised loss is used in the proposed architecture. This loss can fully utilize the information contained in the unlabeled data, and cucumber the limit of the label data. • A performance strategy for contrastive learning is used in the training of the model. The dual net of a student and teacher net is used to dilate the knowledge of the cosine loss. [ABSTRACT FROM AUTHOR]

Data-driven soft modeling has been extensively used for industrial processes to estimate key quality indicators which are hard to measure by some physical devices. However,the existing deep soft methods faces the challenge of training efficiency, gradient diminishing and explosion. Constructing an accurate and robust soft model is still a challenging topic from an application point of view. This paper develops an effective and efficient soft method (SAE-WELM) for processes modeling. First, a stacked autoencoder (SAE) is used to extract the deep features. Then, a top-layer extreme learning machine (ELM) is further applied to a plant-wide industrial aluminum production process. The activation function is wavelet kernel. Finally, the approximation and convergence of the proposed SAE-WELM are theoretically proved. The industrial case demonstrates that SAE-WELM captures the deep features faster than other iterative-based neural networks, and the accuracy and robustness outperform the existing state-of-the-art methods. • The traditional deep architectures contain the issue of gradient diminishing and exploding. To improve the training efficiency and generalization ability, a novel deep soft model based on ELM with a wavelet kernel function is proposed. • A data-driven soft model was developed which can preserve the universal approximation ability and extract the deep and complex features in-process data. In contrast to the Gaussian kernel, the proposed approach combines the wavelet kernel in ELM, the generalization ability and training speed can be greatly improved. • To obtain a stable and applicable model, the approximation ability and convergence of the proposed SAE-WELM are proved by constructing a Lyapunov function. [ABSTRACT FROM AUTHOR]

Koopman operator with numerical approximation method for modelling nonlinear systems has become a popular data-driven approach in the past five years. However, when the system contains time-varying parameters, the data-driven Koopman operator-based model produces deviations between the nominal model and the true one. It affects the control performance when it serves as the nominal model in Model Predictive Control (MPC). To solve this issue, the Koopman operator-based model learned by using a multi-step prediction autoencoder and the strategy of online updating model are proposed. In the first step, a neural network with encoder-decoder structure is trained to search for the optimal lifted functions of the corresponding nonlinear system with fixed parameters. In the second step, an online updating strategy is presented to update the Koopman operator-based model using the collected data, when the nominal model is applied in the closed loop operation of MPC. In order to reduce the additional time consumption on updating the nominal model, the matrix inversion lemma is applied and it leads the strategy of updating to be presented in a recursive form. In general, this paper proposed an iterative MPC algorithm for online updating the Koopman operator-based model in a recursive form based on the optimal lifted function using autoencoder. Numerical simulations on three nonlinear cases with time-varying parameters show good performances on tracking control using the proposed Koopman operator-based model using autoencoder and online updating strategy. • A method based on autoencoder to learn the lifted functions was presented to obtain the approximation of Koopman operator. • The learned nominal model shows good prediction on evolution behaviors of nonlinear systems. • Adaptive model predictive control (MPC) was proposed to address the model mismatch on parameter-varying nonlinear systems. • The proposed Koopman operator-based Adaptive MPC can achieve high computational efficiency and good control performances. [ABSTRACT FROM AUTHOR]

In practice, because of complex mechanism processes, such as heating process, volume heterogeneity, and various chemical reaction characteristics, there is a nonlinear relationship among variables in industrial systems. The nonlinearity brings some difficulties to process monitoring. In order to ensure that the process monitoring system can work normally in nonlinear production processes, the nonlinear relationship between variables ought to be considered. In this work, a new fault detection and isolation method based on kernel dictionary learning is presented. In detail, the linearly inseparable data is mapped to a high-dimensional space. Then, a new nonlinear dictionary learning method based on kernel method was proposed to learn the dictionary. After obtaining the dictionary, the control limit can be calculated from the training data according to the kernel density estimation (KDE) method. When new data arrive, they can be represented by the well-learned dictionary, and the kernel reconstruction error can be used as a classifier for process monitoring. As for the fault data, the iterative reconstruction based method is proposed for fault isolation. In order to evaluate the effectiveness of the proposed process monitoring method, some extensive experiments on a numerical simulation, the continuous stirred tank heater (CSTH) process, and a real industrial aluminum electrolysis process are conducted. The proposed method is compared with several state-of-the-art process monitoring methods and the experimental results show that the proposed method can provide satisfactory monitoring results, especially for some small faults, thus it is suitable for process monitoring of nonlinear industrial processes. • A kernel dictionary learning method is proposed for nonlinear process monitoring. • The proposed method can avoid solving the nonlinear optimization problem. • The effectiveness of the method is illustrated by three experiments. • The method can achieve fault isolation, which is meaningful for industrial system. [ABSTRACT FROM AUTHOR]

Belief rule-based classification system (BRBCS) is a useful model to handle classification problems. In our previous work, a novel data-driven BRBCS with batch-by-batch observation, online learning and multi-weights (BRBCS-BOM) is proposed. It can powerfully obtain knowledge from data. However, in some applications of industrial engineering, it can be hard to obtain enough training samples. The knowledge contained in data is not enough to deal with the problem well. It is necessary to adopt expert-driven BRBCS as an important supplement (hybrid-driven). In this paper, a hybrid-driven BRBCS-BOM with expert intervention (HBRBCS-BOM/E2) is proposed. It adopts a modified inheriting-and-learning hybrid-driven mode. Generally, it inherits the belief rules generated from training samples and make a secondary optimization for these inherited belief rules based on learning from expert-driven BRBCS. Moreover, a novel mode of expert intervention is proposed, based on the reliability evaluation. It can diversely-and-precisely obtain some important online new training samples for enhancement of data-knowledge, making hybrid-driven model better. The related experiments on an industrial engineering classification problem, called abnormity recognition of synthetical balance of material and energy, have demonstrated that the proposed HBRBCS-BOM/E2 not only makes an effective improvement for data-driven BRBCS-BOM from above two ways, but also has a more advanced performance compared with other existing high-performance BRBCS. • A hybrid-driven mode is proposed based on swapping the inheriting-and-learning. • A mode of expert intervention is proposed based on the reliability evaluation. • Proposed model can handle the abnormity recognition in electrolytic cell well. [ABSTRACT FROM AUTHOR]

In many industrial processes, especially chemistry and metallurgy industry, the plant is slow for feedback and data test because of complex and varying factors. Considering the multi-objective feature and the complex problem of production stability in optimal control, this paper proposed an optimal control strategy based on genetic programming (GP), used as a multi-step state transferring procedure. The fitness function is computed by multi-step comprehensive evaluation algorithm, which provides a synthetic evaluation of multi-objective in process state based on single objective models. The punishment to process state variance is also introduced for the balance between optimal performance and stability of production. The individuals in GP are constructed as a chain linked by a few relation operators of time sequence for a facilitated evolution in GP with compact individuals. The optimal solution gained by evolution is a multi-step command program of process control, which not only ensures the optimization tendency but also avoids violent process variation by adjusting control parameters step by step. An optimal control system for operation direction is developed based on this strategy for imperial smelting process in Shaoguan. The simulation and application results showed its effectiveness for production objects optimization in complex process control. [Copyright &y& Elsevier]