EP0392036B1 - Method and compensation of the disturbed parts of the signals in a measuring system - Google Patents

Abstract

The system has a specific number (i = 1 to I) of measuring channels, each measuring signal (x<(i)>) having a channel-independent component(s) which is weighted with a channel-dependent compensation coefficient (A<(i)>), and an additive, channel-dependent interfering component (u<(i)>). Initially, a preliminary compensation value (sum) is formed from the measuring signals and the compensation coefficients and preliminary compensated measuring signal values (sk<(i)>) are formed with said compensation value. Measuring channels (j = 1 to J) with a high degree of interference are then sought. For these channels a respective approximation value (ñ) of the interference signal is determined and used to form a correction value (@). A final compensation value (sum'), with which the final compensated measuring signal values (sk') are calculated is formed using the correction values (@).

Description

The invention relates to a method for compensating interfering signal components in the measuring signals of a measuring system with a certain number of measuring channels, each measuring signal having a channel-independent component, which is weighted with a channel-dependent compensation coefficient, and an additive, channel-dependent interfering component (cf. e.g. EP-A-0 141929 and US-A-4 684 931).

, liefern. Jedes Meßsignal wird durch eine kanalunabhängige Komponente (gemeinsamen Signalanteil) s, die mit einem kanalabhängigen, im allgemeinen auch zeitabhängigen Kompensationskoeffizienten A⁽ⁱ⁾ gewichtet ist, sowie durch eine additive, kanalabhängige störende Komponente (individueller Signalanteil bzw. Störkomponente) n⁽ⁱ⁾ gemäß folgender Beziehung modelliert:

${\text{x}}^{\text{(i)}}{\text{= A}}^{\text{(i)}}{\text{. s + n}}^{\text{(i)}}\text{ (1)}$

Aufgabe der Erfindung ist die Kompensation dieser Signale untereinander mit dem Ziel, den gemeinsamen Signalanteil in allen Signalen von individuellen Signalanteilen in einem einzelnen Signal zu trennen. Dabei soll eine bessere Kompensation von wetterbedingten Signalverläufen erreicht werden, beispielsweise für einen kapazitiven Schutzzaun, mit dem Ziel einer verbesserten Erkennung von Eindringversuchen.Given is a measuring system with, for example, I measuring channels, which generally have time-dependent measuring signals x ⁽ⁱ⁾ $\text{i = 1 to I}$

, deliver. Each measurement signal is characterized by a channel-independent component (common signal component) s, which is weighted with a channel-dependent, generally also time-dependent compensation coefficient A ⁽ⁱ⁾ , and by an additive, channel-dependent interfering component (individual signal component or interference component) n ⁽ⁱ⁾ according to modeled the following relationship:

${\text{x}}^{\text{(i)}}{\text{= A}}^{\text{(i)}}{\text{. s + n}}^{\text{(i)}}\text{(1)}$

The object of the invention is to compensate these signals with one another with the aim of separating the common signal component in all signals from individual signal components in a single signal. In this case, better compensation of weather-related signal profiles is to be achieved, for example for a capacitive protective fence, with the aim of improved detection of intrusion attempts.

• a) aus den Meßsignalen x⁽ⁱ⁾ und den Kompensationskoeffizienten A⁽ⁱ⁾ wird in einer Summiereinrichtung ein vorläufiger Kompensationswert sum entsprechend der Beziehung:
  
  gebildet;
• b) aus den Meßsignalen x⁽ⁱ⁾ und aus den mit dem vorläufigen Kompensationswert sum gewichteten Kompensationskoeffizienten A⁽ⁱ⁾ wird in einer Subtrahiereinrichtung für jeden Meßkanal (i) ein vorläufiger kompensierter Meßsignalwert:
  
  ${\text{sk}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum (3)}$
  
  
  
  gebildet;
• c) aus den vorläufigen kompensierten Meßsignalen sk⁽ⁱ⁾ werden mittels bestimmter Suchstrategien schwach gestörte Meßkanäle (k bzw. m,1) oder stark gestörte Meßkanäle (j) ausgewählt;
• d) für die stark gestörten Meßkanäle $\text{(j = 1 bis J)}$
  
  wird ein Näherungswert ñ ^(j) der Störkomponenten (n⁽ⁱ⁾) nach der Beziehung:
  
  ermittelt;
• e) aus den Näherungswerten ñ ^(j) werden Korrekturwerte n̂ ^(j) für die stark gestörten Meßkanäle (j) abgeleitet;
• f) aus dem vorläufigen Kompensationswert sum und den Korrekturwerten n̂ ^(j) wird ein endgültiger Kompensationswert
  
  berechnet;
• g) mit dem endgültigen Kompensationswert sum' werden für jeden Meßkanal die endgültigen kompensierten Meßsignalwerte
  
  ${\text{sk'}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum' (3')}$
  
  
  
  ermittelt.

