Classifications

• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B11/00—Automatic controllers
  • G05B11/01—Automatic controllers electric
  • G05B11/36—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential
  • G05B11/42—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential for obtaining a characteristic which is both proportional and time-dependent, e.g. P. I., P. I. D.
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
  • H03H17/00—Networks using digital techniques
  • H03H17/02—Frequency selective networks
  • H03H17/0248—Filters characterised by a particular frequency response or filtering method
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
  • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
  • G05B13/0205—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric not using a model or a simulator of the controlled system
  • G05B13/021—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric not using a model or a simulator of the controlled system in which a variable is automatically adjusted to optimise the performance

Definitions

• the invention relates to a method and a corresponding device for filtering a signal, in which a noisy input signal is continuously checked as to whether it is inside or outside a deadband.
• the invention further relates to a control device for a process.
• Control structures are used to automate process engineering processes.
• a control loop is composed of this
• control algorithm which determines from the control difference, how the actuator (eg valve, damper, Mo ⁇ gate, ...) is to be moved so that the controlled variable approaches the reference variable,
• the controlled variable of a component of a technical plant by means of the controller ⁇ rule should be as good as possible maintained at a desired value for the control variable.
• a problem with control processes generally represents the noise of the controlled variable, the cause of which may lie in the metrological detection of the controlled variable, a process noise or the like caused by disturbing influences.
• Unwanted oscillations of the actual value of the controlled variable can be avoided, for example, with a method which is known from EP 1 490 735 Bl.
• the actual value of the control ⁇ size is continuously determined and the gain of a PI controller in dependence on the temporal behavior of the actual value changed as long as until the actual value is within a predetermined tole ⁇ ranzbandes respect to the target value remains.
• the Amplification factor only changed again when the actual value of the controlled variable is in value removed from the tolerance band.
• the ⁇ se re constant change in gain overall schieht only as long again until the actual value plunges back into Tole ⁇ ranzband and remains there.
• the noise of the controlled variable leads to a corresponding noise of the control difference and thus to a constant excitation of the controller. This will then cause the actuator to constantly perform small actuating movements. However, the noise can not be eliminated by this adjusting movement, it is possibly even amplified.
• the drive itself is heavily loaded by the constant, unnecessary back and forth movements, the wear is correspondingly large.
• the signal curve for the controlled variable or for the control difference must be filtered in control engineering.
• a signal must be generated to advertising that reflects the history of the original signal in Wesentli ⁇ chen but the rapid, small signal variations does not include.
• a simple method of implementing such a filtering at the input of a regulator is the use of a dead band, which is usually sufficient for signal filtering in power plant technology. Where higher-quality filter algorithms Need Beer Untitled ⁇ , they will be additionally installed.
• f (x) x - T B for x> a
• the value 0 is output as output signal. Outside this range, the value of the input signal is reduced or increased by half the deadband width. There may also be a corresponding offset as expressed here by the constant a. Accordingly, with such a deadband function, it is possible to smooth out signals which fluctuate around the value 0. As an extension of Totbandfunktion also signals can be filtered, which fluctuate around values un ⁇ zero (eg here a). In control systems, controller modules are usually equipped with such dead band functions as standard. The control difference, which fluctuates around the value zero due to the fact that the control variable is regulated to its setpoint, is then switched to the deadband. If the control difference changes only within the set deadband, by definition the value zero will always be present at the output of the filter and the controller will thus not be excited. Only major changes the rule difference come through. The influence of the noise is thus eliminated.
• the width of the deadband must be set individually for each control loop. If the deadband is too small, some of the noise is still coming through. If it is too large, the controller reacts too late to actually occurring control differences.
