METHOD AND SYSTEM FOR ANALYSING INDUSTRIAL PROCESS PERFORMANCE
FIELD OF THE INVENTION
The invention relates to a method and system for measuring and analysing an industrial process performance visually from a graphical user interface.
BACKGROUND OF THE INVENTION
Process control systems control an industrial process by means of various field devices, e.g. regulating devices, control devices, sensors, trans- ducers and the like, which are connected to the process. A typical field device is a control valve provided with a valve controller. So-called intelligent field devices are provided with a control logic and software, which make it possible to control the field device locally for instance by means of a suitable control algorithm, to collect status and measurement data and/or to communicate with an automation system or a field device management system. A field device, such as an intelligent control valve, is typically controlled by a process controller using a suitable control algorithm on the basis of the measurement results (feedback) obtained from the process and the set values. Thus, a so-called control loop is formed, and a big industrial process may include several, even hundreds of such control loops.
Control loops (control circuits) are tuned during installation to produce a desired process operation as well as possible, and they can be controlled to improve the process performance or for some other reason. There is a variety of indexes and measurements representing the performance of a control system and a process. They all illustrate this important matter from different points of view. Suitable indexes and measurements should be selected each time among them to describe the process performance in a specific situation. Performance indexes are also dependent on one another, and improving the performance on the basis of one index may weaken the per- formance according to some other performance index. Further, when a large number of control circuits and control loops are used, it is difficult for a control room personnel to perceive and analyse the effect of different process controls and the real performance of a control loop or a sub-process in relation to the desired performance.
SUMMARY OF THE INVENTION
The object of the present invention is a method and a system, which allow the operating personnel of an industrial plant to better analyse and evaluate the performance of a control loop or a sub-process in an industrial process.
This is achieved by a method of analysing an industrial process performance, as claimed in claim 1.
The invention also relates to a system implementing the method. The basic idea of the invention is to implement a visual and user- friendly user interface for a performance analysis of a control loop. In the user interface, performance index information of a control loop is combined as an illustrative whole, which allows to immediately detect visually, if the performance of the control loop is weakening. The performance is displayed on the user interface graphically as two polygons, in each corner of which there is one piece of performance index information. One of the polygons is a reference polygon, and each corner point represents a normalised reference value of one performance index. These reference index values represent ideal operation and performance, which have served as a basis for design. For example, the target speed of various control loops can be told (as a rise time or a time constant) already in advance, in the design stage of the circuit. On the basis of this design information or other foreknowledge, reference values for the performance indexes are formed, and the real performance is compared with these values. Different performance indexes are also preferably scaled and normalised such that each corner point of the reference polygon is at an equal distance from the centre (origin) of the polygon, and a regular polygon is formed. Overlapping with the reference polygon, another polygon, i.e. performance polygon, is displayed, and each corner point of this polygon represents the real value, calculated on the basis of the measurements, of one performance index. The performance indexes of the performance polygon are scaled according to the reference polygon. Thus, if each performance index value of the control loop to be analysed is the same as the ideal reference value of the control loop, the corner points of the performance polygon are aligned with the corner points of the reference polygon, and the shape of the polygons is similar. If one of the real performance index values differs from the corresponding reference value, the respective corner point of the performance polygon is offset from the corresponding corner point of the reference polygon,
and the shape of the performance polygon differs from the regular shape of the reference polygon. The direction of change depends on whether the real performance is weakening or improving, when compared to the reference performance. When all real performance index values differ from the reference values, the shape of the performance polygon may differ from the shape of the reference polygon considerably. If such non-overlapping areas of the polygons that represent relatively weakened real performance as against the reference performance are illustrated with a different colour than the other areas, the performance of the control loop can be visualised in a very illustrative and clear manner. This performance presentation formed by two overlapping polygons is called a control diamond herein. A user interface may simultaneously display a large number of control diamonds according to the invention, and on the basis of a changed shape of the diamond, the user can perceive at a glance, if the performance of a control loop is weakening or changing. As each corner point of the polygons represents a specific performance index, the user may also detect easily, which performance index has led to a change in the performance. In a preferred embodiment of the invention, complementary performance indexes (when one improves, the other one weakens and vice versa) are located to the opposite corner points of the polygons. By means of the control diamond of the invention, a number of performance indexes and their co-operative action can be displayed for the user in an illustrative and a visually clear manner. In addition, the reference polygon offers a clear point of comparison for the index data obtained from the process measurements and helps the user to evaluate the relative perform- ance. This is a significant advantage: if measurement data obtained only from the process values were used for evaluating the control loop performance and for calculating the performance index, it could only be compared with the previous behaviour of the control loop in question. A drawback of such a performance indicator would be that in each control loop, each performance index would have been scaled to its own natural value range. Thus, drawing general conclusions from the functions of different control loops is difficult. "Ideal performance information", on which the reference polygon of the invention is based, also includes scaling information of these signal-dependent performance indexes, and this scaling information changes every time when the level of performance indexes is rising because of the improved performance.