In a measuring system described in the introduction, this task is solved by the following process steps:

• a) the measurement signals x ⁽ⁱ⁾ and the compensation coefficients A ^{(i) are} used in a summing device to produce a provisional compensation value sum according to the relationship:
  
  educated;
• b) from the measurement signals x ⁽ⁱ⁾ and from the compensation coefficients A ⁽ⁱ⁾ weighted with the provisional compensation value sum, a provisional compensated measurement signal value is made in a subtracting device for each measurement channel (i):
  
  ${\text{sk}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum (3)}$
  
  
  
  educated;
• c) from the provisionally compensated measurement signals sk ⁽ⁱ⁾ , weakly disturbed measurement channels (k or m, 1) or severely disturbed measurement channels (j) are selected by means of certain search strategies;
• d) for the strongly disturbed measuring channels $\text{(j = 1 to J)}$
  
  becomes an approximation ñ ^{(j) of} the interference components (n ⁽ⁱ⁾ ) according to the relationship:
  
  determined;
• e) correction values n̂ ^(j) for the severely disturbed measuring channels (j) are derived from the approximate values ñ ^(j) ;
• f) the provisional compensation value sum and the correction values n̂ ^(j) become a final compensation value
  
  calculated;
• g) with the final compensation value sum ', the final compensated measurement signal values for each measuring channel
  
  ${\text{sk '}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum '(3')}$
  
  
  
  determined.

In the method according to the invention, a preliminary compensated measurement signal sk ⁽ⁱ⁾ is first determined, as previously. A preliminary compensation value sum is determined from the measurement signals x ⁽ⁱ⁾ by weighted averaging, a kind of estimated value of the common component s, which is inversely weighted and subtracted from the individual signals, as described above by the two relationships (2) and (3) is expressed. The signal value sk ^{(i) obtained in this way} , which is referred to as a provisionally compensated measurement signal value, basically represents an approximation of the individual signal component n ⁽ⁱ⁾ sought. Since strongly additive interference components n ⁽ⁱ⁾ in any signal x ⁽ⁱ⁾ cause the sum value sum to deviate greatly from the actual ideal value s, this has the disadvantageous consequence that such an additive interference component is found not only in the provisionally compensated signal sk ^{(i) of} the actual channel (i) concerned, but also in maps the signals sk ^(k) with k ≠ i of all channels with different amplitudes. Therefore, advantageously, the compensation method is further developed to the effect that approximate values ñ ⁽ⁱ⁾ of the additive noise components n ⁽ⁱ⁾ is determined, and that the provisional compensation value sum depending on determined therefrom correction values n ⁽ⁱ⁾ according to the above-mentioned relationship (4 ') is corrected.

With the final compensation value sum ', the final compensated measurement signal values are then determined for each measuring channel in accordance with equation 3' listed above.

Das Gleichungssystem (5) ist allerdings unterbestimmt und daher nur bei Vorliegen zusätzlicher Informationen auflösbar. Diese lassen sich dadurch gewinnen, daß man für einen oder einige der Kanäle - diese sind im folgenden mit dem Index k versehen - die bestimmten Störkomponenten n^(k) als bekannt annimmt. Wenn bestimmte Störkomponenten n^(k) als vernachlässigbar klein angesehen werden können, gilt n^(k) = 0. So können dann entsprechende Spalten und Zeilen des Gleichungssystems (5) gestrichen werden, so daß J-Gleichungen (J<I) übrigbleiben, aus denen die Unbekannten ñ⁽ⁱ⁾, die Näherungswerte, nunmehr berechenbar sind. Für dieses reduzierte Gleichungssystem existieren mehrere Lösungen. Eine besonders einfache Lösung läßt sich in vorteilhafter Weise unter Ausnutzung der regelmäßigen Struktur dieses Systems gemäß folgender Beziehung erzielen:

Daraus werden die Korrekturwerte n̂ ^(j) für die stark gestörten Meßkanäle (j) aus den Näherungswerten ñ ^(j) abgeleitet. In einfachster Weise können die Korrekturwerte unmittelbar den Näherungswerten entsprechen. Es ist jedoch wesentlich vorteilhafter, die Korrekturwerte aus den Näherungswerten in Abhängigkeit von einem vorgebbaren Schwellenwert n_o ^(j) zu ermitteln. Bei der Ermittlung des Korrekturwertes der stark gestörten Meßkanäle $\text{j = 1 - J}$

gilt dann folgende Beziehung:

Für die Auswahl eines Meßkanals oder mehrerer Meßkanäle mit vernachlässigbar kleiner additiver Störkomponente, d.h. schwach gestörte Meßkanäle, gibt es mehrere Suchstrategien. Eine zweckmäßige Suchstrategie besteht darin, daß bei mehreren solchen Meßkanälen die Werte

identisch sind. Zwei Kanäle mit vernachlässigbar kleiner additiver Störkomponente, die im folgenden mit dem Index m und 1 bezeichnet werden, werden daher in vorteilhafter Weise durch die Suche nach denjenigen Komponenten sk^(m) und sk⁽¹⁾ ermittelt, deren Werte sich möglichst wenig unterscheiden, also ein Minimum bilden. Die Ermittlung des Minimums erfolgt gemäß folgender Beziehung:

Die Suche nach schwach gestörten Kanälen kann auf eine beliebige Zahl von Kanälen ausgedehnt werden.The above approximation values ñ ⁽ⁱ⁾ can be obtained by solving the following system of equations (5):

However, the system of equations (5) is undetermined and can therefore only be solved if additional information is available. These can be obtained by assuming the specific interference components n ^(k) as known for one or some of the channels - these are given the index k below. If certain interfering components n ^(k) can be regarded as negligibly small, then n ^(k) = 0. Corresponding columns and lines of the system of equations (5) can then be deleted so that J equations (J <I) remain to which the unknown ñ ⁽ⁱ⁾ , the approximate values, can now be calculated. There are several solutions for this reduced system of equations. A particularly simple solution can advantageously be achieved using the regular structure of this system according to the following relationship:

The correction values n̂ ^(j) for the severely disturbed measuring channels (j) are derived from the approximate values ñ ^(j) . In the simplest way, the correction values can correspond directly to the approximate values. However, it is much more advantageous to determine the correction values from the approximate values as a function of a predefinable threshold value n _o ^(j) . When determining the correction value of the strongly disturbed measuring channels $\text{j = 1 - J}$

the following relationship then applies:

There are several search strategies for selecting a measuring channel or several measuring channels with a negligibly small additive interference component, ie weakly disturbed measuring channels. An expedient search strategy consists in the fact that the values for several such measurement channels

are identical. Two channels with negligibly small additive interference components, which are referred to below with the index m and 1, are therefore advantageously determined by searching for those components sk ^(m) and sk ⁽¹⁾ whose values differ as little as possible, that is form a minimum. The minimum is determined according to the following relationship:

The search for weakly disturbed channels can be extended to any number of channels.

Auf diese Weise wird eine weitere Fehlerquelle bei der Ermittlung kompensierter Signalwerte reduziert. Mit dem erfindungsgemäßen Kompensationsverfahren wird die Korrektur der gemessenen Signalwerte durch Korrekturwerte vor der endgültigen Berechnung der kompensierten Signalwerte erreicht, wobei die Korrekturwerte aus näherungsweise berechneten, additiven Komponenten ermittelt werden.In this compensation method according to the invention, the compensation coefficient A ^{(i) was used} as a known empirical value of constant size. However, the compensation coefficients can also be determined in a suitable manner from the measurement signals x ⁽ⁱ⁾ , for example by energy nomination of the measurement signals in a predetermined time window. Another advantage of erfindungsgemäßens method is that in such from the measurement signal x ⁽ⁱ⁾ continuously calculated compensation coefficients A ⁽ⁱ⁾ the previously determined approximate values ñ ⁽ⁱ⁾ can be subtracted also from the respective measuring signals and therefore significantly better coefficients A ⁽ⁱ⁾ can be determined according to the relationship:

${\text{A}}^{\text{(i)}}{\text{= f {(x}}^{\text{(i)}}{\text{- ñ}}^{\text{(i)}}\text{)}; (10)}$

In this way, a further source of error in the determination of compensated signal values is reduced. With the compensation method according to the invention, the correction of the measured signal values is achieved by correction values before the final calculation of the compensated signal values, the correction values being determined from approximately calculated additive components.