• the controlled variable signal must therefore be examined and the width of the noise determined in order to set the dead band correctly.
• the setting of all deadbands is therefore a very complex process. This is especially true when you consider that the noise can vary z. B. as a function of the current operating point of the system, over time by changing the external interference or by signs of wear o. ⁇ .
• the settings for the Tot ⁇ tapes must therefore be readjusted several times under certain circumstances. This is very cumbersome because the adjustment of the deadbands in the power plant control nowadays by hand by an engineer in the context of procedural commissioning and optimization.
• the object is achieved according to the invention by a method and a corresponding device in which a noisy input signal is constantly checked to the effect that whether it is inside or outside a deadband.
• the Totbandbrei ⁇ te in contrast to the prior art is not fixed or set in advance, but is changed during operation as a function of the time behavioral the input signal and a selected system time constant.
• the zero point of the deadband function is changed and thus adjusted by means of the algorithm according to the invention.
• ⁇ at advantageously a self-learning algorithm, for example, in an industrial plant -. Always moves along - even angesto ⁇ KISSING .
• the deadband width increased by a factor when the input ⁇ signal within a first, preferably short period of time a certain number n times by the dead band pass Gegan ⁇ gene.
• a change to be adapted parameter where the gain factor of a regulator
• the parameter change (here the adjustment of the dead band width) is performed. This gives more flexibility with regard to the setting of the deadband width.
• the deadband width is continuously nert over time verklei ⁇ when the input signal during a second, be ⁇ vorzugt long period of time is within an interior region of the dead band and remains there.
• a second filtered output signal such as a corrected around the mean of the noise is Totbandsignal is output.
• Counting the Totband trim sauté is reset to zero advantageous if the time interval between a maximum and minimum of the input signal is greater than a predetermined maxi ⁇ male half period. In this way, beneficial way ⁇ noise from "real" signal movement is distinguished.
• a particularly advantageous possibility of OF INVENTION ⁇ to the invention process is in control technology.
• a control device comprising a device for filtering a signal using an adaptive dead band in the current Be ⁇ shoot, increases in a technical installation, the Quali ty ⁇ the control and thus also of the system to essentially lie ⁇ constricting process. Not only in the power plant sector is the trend towards the fully automatic adaptation of regulations of the individual components.
• the control device with adaptive, fully automatic signal filtering during operation of a system is universally applicable and is capable of optimi ⁇ male and plant-friendly control results to achieve.
• the control device consists of an apparatus for adaptive signal filtering according to the invention and a downstream PI controller with adaptively adjustable gain factor.
• Avoid vibrations of the input signal and the controller can be adjusted so that an optimal control performance is achieved, ie that the controlled variable follows its setpoint as closely as possible.
• the combination of both components results in many advantages: There are fewer costs, because work that was previously carried out manually now runs automatically. The regulations show less wear and will not get worse over time.
• a Anla ⁇ ge containing the control modules mentioned above, can be optimized more quickly, making it fast again ⁇ ler available.
• FIG. 2 shows a further exemplary time profile of the input signal, enlarged in its scale
• control device which is a device for
• Filtering according to the present invention comprises.
• FIG. 1 shows a computer printout of a screen display of a guidance system in which the method according to the invention is implemented.
• Fig. 1 a graphical representation of waveforms to see the lower part comprises a table with further information on the illustrated signal curves.
• the description of the embodiment relates We ⁇ sentlichen to those illustrated in the upper segment waveforms, with the abscissa representing the time t Sig ⁇ nalvercrest S are plotted (here as the time in minutes intervals) and on the ordinate.
• the input signal IN has a high-frequency noise. This can be any process signal, a control difference or the pure actual value of a measurement signal.
• At least one filtered output signal is continuously (online) calculated during the current operation of a system starting from the noisy input signal IN and other predetermined or predefinable values, output immediately and displayed.
• the starting value zero is specified for the dead band width DB.
• a starting value greater than zero would be possible.
• the learning algorithm is released. Then it takes a while for the algorithm to learn until the noise is gone (see area A in Figure 1).