BRIEF DESCRIPTION OF THE DRAWINGS
In the following the invention will be described by means of the preferred embodiments with reference to the attached drawings, in which
Figure 1 is a functional block diagram illustrating a performance measurement system of the invention,
Figures 2A and 2B illustrate a hexagonal control diamond of the invention in two performance situations,
Figure 3 shows a general view of a graphical user interface, in which (quadrangular) control diamonds of four different control loops are dis- played.
DETAILED DESCRIPTION OF THE INVENTION
The present invention can be applied to all industrial processes or the like, in which the performance of a control loop, a sub-process or the like is displayed on the user interface in an illustrative and clear manner. In Figure 1 , a process automation system is composed of e.g. a conventional distributed control system (DCS), which controls the process. The process automation system typically comprises a supervising computer, which contains process controllers or which is connected to the process controllers via a data network. Process controllers are typically connected by field buses to field devices, e.g. control valves, which may have their own control unit. A control loop may comprise e.g. a process controller, a field device and a measured feedback from the process to the process controller. A process automation system also typically comprises software or hardware for tuning control loops. Process automation systems, control loops and tuning systems are described in the co-pending Finnish patent application 990360, for example, which is incorporated as a reference herein. In addition, the process automation system may be connected to a field device management system FP, through which status and condition data can be collected from a field device. Such a management system is described for in the co-pending Finnish patent application 981235, for example. A process automation system also typically comprises a history database, to which measurement data and other information on the system are collected for later use. Only a few examples of process automation systems are described above. It is to be noted that the implementation method of an industrial process and a process automation system is not relevant for the present invention.
In Figure 1 , the functions according to the invention are illustrated with a performance analyser (computing and data processing unit), a database and a graphical user interface. They can be implemented in a conventional personal computer (PC), for example. In Figure 1 , the performance of a control loop is mainly calculated in the process automation system, in a data preprocessing block in connection with each control loop. Alternatively, this calculation can be performed in a performance analyser. As initial information, the preprocessing block obtains process measurement results and other information, concerning for example measurement information on the current value, set value and control and mode information of the control loop (automatic control, manual control, cascade connection of controllers) and error bit information of the measurements (indicating, whether there is an error in the measurement results or not). On the basis of this source information, the preprocessing block can calculate the performance index information, which is transferred to the above mentioned history database, for example. The data in the history database are still entirely unsealed.
The performance analyser reads the latest performance index information from the (on-line) automation system and displays it on a graphical user interface, i.e. on the monitor display of the control room. The performance analyser may also store performance information of field devices in its own project database.
The performance analyser contains a variety of performance indexes of control loops, on the basis of which the total performance is dis- played on the graphical user interface (in the control diamond) of the invention. Four or six of the following performance indexes are used in the preferred embodiments of the invention: control speed index, CSI, from a closed-circuit speed tuning test, robustness index, Rl, from a controller robustness tuning test, variability index, VI, variability of an error parameter or a measurement, control travel index, CTI, a control variable travel (proportional to the work done by a regulating unit) integral absolute error, IAE, an integral of the absolute value of an error parameter (control error)
statistical performance index, SPI, a statistical performance (based on the correlation) oscillation index, 01, an oscillation index from on-line measurements. Determining the performance of a sub-process is always a much more complex problem. Generally, the most suitable indicators have to be tailored to each sub-process separately. Some examples of performance indexes of a sub-process are as follows:
Automation grade, the number of control loops (including feedfor- wards and other advanced controls) operating in an automatic control mode), storing indexes, direct quality and capacity indicators, specific energy, raw material and chemical consumption. As described above, the control diamond of the invention is a visual manner of representation, which combines the performance index information of a control loop as an illustrative whole, in which it can immediately be detected, if the circuit is weakening. The control diamond in Figures 2A and 2B is a hexagon, and in each corner of the hexagon there is one piece of performance index information. In other words, the control diamond visualises the in- formation of six performance indexes from the performance of one single control loop. All six indexes CSI, Rl, VI, CTI, Ol and SPI (it is possible to replace SPI with the performance index IEA) are scaled such that the index, which equals to one, refers to nominal desired control quality. Nominal quality means a reference value, which is comparable with similar typical process control so- lutions. These typical process control reference values are tabulated in a specific process list. The bigger the performance index, the worse quality it refers to. The control diamond is used for visualising control indexes.
