Chao Shang

Verified

Chao verified their affiliation via an institutional email.

Learn more

Verified

Chao verified their affiliation via an institutional email.

Learn more

• Ph.D.
• Professor (Associate) at Tsinghua University

Scenario programs have established themselves as efficient tools towards decision-making under uncertainty. To assess the quality of scenario-based solutions a posteriori, validation tests based on Bernoulli trials have been widely adopted in practice. However, to reach a theoretically reliable judgement of risk, one typically needs to collect mass...

We present a novel distributionally robust optimization approach for integrated design and assessment of fault detection system. Its salient feature is the guaranteed robustness against the inexactness of probability distribution of unknown disturbances. The integrated design problem is formulated as a distributionally robust chance constrained pro...

Latent variable (LV) models have been widely used in multivariate statistical process monitoring. However, whatever deviation from nominal operating condition is detected, an alarm is triggered based on classical monitoring methods. Therefore they fail to distinguish real faults incurring dynamics anomalies from normal deviations in operating condi...

As a useful and efficient alternative to generic model-based control scheme, data-driven predictive control (DDPC) is subject to bias–variance tradeoff and is known to not perform desirably in face of uncertainty. Through the connection between direct data-driven control and subspace predictive control (SPC), we gain insight into the reason being t...

For decades, subspace identification method (SIM) has been widely adopted for modeling multiple-input multiple-output processes. However, conventional SIMs yield unsatisfactory performance in modeling processes with evident dead time characteristics. To tackle this challenge, we develop in this work an efficient SIM scheme with consideration of tim...

Direct data-driven control methods are known to be vulnerable to uncertainty in stochastic systems. In this paper, we propose a new robust data-driven predictive control (DDPC) framework to tackle the uncertainty in dynamic systems. By analyzing non-unique solutions to behavioral representation, we first shed light on the lack of robustness in subs...

The robustness of fault detection algorithms against uncertainty is crucial in the real-world industrial environment. Recently, a new probabilistic design scheme called distributionally robust fault detection (DRFD) has emerged and received immense interest. Despite its robustness against unknown distributions in practice, current DRFD focuses on t...

In this paper, we describe a novel unsupervised learning scheme for accelerating the solution of a family of mixed integer programming (MIP) problems. Distinct substantially from existing learning-to-optimize methods, our proposal seeks to train an autoencoder (AE) for binary variables in an unsupervised learning fashion, using data of optimal solu...

The pervasiveness of PID control in process industries stipulates the critical need for efficient autotuning techniques. Recently, the use of Bayesian optimization (BO) has been popularized to seek optimal PID parameters and automate the tuning procedure. To evaluate the overall risk-averse performance of PID controllers, scenario programming that...

Sparse canonical correlation analysis (CCA) is a useful statistical tool to detect latent information with sparse structures. However, sparse CCA, where the sparsity could be considered as a Laplace prior on the canonical variates, works only for two data sets, that is, there are only two views or two distinct objects. To overcome this limitation,...

Recently, dynamic latent variable (DLV) models have been prevalent in dynamic data modeling and process monitoring. They maximize the covariance or canonical correlation between latent components and one-step ahead prediction thereof; however, auto-correlations may still be present in residuals, resulting in unmodeled dynamics and hence compromisin...

This paper proposes a novel state estimation strategy with globalized robustness for a class of systems under uncertainty. Departing from the classical minimax estimation, this paper focuses on the globalized robust estimation (GRE), which minimizes the estimator's fragility to attain an acceptable loss compared with the nominal model. The GRE prob...

The alternating direction method of multipliers (ADMM) has been widely adopted in low-rank approximation and low-order model identification tasks; however, the performance of nonconvex ADMM is highly reliant on the choice of penalty parameter. To accelerate ADMM for solving rank-constrained identification problems, this paper proposes a new self-ad...

As a useful and efficient alternative to generic model-based control scheme, data-driven predictive control is subject to bias-variance trade-off and is known to not perform desirably in face of uncertainty. Through the connection between direct data-driven control and subspace predictive control, we gain insight into the reason being the lack of c...

Relay feedback tuning has been a state-of-the-art autotuning technique in process control, where frequency-domain analysis is generically carried out using limit cycle data for model identification and controller tuning. However, due to noise and disturbance, nonnegligible approximation errors exist in frequency information distilled from data, res...

Recent years have witnessed a booming interest in data-driven predictive control of dynamical systems. However, ill-conditioned solutions may occur in face of stochastic systems, causing inaccurate predictions and unexpected control behaviours. In this article, we develop new data-driven solutions to output prediction and control tasks of stochasti...