Claims (7)

1. Method for compensating disturbing signal components in the measuring signals of a measuring system having a specific number I of measuring channels, each measuring signal (x⁽ⁱ⁾) having a channel-independent component (s) which is weighted by a channel-dependent compensation coefficient (A⁽ⁱ⁾) and an additive, channel-dependent disturbing component (n⁽ⁱ⁾) in accordance with the relationship:
   
   ${\text{x}}^{\text{(i)}}{\text{= A}}^{\text{(i)}}{\text{. s + n}}^{\text{(i)}}\text{, where (i = 1 to I),}$

   

   
   
   characterised by the following method steps:

   a) a provisional compensation value sum is formed from the measuring signals x⁽ⁱ⁾ and the compensation coefficients A⁽ⁱ⁾ in a summing device in accordance with the relationship:

   

   b) for each measuring channel (i) a provisional compensated measuring signal value:
   
   ${\text{sk}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum}$

   

   
   
   is formed from the measuring signals (x⁽ⁱ⁾) and from the compensation coefficients A⁽ⁱ⁾ weighted with the provisional compensation value in a subtracting device;

   c) weakly disturbed measuring channels (k or m,1) or strongly disturbed measuring channels (j) are selected from the provisional compensated measuring signal values sk⁽ⁱ⁾ by means of specific search strategies;

   d) an approximate value n̂^(j) of the disturbing components (n⁽ⁱ⁾) is determined for the strongly disturbed measuring channels $\text{(j = 1 to J)}$

   

   in accordance with the relationship:

   

   e) correction values n̂^(j) for the strongly disturbed measuring channels (j) are derived from the approximate values ñ^(j);

   f) a final compensation value

   

   is calculated from the provisional compensation value sum and the correction values n̂^(j);

   g) the final compensated measuring signal values
   
   ${\text{sk}}^{\text{(i)}}{\text{= x}}^{\text{(i)}}{\text{- A}}^{\text{(i)}}\text{. sum' (3')}$

   

   
   
   are determined for each measuring channel by means of the final compensation value sum'.
2. Method according to Claim 1, characterised in that the provisionally compensated measuring signal values sk⁽ⁱ⁾ and the compensation coefficients A⁽ⁱ⁾ are used in accordance with the relationship:

   

   to search for a strongly disturbed measuring channel (j).
3. Method according to Claim 1, characterised in that the provisionally compensated measuring signal values sk⁽ⁱ⁾ and the compensation coefficients A⁽ⁱ⁾ are used in accordance with the relationship:

   

   to search for two weakly disturbed measuring channels (ℓ,m).
4. Method according to Claim 1, characterised in that the correction value n̂^(j) corresponds to the approximate value ${\text{ñ}}^{\text{(j)}}\text{. (}{\hat{\text{n}}}^{\text{(j)}}{\text{= ñ}}^{\text{(j)}}\text{) (7a)}$

   
5. Method according to Claim 1, characterised in that the correction value n̂^(j) is determined from the approximate value ñ^(j) as a function of a prescribable threshold value n_o ^(j):

   
6. Method according to one of the preceding claims, characterised in that the compensation coefficients A⁽ⁱ⁾ either are known empirical values of constant magnitude or are determined from the measuring signals x⁽ⁱ⁾:
   
   ${\text{A}}^{\text{(i)}}{\text{= f {x}}^{\text{(i)}}\text{}. (9)}$

   
7. Method according to Claim 6, characterised in that the compensation coefficients A⁽ⁱ⁾ are calculated adaptively from the measured values x⁽ⁱ⁾ and approximate values ñ⁽ⁱ⁾:
   
   ${\text{A}}^{\text{(i)}}{\text{= f {(x}}^{\text{(i)}}{\text{- n}}^{\text{(i)}}\text{)}. (10)}$