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 the following steps: monitor operating controller performance and detect performance degradation, diagnose probable root causes of control system malfunctions, and take specific actions to resolve associated problems. In this article, a comprehensive overview of the mainstream of control loop monitoring and diagnosis is provided, and some existing problems are also analyzed and discussed. From the viewpoint of synthesizing abundant information in the context of big data, some prospective ideas and promising methods are outlined to potentially solve problems in industrial applications.

In a typical continuous process industry facility, the number of control loops can be about between 500 and 5000. In contrast to the wide applications of controllers, only a small portion of industrial controllers operates at healthy states, as pointed out by some existing investigations . It is therefore necessary to monitor industrial controllers and evaluate their closed-loop performance. However, the labor cost of manual monitoring is huge due to the large number of control loops, and thus achieving online and automated controller monitoring is an urgent demand from a practical perspective. To this aim, the technique of controller performance assessment (CPA), the objective of which is to detect performance degradation by analyzing routine closed-loop operating data, has gained considerable attention from both academia and industry in recent decades.

Detection of performance degradation is a prerequisite to improve controller performance. Then an important step afterwards is to diagnose probable root causes, which furnishes useful information for further maintenance. In general, possible causes of performance degradation can be enumerated as controller problems, equipment malfunctions and external disturbances. For controller problems, there is a need of intelligent guidelines about controller maintenance, and for problems like equipment malfunctions and external disturbances, it is imperative to report the corresponding faults timely such that specific actions can be taken to fix the related problems. From a pragmatic standpoint, an Orient–Decide–Act–Improve workflow should be followed, which includes the following steps

#8226; Orient: detect control loop performance degradation.

#8226; Diagnose: diagnose root causes of control system problems.

#8226; Act: repair and maintain the control system according to the diagnosis results.

#8226; Improve: assess whether closed-loop performance is improved

There are many techniques available for the design of feedback control strategies. These methods may be very simple, such as those used in statistical process control, or they may be more elaborate strategies, such as pole placement designs and linear quadratic controllers. Regardless of the control strategy, it is important to have some benchmark against which its performance can be evaluated. The theoretical best achievable control, as measured by the mean square error, is such a benchmark. If the theoretical best achievable control represents a significant improvement over the current performance, alternate controller tuning or feedback control strategies can be considered if this improved performance is warranted. However, the best achievable performance itself may not be adequate. In these cases alternate approaches, such as feedforward control, reduction of deadtime and different loop pairings must be used to achieve a reduction in variability.

The purpose of this paper is to describe a very simple technique for ascertaining the best theoretically achievable feedback control performance as measured by the output mean square error. A univariate time series model is tit to process data. From this time series model, the best control performance can be estimated if the number of whole periods of delay is known. An important feature of this method is that it is not necessary to 'perturb' the process with extraneous test signals. Data may be used from any linear process which is operating under normal closed loop control with a linear, time-invariant feedback control strategy. The outline of this paper is as follows. In the first section the process description is given.

This is followed by a brief derivation of a minimum variance controller, and of the resulting properties of a process operating under minimum variance control. It is then shown that the theoretical best achievable performance in the mean square sense can be estimated from process data collected under 'normal' closed loop conditions. A method for estimating this mean square performance is discussed. This is followed by a Monte-Carlo simulation to illustrate the statistical properties of the estimator. Finally, this technique is applied to pilot plant and production data.