• the deadband width is adaptively matched to the waveform, ie, the deadband width is kon ⁇ continuously changed in dependence of the temporal behavior of the input signal IN.
• the current total value DB of the deadband is the current total value DB of the deadband.
• the dead band may be for example symmetrically around the mean value of the minima and maxima of the amplitude of the input signal is classified ⁇ and extend similarly an envelope in the immediate vicinity of the minima and maxima of the amplitude of the input signal.
• the upper limit values of the dead band are designated UL DB, a respective current value is called UL DB ACT.
• the lower limits of the deadband are denoted by LL DB ⁇ net, each current value is called LL DB ACT.
• a dead band width DB it is checked whether the input signal lies inside or outside the dead band with its current width (area B of FIG. 1). If the value is within the current deadband width is output as an output signal, for example, a mean value of the preceding Gegan ⁇ genes vibration amplitudes. In the DAR identified in Fig. 1 embodiment is in the off ⁇ output signal OUT DB to the output of the dead band. For example, if the input signal is a control difference, the value zero is output for OUT DB, which means that the input signal is within the current deadband width.
• the noise shows a strong asymmetry, ie if the time average of the input signal is not exactly in the middle between the maximum and the minimum of the amplitude, this can be read off by means of a second output signal OUT.
• the time average of the noise- reduced signal is reproduced correctly by means of the signal OUT.
• the signal profile for the signal OUT is therefore not as smooth as the profile of the signal OUT DB, since the filter effect is somewhat reduced by the greater consideration of the signal fluctuations. Nevertheless, even here, the high vibration frequencies are no longer available.
• Fig. 1 clearly shows that the deadband width DB is changed continuously over time. Even at first glance, it can be seen that the dead band width is automatically increased or decreased, and like a tube lies around the amplitude extrema of the input signal and thus follows only the "real" fluctuations of the input signal. "Real" fluctuation means that here the mean value the maxima and minima of the input signal changes.
• a time constant In order to be able to make adaptive changes to the dead band width, a time constant must first be specified so that time spans or time periods can be defined and thus also rates of change of the input signal can be determined. A so-called system time constant thus describes the dynamic system behavior and depends on the considered overall system.
• a temperature control system for example Systemzeitkonstan ⁇ ten between 30 and 60 would have s, while a pressure control ⁇ system time constants 5-10 would comprise s.
• the signal noise of "real" signal changes can be distinguished.
• a temperature can s not change, for example, within a period of 5 several times. However, if the temperature reading shows such a behavior, it must be a Sig ⁇ nalrauschen and not to be a real signal change.
• a time duration of 20 s should be assumed here as the system time constant. If multiple oscillations of the input signal are observed within this period of time, this is noise in the case of a temperature measurement value, but in the case of a pressure measurement value it is a fast signal change. Also terms such as “long” or “short” periods of time can be quantified according to the system time constants.
• a reduction of the dead-band width takes place continuously, when the A ⁇ input signal IN has already spent within a second time period in ⁇ nerrenz an interior region of the dead band, and there is still always, wherein the second time period is determined by the system time constant and the width of the Indoor area is specified.
• the width of the inner region of the dead band is here case ⁇ set.
• a width of 95% of the total width of the dead band is taken at ⁇ for ⁇ for indoor rich.
• the input signal is continuously observed personally observed personallyge- thus starting, whether it is of a ⁇ in this case reduced to 5% width within or outside a Totban.
• This deadband will be referred to below as a 95% dead band.
• the dead band is continuously reduced. This can be clearly seen in FIG. 1 in region D.
• the oscillations of the input signal is run within the draw ⁇ th deadband limits over a period of nearly 3 minutes.
• the deadband width changes accordingly continuously.
• the rate of reduction of the deadband width is determined as a function of the system time constant. The slower the system, the slower the dead band is reduced.
• the speed of reduction of deadband width is also reduced the smaller the deadband is already.
• the reduction of the deadband width is stopped when the input signal IN is in the interior of the
• Deadband leaves again and / or the dead band width reaches a lower limit (see area E in Fig. 1).
• the mean value of the input signal changes conspicuously by about 25%.