The alternating direction method of multipliers (ADMM) has been widely adopted in low-rank approximation and low-order model identification tasks; however, the performance of nonconvex ADMM is highly reliant on the choice of penalty parameter. To accelerate ADMM for solving rankconstrained identification problems, this paper proposes a new self-ada...

In this work, we investigate the data-driven safe control synthesis problem for unknown dynamic systems. We first formulate the safety synthesis problem as a robust convex program (RCP) based on notion of control barrier function. To resolve the issue of unknown system dynamic, we follow the existing approach by converting the RCP to a scenario con...

For using the infrared spectroscopy to conduct in-situ measurement of slurry component concentrations during batch crystallization or fermentation process in engineering practice, a novel just-in-time learning (JITL) based functional modelling method is proposed for spectral data analysis in this paper to improve on-line measurement accuracy, by ad...

In this letter, we propose a simple yet effective singular value decomposition (SVD) based strategy to reduce the optimization problem dimension in data-enabled predictive control (DeePC). Specifically, in the case of linear time-invariant systems, the excessive input/output measurements can be rearranged into a smaller data library for the non-par...

Current data-driven predictive control (DDPC) methods heavily rely on data collected in open-loop operation with elaborate design of inputs. However, due to safety or economic concerns, systems may have to be under feedback control, where only closed-loop data are available. In this context, it remains challenging to implement DDPC using closed-loo...

In this article, we study the challenging few-shot fault diagnosis (FSFD) problem where limited faulty samples are available. Metric-based meta-learning methods have been a prevalent approach toward FSFD; however, most of them rely on learning a generalized distance metric and fall short of leveraging intraclass and interclass distribution informat...

The need for exact distributions in probabilistic fault detection design is hardly fulfilled. The recent moment-based distributionally robust fault detection (DRFD) design secures robustness against inexact distributions but suffers from over-pessimism. To address this issue, we develop a new DRFD design scheme by using unimodality, a ubiquitous pr...

For an ethylene cracking furnace system (ECFS), the cyclic scheduling is significant for increasing the benefit. However, the scheduling period is always more than three hundred days, the frequently changed prices of raw materials and products may lead to suboptimal results. This paper proposes a data-driven robust optimization (DDRO) approach to h...

This article is concerned with data-driven realization of fault detection (FD) for nonlinear dynamic systems. In order to identify and parameterize nonlinear Hammerstein models using dynamic input and output data, a stacked neural network-aided canonical variate analysis (SNNCVA) method is proposed, based on which a data-driven residual generator i...

Moment-based ambiguity sets are mostly used in distributionally robust chance constraints (DRCCs). Their conservatism can be reduced by imposing unimodality, but the known reformulations do not scale well. We propose a new ambiguity set tailored to unimodal and seemingly symmetric distributions by encoding unimodality-skewness information, which le...

To measure the moisture content of granules during the industrial fluidized bed drying (FBD) process, a semi-supervised calibration model is proposed for using the near-infrared (NIR) spectroscopy to conduct in-situ measurement. To solve the dilemma of lacking sufficiently labeled samples as often encountered in various batch FBD processes, a semi-...

Fisher discriminative analysis (FDA) has been recognized a prototypical approach to fault classification and diagnosis. To enhance model performance with time-series data used, it is customary to encompass lag measurements into the model. This not only increases model complexity prohibitively but also reduces the interpretability of fault diagnosis...

In this work, we propose a data-driven robust model predictive control (DDRMPC) framework that utilizes stem water potential (SWP) as a basis for effective irrigation control of high value-added crops. By linearizing and discretizing a nonlinear dynamic model of water dynamics, we develop a state-space model that predicts the dynamic state of SWP....

Refinery planning under uncertainty has gained tremendous attention, and this paper bridges deep learning and robust optimization to address this issue. First, we propose a large-scale mixed-integer linear programming model for refinery planning, where the fixed-yield models of the processing units are used. Prices of final products are considered...

Probabilistic methods have attracted much interest in fault detection design, but its need for complete distributional knowledge is seldomly fulfilled. This has spurred endeavors in distributionally robust fault detection (DRFD) design, which secures robustness against inexact distributions by using moment-based ambiguity sets as a prime modelling...

Coal gasification technology has gained increasing popularity in recent years, but the optimization of operating conditions remains inefficient. The operation optimization of the Shell coal gasification process (SCGP) is investigated in this paper using an operation optimization model integrating data analytics and mechanism analysis. The objective...

Establishing a quantitative similarity between different datasets has gained prevalence and significance in many applications of process control. In industrial practice, process data are usually multi-dimensional, nonlinearly correlated, and with unknown time-varying distribution, which raise immense challenge for reasonably evaluating similarity....

In the presence of low-quality industrial process data, generic step response identification methods typically show unsatisfactory performance and heavily rely on manual intervention of technical personnel. This erects obvious obstacles for the advancement of intelligent manufacturing in process industries. To address these challenges, we propose a...