The performance indexes are preferably arranged such that complementary performance indexes are placed in the opposite corners of the control diamond: the axes CS-RI and CT-VI and CT-IAE. The oscillation index Ol is independent of other performance indexes and it does not have a complementary pair. A complementary pair means that when one performance index of the complementary pair is improving, the other one is getting weaker, and vice versa. In Figures 2A and 2B, in the foreground (on top) of the control diamond there is a regular hexagon 21 (for example of green colour on the
graphical display), (a reference hexagon), which illustrates the operation of an ideal circuit. In other words, at each corner point of the hexagon 21 there is a reference value for the performance index, the value being scaled such that the distance of each corner point from the centre (origin) of the hexagon is one. Under the reference hexagon 21 there is a hexagon 22 (performance hexagon) indicating the real performance. The performance indexes, which are calculated on the basis of the process measurements and/or tuning information and/or process models and which are scaled according to the ideal control loop, are located at the corner points of the reference hexagon. The performance hexagon can be of red colour, for example. When only green colour, i.e. the reference hexagon 21 , is visible, the performance is good. In this case, the real values of all performance indexes are lower (better) than the corresponding reference values. The quality of the performance can be observed by following the outlines of the performance hexagon 22 inside the green reference hexagon 21. As the performance is weakening, the value of at least one performance index is rising, and the corresponding corner point of the performance hexagon 22 moves outside of the reference hexagon 21 and the red colour becomes visible. The red colour indicates that something in the control loop operation has got worse. For example, the value of the index VI in Figure 2A is higher than the corresponding reference value, whereby the corresponding corner of the hexagon 22 crosses the lines of the reference hexagon 21 and becomes red.
When observing the control diamond of Figure 2A, it can be seen that almost all indexes are on the green area. Only the variability range (variability index) is somewhat on the red area. Although the circuit is now tuned to be faster than the reference circuit (CSI is almost in the middle of the green area), the complementary robustness has not weakened either (Rl in the middle, too). In other words, the circuit could easily be tuned faster without having to meet any actual problems in stability. For some reason, however, a considerably slower closed circuit speed (CSI) has been selected as the ideal speed of the circuit. Before accelerating the tuning, it should be made sure that the ideal speed has not been defined on the basis of wrong information. The. oscillation index Ol also indicates a small growth (in the ideal state, it would be zero), so the circuit should also be monitored later on. In Figure 2B, a control loop is tuned almost in the same manner as in Figure 2A. The real values of the performance indexes VI, Ol and CTI are
considerably higher (worse) than the corresponding reference values, and index areas of the hexagon 22 become visible outside of the reference hexagon 21 and they are red-coloured. The oscillation index Ol in particular indicates a clear oscillation. The controller tries to compensate the situation by strong control steps (since the circuit is tuned to be fast). The oscillation can also be seen in the growth of the variability range (VI).
On the basis of the control diamond display of Figure 2B, the operating personnel may easily detect a weakly operating control loop among better operating control loops, which have displays that are similar to Figure 2A, and analyse the reasons for the weak performance, as was illustrated above. Figure 3 illustrates a graphical user interface, which displays control diamonds of four different control loops simultaneously. In Figure 3, the reference polygon and the performance polygon are squares. Unlike in Figures 2A and 2B, the reference polygon 31 is located under the performance polygon. The reference polygon 21 is for example red and the performance polygon 32 green. The red square 31 illustrates the desired behaviour of the control loop, according to the behaviour of its model loop. The upper comer point of the red square 31 on the vertical axis illustrates the desired speed of the closed loop (CS) and the complementary index in the lower corner on the vertical axis indi- cates the loop robustness (Rl), in the same manner as it should be in the ideal model loop. The corresponding corner points of the green square 22 describe the real speed and robustness of the closed loop (in relation to the reference indexes), which are obtained from the controller tuning tests. When red colour can be seen behind the green square 22, a conclusion can be drawn that the loop is either too slow or too unrobust or both.
The corner points of the red square 31 on the horizontal axis indicate the variability VI of the process value and the control travel index of the control signal CTI. The model loop cannot define typical values for these indexes, since they are always dependent on the real existing disturbance level (noise level) of that particular control loop. During the start-up, the user inputs suitable initial values for these indexes, which are stored as loop-specific values in the database. Whenever the user wants, he can change these values afterwards, and the history data is still comparable with the new data. The real measured variability and control travel (green corners) are compared with these values. The data on the vertical axis is based on the control loop tuning information, and it is collected a couple of times in a year (the data on the
horizontal axis, however, is entirely based on the measured signals, and the sampling rate may vary from few seconds to hours).
The control diamond displays of the invention may also have separately determined warning and alarm limits, and if these limits are exceeded, the information on the changes is given for example as text in connection with the corresponding control diamond display. The performance analyser can also collect history data into a database. The performance of sub-processes can be analysed afterwards on the basis of the history data in various ways. One central property is that the control diamond displays can be categorised in versatile ways for example on the basis of running method, type, quality, operating point or a moment of the day. The best possible running method (automatic, manual, cascade) can thus be determined for each type. Statistical methods, on the other hand, provide conventional means of analysis for searching the history data. The invention and the embodiments thereof are not restricted to the examples described above, but may be modified within the scope of the claims.