• the dead band follows the resultsssig ⁇ nal and adapts to this. The zero point is shifted according to the waveform as a present "real" signal change. No major change to the deadband width is needed in these areas, as the noise power not än ⁇ changed.
• an increase in the deadband width by a factor occurs when the input signal IN ⁇ n times alternately upwardly and downwardly within crosses the dead band boundaries for a first period of time, where the number n of passes is dictated by the dead band boundaries, and the first time duration is determined by the system time constant.
• the first time period is a relatively short period of time.
• the dead-band width is at any other time in which the A ⁇ input signal passes in a short time by the dead band is increased by a factor.
• the magnification of the deadband width is stopped when the input signal no longer passes through or within a short period of time by the dead band, the dead ⁇ bandwidth reaches an upper limit.
• the counting of the Totband trim sautician is n based on the parameter is reset to zero at ⁇ when the input signal remains within the deadband again.
• FIG. 1 An example of the case of increasing the deadband width is shown in FIG. Here is a part of the area C of FIG. 1 has been enlarged.
• the input signal IN remains within the upper and lower deadband limits, which are shown here with thick lines. Furthermore, the 95% dead band is indicated by a thin line.
• the signal IN now exits the deadband and passes through the deadband DB at the point 2 for the first time.
• the signal IN now alternately goes down and up through the deadband 2 times within a short period of time.
• the time Tmax was previously set as the maximum time for a pass.
• the signal goes IN intra ⁇ half a shorter time period (T ⁇ Tmax) by the deadband limits.
• T ⁇ Tmax time period
• both conditions are met at point 3, which have an increase in the deadband width by a factor result.
• the signal remains within the deadband limits until it assumes a smaller amplitude at point 5.
• the signal now remains within the 95% deadband for a relatively long period of time, Tgr.
• the deadband is continuously reduced until the signal crosses the 95% deadband at point 6. Since the signal IN now remains within the outer dead band marked with a thick line, the dead band width is maintained.
• FIG. 3 shows a device F according to the invention for filtering a signal IN with a downstream regulating device R.
• a first input of the device F can be acted upon by the input signal IN.
• at least one second input XY is present to accommodate further parameters or values.
• Be such a input of the filter device F may, for example, the system time constant or the number n of passages of the input signal by the deadband limits, wel ⁇ che to increase the deadband width is needed supplied ⁇ leads.
• the filter device F comprises a calculation unit BE, by means of which the adaptive filtering of the input signal is carried out according to the present invention. At least one signal output for outputting the filtered output signal OUT is present at the output of the filter device F.
• At least a second output is provided for a second output signal OUT DB.
• At least one of the outputs shown can be connected to a regulator R.
• the controller R is used to control at least one component of a technical system and can be ⁇ example designed as a PI controller.
• the controller R can now be optimized by another block BS2.
• a combination of the inventive filter device F (or also block 1, BS1) with a PI controller R which is adapted according to EP 1 490 735 Bl, an increased control quality is achieved.
• the actual value of the controlled variable is continuously determined in block 2 BS2 and the gain K, and a reset time of a PI controller in dependence on the temporal behavior of the actual value changed as long as the actual value to within a pre give ⁇ NEN tolerance band remains with respect to the target value.
• the op ⁇ -optimized gain factor K, and a reset time is supplied to the controller R, which outputs a control signal ST, which in turn affects the controlled variable.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• Automation & Control Theory (AREA)
• General Physics & Mathematics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Medical Informatics (AREA)
• Software Systems (AREA)
• Evolutionary Computation (AREA)
• Health & Medical Sciences (AREA)
• Artificial Intelligence (AREA)
• Computer Hardware Design (AREA)
• Mathematical Physics (AREA)
• Feedback Control In General (AREA)

Priority Applications (1)

Applications Claiming Priority (4)

Publications (1)

Family

ID=45495935

Family Applications (1)

Country Status (4)

Families Citing this family (6)

Citations (1)

Family Cites Families (12)

• 2012
  • 2012-01-10 WO PCT/EP2012/050285 patent/WO2012095407A1/fr active Application Filing
  • 2012-01-10 US US13/979,159 patent/US20130346460A1/en not_active Abandoned
  • 2012-01-10 EP EP12700385.3A patent/EP2663903A1/fr not_active Ceased
  • 2012-01-10 CN CN201280012282.6A patent/CN103415818B/zh not_active Expired - Fee Related

Patent Citations (1)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events