Robust optimization (RO) has been broadly utilized for decision-making under uncertainty ; however, as a key issue in RO the design of the uncertainty set could exert significant influence on both the conservatism of solutions and tractability of induced problems. In this paper, we propose a novel multiple kernel learning (MKL)-aided RO framework f...

Multivariate statistical process monitoring (MSPM) methods provide sensitive indicators of process conditions by harnessing the value of massive process data. Large-scale industrial processes are subject to wide-range time-varying operating conditions such that some variables inevitably exhibit non-stationary behavior, which poses significant chall...

We present a novel distributionally robust optimization approach to integrated design and assessment of fault detection system. The salient feature is the guaranteed robustness against the inexactness of probability distribution of unknown disturbances. The integrated design problem is formulated as a distributionally robust chance constrained prog...

Scenario programs have established themselves as efficient tools towards decision-making under uncertainty. To assess the quality of scenario-based solutions a posteriori, validation tests based on Bernoulli trials have been widely adopted in practice. However, to reach a theoretically reliable conclusion, one typically needs to collect massive val...

In classical paradigm of model identification, a single prediction value is returned as a point estimate of the output. Recently, the interval prediction model (IPM) has been receiving increasing attentions. Different from generic models, an IPM gives an interval of confidence as the prediction that covers the majority of training data while being...

We propose a novel probabilistic fault detection scheme with adjustable reliability estimates. Our scheme consists of two phase, the first is the modelling phase, where a probabilistic fault detection design is devised, while the second is the validation phase, where reliability estimates of the design are adjusted online according to new operation...

In industrial processes, causality analysis plays an important role in fault detection and topology building. Aiming to attenuate the influence of common correlation and noise, a feature based causality analysis method is proposed. By using the orthogonality and de-noising in feature analysis, it can capture more efficient causal factors. Moreover,...

In process industries, it is necessary to conduct fault diagnosis after abnormality is found, with the aim to identify root cause variables and further provide instructive information for maintenance. Contribution plots along with multivariate statistical process monitoring are standard tools towards this goal, which, however, suffer from the smear...

Safe, efficient, and sustainable operations and control are primary objectives in industrial manufacturing processes. State-of-the-art technologies heavily rely on human intervention, thereby showing apparent limitations in practice. The burgeoning era of big data is influencing the process industries tremendously, providing unprecedented opportuni...

We develop a novel data-driven robust model predictive control (DDRMPC) approach for automatic control of irrigation systems. The fundamental idea is to integrate both mechanistic models, which describe dynamics in soil moisture variations, and data-driven models, which characterize uncertainty in forecast errors of evapotranspiration and precipita...

In industrial processes, soft sensor models are commonly developed to estimate values of quality-relevant variables in real-time. In order to take advantage of the correlations between process variables, two convolutional neural network (CNN) based soft sensor models are developed in this work. By making use of the unique architecture of CNN, the f...

Stochastic model predictive control (SMPC) has been a promising solution to complex control problems under uncertain disturbances. However, traditional SMPC approaches either require exact knowledge of probabilistic distributions, or rely on massive scenarios that are generated to represent uncertainties. In this paper, a novel scenario-based SMPC...

We develop a novel data-driven robust model predictive control (DDRMPC) approach for automatic control of irrigation systems. The fundamental idea is to integrate both mechanistic models, which describe dynamics in soil moisture variations, and data-driven models, which characterize uncertainty in forecast errors of evapotranspiration and precipita...

Stochastic model predictive control (SMPC) has been a promising solution to complex control problems under uncertain disturbances. However, traditional SMPC approaches either require exact knowledge of probabilistic distributions, or rely on massive scenarios that are generated to represent uncertainties. In this paper, a novel SMPC approach is pro...

Recently, a new process monitoring and fault diagnosis method based on slow feature analysis has been developed, which enables concurrent monitoring of both operating point and process dynamics. In this article, a recursive slow feature analysis algorithm for adaptive process monitoring is put forward to accommodate time-varying processes, by updat...

Latent variable (LV) models have been widely used in multivariate statistical process monitoring. However, whatever deviation from nominal operating condition is detected, an alarm is triggered based on classical monitoring methods. Consequently, they cannot distinguish real faults incurring dynamics anomalies from normal deviations in operating co...

In this chapter, we further investigate the implication of SFA in control performance monitoring and diagnosis. It will be clarified that the process dynamics described by SFA essentially is related to control performance. Then we propose a new data-driven control performance monitoring approach based on SFA. In addition, by employing the contribut...

The common occurrence of slow drifts in industrial processes ask for the need of adaptive monitoring. In this chapter, a recursive slow feature analysis algorithm for adaptive process monitoring is developed to handle time-varying processes. An algebraic property of slow feature analysis is first revealed. We then show that such a property may be v...

