Method for the monitoring of a plant

Details Method for the monitoring of a plant Method for the monitoring of a plant

2001-08-06

2003-07-01

Picard, Leo (Department: 2125)

Data processing: generic control systems or specific application

Generic control system, apparatus or process

Optimization or adaptive control

C700S029000, C703S002000

Reexamination Certificate

active

06587737

ABSTRACT:

The invention relates to a method for the monitoring of a plant having a plurality of sub-systems in accordance with the preamble of the independent claim.
Plants such as hydroturbines or gas turbines with driving generators for the generation of electrical energy in power stations, turbocompressors and piston compressors to compress gases, pump stations or aeronautical engines are typically very complex systems which frequently work at different working points and whose respective operating status is influenced by a large number of process parameters. It is necessary, as a rule, to monitor the status of such plants in order to detect operating malfunctions, that is deviations from normal operating characteristics, at the earliest possible time or to monitor the state of wear of individual components so that necessary maintenance work can be planned in time and carried out efficiently.
It is, for example, possible for this purpose to inspect the plant at regular intervals. However, it is usually necessary to switch the plant off for this, which is disadvantageous from an economic aspect.
A plurality of process parameters such as, for example, pressure, temperatures at different points of the plant, flow rates, speeds, power, bearing temperatures, etc. is frequently determined by measurement and, for example, stored as a function of time or presented in graphical form. Such plants do not, however, usually work at a fixed operating or working point so that the time-dependent course of the measured parameters detected for the monitoring also shows great fluctuations in normal operation, that is operation free from malfunction. It is therefore only possible to judge whether the plant is working without malfunction with great difficulty—if at all—by using the time-dependent course of the measured parameters. While fixed threshold values can be set for some process parameters detected by measurement, with an alarm being triggered if these are not reached or exceeded, this procedure also causes the disadvantage due to the variable working points that false alarms are frequent or that actually existing operating malfunctions are not recognised in time. It is furthermore difficult to recognise and evaluate changes which take place very slowly over time such as occur, for example, due to operation-related wear.
It would be possible in principle to make a physical model of the whole plant, that is to calculate the physical relationships between the individual process parameters and then to carry out an evaluation of the operating status by a comparison of such physical model calculations and the parameters detected by measurement. However, this approach is frequently much too cost-intensive and complex in practice so that it is less suitable, in particular for industrial applications. One reason for this is that such plants have an enormous complexity with a plurality of sub-systems in mutual interaction with one another so that a more or less reliable physical model must consider a plurality of relationships between the individual process parameters, whereby its preparation is made into an extremely difficult task which is both time and cost intensive.
For this reason, a monitoring method is proposed in EP-A-0 895 197 which is based on an experimental modelling, that is without using a prior calculation of the physical relationships between the process parameters. Respective measured values are detected for a fixed set of process parameters at pre-settable time intervals. Measured values are detected for the process parameters for as many different working points as possible in a first, so-called modelling phase, with a check being made that the plant is working without malfunction during this modelling phase. An experimental model for the operating characteristics is prepared using the measured values detected during the modelling phase, with the input variables of the model being at least a part of the set of process parameters and with the output values comprising a model value for at least one of the process parameters. A respective residual value is determined by a comparison of the respective model value and the actual measured value of the modelling phase corresponding thereto and the model is optimised by determining model parameters such that a model error determinable from the residual values becomes minimal. A simple mathematical relationship, which as a rule has no physical significance, is usually selected as the model structure. When the modelling phase has been completed, the experimental model has thus “learned” how the plant, i.e. the individual process parameters, behaves at different working points.
In the second phase, the normal operating phase of the plant, at least one monitoring parameter, which is independent of the respective current working point, is determined at pre-settable time intervals using the model for the operating characteristics. This monitoring parameter is preferably the residual value resulting from the difference between the respective current measured value and the model value corresponding thereto. The time-dependent course of the monitoring parameter is used to evaluate the wear in the sub-systems of the plant and/or to detect operating malfunctions.
The method disclosed in EP-A-0 895 197 has the advantage that it essentially works without physical modelling and is therefore very simple and suitable for industrial applications. Moreover, it takes into account the respective current working point of the plant and can also recognise slowly progressing changes such as are caused, for example, by wear, at an early point. Complex plants can also be reliably monitored in this way. Furthermore, an efficient planning of the maintenance work is possible, which results in a reduction of maintenance and operating costs.
Even though the method in accordance with EP-A-0-895 197 has proved itself in practice, there is nevertheless a need for improvement. The isolation and identification of faulty process parameters is actually relatively difficult and costly. An unrecognised fault in a process parameter can result in the triggering of a false alarm, which represents a limitation, in particular under economic aspects.
It is therefore an object of the invention to modify and improve such a method for the monitoring of a plant having a plurality of sub-systems such that a fault in a process parameter is reliably recognisable, such that the faulty process parameter can be identified more easily and such that false alarms are avoided as much as possible.
The method which satisfies this object for the monitoring of a plant having a plurality of sub-systems which is operable at variable working points is characterised by the features of the independent claim. Advantageous measures and preferred embodiments of the invention can be seen from the dependent claims.
The method in accordance with the invention therefore comprises the following steps: respective measured values are detected for a fixed set of process parameters at pre-settable time intervals during the operation of the plant. The measured values detected in a learning phase for different working points are used to prepare models for the operating characteristics of the sub-systems, with the input values of each model being at least a part of the process parameters and the output value of each model comprising a model value for at least one of the process parameters, and with the models being optimised by comparing the model values with the measured values. At least one monitoring parameter, which is independent of the respective current working point, is determined at pre-settable time intervals in an operating phase using the models and the time-dependent course of the monitoring parameter is used for the monitoring of the plant. A pre-check is carried out prior to the determination of the monitoring parameter, in which a check is made whether at least the measured values for those process parameters which are operating parameters are within a pre-determined range.
The method in accordance with the i

Affiliated with

Also associated with

Rating

Method for the monitoring of a plant does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Method for the monitoring of a plant, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for the monitoring of a plant will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3044509