Traditional data-driven static soft sensors only utilize single snapshots of process samples for quality prediction, thereby falling short of addressing process dynamics appropriately. Hence, a series of limitations in practice are incurred, such as sensitivity to temporal noises and inadequate descriptions to process dynamics. Because of these con...

In this chapter, we put forward a new nonlinear dynamic soft sensing model based on finite impulse response (FIR) and support vector machine (SVM). The whole model has a Wiener structure, in which nonlinearity and dynamics are described separately. In addition, a novel four-level Bayesian framework is developed to probabilistically illustrate and i...

Latent variable (LV) models provide explicit representations of underlying driving forces of process variations and retain the dominant information of process data. In this chapter, we develop a new soft sensor model called probabilistic slow feature regression (PSFR). Slow features as temporally correlated LVs are first derived using probabilistic...

This chapter summarizes the whole work of this thesis, and some challenges on data-driven modeling methodologies and industrial applications are pointed out as the future work.

Distributionally robust optimization (DRO) has been gaining increasing popularity in decision-making under uncertainties due to its capability in handling ambiguity of distributions. However, it is a nontrivial task to approach the scheduling problem within the DRO framework. In this paper, we propose a novel DRO scheduling model under uncertain de...

Slow feature analysis has proven to be an effective process monitoring and fault diagnosis approach. By isolating temporal behaviors from steady-state variations in process data, slow feature analysis enables a concurrent monitoring of operating condition and process dynamics, based on which false alarms triggered by nominal operating condition dev...

Process network planning is an important and challenging task in process systems engineering. Due to the penetration of uncertainties such as random demands and market prices, stochastic programming and robust optimization have been extensively used in process network planning for better protection against uncertainties. However, both methods fall...

Data-driven robust optimization has attracted immense attentions. In this work, we propose a data-driven uncertainty set for robust optimization under high-dimensional uncertainty. We propose to first decompose the high-dimensional data space into the principal subspace and the residual subspace by employing principal component analysis, and then a...

Distributionally robust optimization (DRO) is an emerging and effective method to address the inexactness of probability distributions of uncertain parameters in decision-making under uncertainty. We propose an effective DRO framework for planning and scheduling under demand uncertainties. A novel data-driven approach is proposed to construct ambig...

This paper presents a study on using deep learning for the modelling of a post-combustion CO2 capture process. Deep learning has emerged as a very powerful tool in machine learning. Deep learning technique includes two phases: an unsupervised pre-training phase and a supervised back-propagation phase. In the unsupervised pre-training phase, a deep...

To design robust PID controllers for second-order plus time delay (SOPTD) processes with parameter uncertainty, a reference model approximation method is proposed in this article. The central idea is to allow the frequency response of the PID controller to approximate that of a user-specified reference model. A convex hull is used to approximate th...

Due to process nonlinearities and operating condition changes, industrial processes frequently encounter significant dynamics variations, which would compromise the long-term effectiveness of controller monitoring schemes and leads to superfluous alarms. To address these issues, a novel performance benchmark based on the min-max principle is develo...

We propose piecewise linear kernel-based support vector clustering (SVC) as a new approach tailored to data-driven robust optimization. By solving a quadratic program, the distributional geometry of massive uncertain data can be effectively captured as a compact convex uncertainty set, which considerably reduces conservatism of robust optimization...

A novel data-driven method for simultaneous performance assessment and retuning of PID controllers is presented in this study. The PID tuning problem is cast as a convex approximation problem of a reference model, which makes the resulting closed-loop step response resemble that of a predefined reference model. The proposed data-driven method only...

Dynamic data reconciliation and gross error detection ask for an accurate physical model, e.g. a state-space model, based on which measurement noise and gross errors can be quantitatively assessed. The model can be established based on either first-principle knowledge or process operation data. This work considers a case with limited first-principl...

To solve the issue of low accuracy of the traditional soft sensor methods of polypropylene melt index, an approach based on deep belief network and extreme learning machine (DBN-ELM) was used to the melt index prediction of polypropylene. Traditional deep belief network (DBN) applied the deep learning to the learning process of the deep neural netw...

Owing to wide applications of automatic control systems in the process industries, the impacts of controller performance on industrial processes are becoming increasingly significant. Consequently, controller maintenance is critical to guarantee routine operations of industrial processes. The workflow of controller maintenance generally involves th...

Recently, slow feature analysis (SFA), a novel dimensionality reduction technique, has been adopted for integrated monitoring of operating condition and process dynamics. By isolating temporal behaviors from steady-state information, the SFA-based monitoring scheme enables improved discrimination of nominal operating point changes from real faults....