JP2010050810A - Noise distortion correction filter for communication line and method of correcting noise distortion for communication line - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a noise distortion correction filter for a communication line for effectively correcting time waveform distortion occurring in a communication signal flowing in the communication line by electromagnetic noise from the outside, and to provide a method of correcting noise distortion for the communication line. <P>SOLUTION: A signal separation section 12 detects the communication signal flowing in the communication line 200 and a detection signal detected by a sensor section 17 as a plurality of detection signals, uses an independent component analysis method to calculate a plurality of estimation signals estimated as an original signal and an electromagnetic noise signal from the plurality of detection signals, and separates the plurality of estimation signals from the plurality of detection signals as the plurality of separation signals. A similarity calculation section 14 calculates similarity between respective detection and separation signals. A signal application section 15 decides separation signals corresponding to the respective detection each based on the similarity. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a communication line noise distortion correction filter and a communication line noise distortion correction method for correcting a communication signal on which a distortion component due to an electromagnetic noise signal is superimposed to an original signal.

  In recent years, electromagnetic interference caused by unnecessary radio waves (electromagnetic noise) generated from radio waves and inverter devices has become a problem. When unwanted radio waves are guided to a communication line, there is a possibility that electromagnetic interference may occur in the communication equipment connected to the end of the communication line and communication quality may be deteriorated. Tends to regulate. An example of such regulations is “IEC / CISPR Pub.22” established by the International Commission on Radio Interference. However, in addition to the decrease in electromagnetic immunity of communication devices due to the increase in speed and power reduction of electronic filters, for example, power line carrier communication (commonly known as “PLC (Power Line Communication)”) that uses power lines as communication lines. As a new communication form is widely used, the electromagnetic induction problem for communication lines is becoming more and more complicated.

  Conventionally, as a means for preventing electromagnetic induction to such a communication line, a method using a common mode noise filter for a communication line (see Patent Document 1), a power circuit of a connected device and a communication circuit when a lightning surge enters, A technique using a method for suppressing a situation in which a potential difference is generated (see Patent Literature 2), a method for detecting an output signal of a communication circuit and improving an S / N ratio (see Patent Literature 3), etc. 1) ") has been proposed. Such a technique can effectively prevent electromagnetic induction when an unnecessary radio wave / electromagnetic interference source that causes electromagnetic induction is clear or when an electromagnetic induction signal in a specific frequency band is removed.

On the other hand, when a plurality of signals are superimposed on the detection signal detected on the communication line, the original signal fluctuation before superposition is estimated using an independent component analysis method (see Non-Patent Document 1 to Non-Patent Document 3). A technique (hereinafter referred to as “second technique”) has been proposed. Also, an apparatus and a program for applying this independent component analysis method to an actual environment have already been disclosed (see Patent Documents 4 to 7).
Japanese Utility Model Publication No. Hei 6-29226 Japanese Patent Laid-Open No. 11-225433 Japanese Patent No. 3321640 Japanese Patent No. 3887247 JP 2004-110404 A JP 2005-258363 A JP 2006-243664 A JP 2007-206037 A Aapo Hyvarinen, “Survey on Independent Component Analysis”, Neural Computing Surveys 2, 1999, p.94-128 A. Hyvarinen, 1 other, "Independent component analysis: algorithms and applications", Neural Networks 13 (4-5), 2000, p.411-430 Aapo Hyvarinen, 1 other, Iku Nemoto, Maki Kawakatsu, “Independent Component Analysis”, Tokyo Denki University Press, 2005, 1st edition, chapters 7-12 Yoshiro Ehara, “Users Digital Signal Processing”, Tokyo Denki University Press, 1991, 1st edition, p.46, p.150 Miyamoto Sadaaki, “Introduction to Cluster Analysis”, Morikita Publishing, 1999, 1st edition, p.5, p.84, p.49

  However, with regard to the first technique, there has been a problem that it is necessary to perform a process of optimizing the filter according to the type of noise or a process of appropriately selecting a suitable filter. For example, in the case of a conventional LC type noise filter, since it has a resonance frequency peculiar to a circuit element, a problem that a desired attenuation cannot be obtained in a certain frequency band, or in the case of a noise filter that cuts a low frequency band in particular, There was also a problem that the element was enlarged. In addition, there is a problem in that it is not possible to flexibly cope with fluctuations in noise frequency. Furthermore, there is a problem that it is difficult to cope with an event in which the frequency band of unnecessary radio waves and electromagnetic interference waves that cause electromagnetic induction cannot be specified in advance.

  In addition, with regard to the second technique, the purpose is to separate superimposed audio signals, biological signals, and the like into individual original signals, to improve the estimation accuracy of the original signals, and to increase the efficiency of calculation processing. Therefore, the original signal is not estimated from a signal on which a distortion component due to the influence of noise is superimposed. Further, even if the original signal is estimated and a plurality of estimated signals estimated as the original signal are separated from a plurality of detected signals respectively detected on a plurality of communication lines as a plurality of separated signals, When the component analysis method is applied, the relationship (detection order of detection signals and estimation order of separation signals) between each detection signal and each separation signal is not guaranteed, that is, any separation signal is assigned to any detection signal. There was a problem that the “orderedness of the separated signals” as to whether or not it was supported was not ensured.

  The present invention has been made in view of the above, and a communication line noise distortion correction filter and a communication line noise that effectively correct a time waveform distortion generated in a communication signal flowing through the communication line due to external electromagnetic noise. It is an object to provide a distortion correction method.

  The present invention according to claim 1 detects the electromagnetic noise signal that causes the distortion component to act in a noise distortion correction filter for a communication line that corrects the communication signal on which the distortion component due to the electromagnetic noise signal is superimposed to an original signal. Detection means, the communication signal flowing through the communication line, and the detection signal detected by the detection means are detected as a plurality of detection signals, respectively, and the original signal and the electromagnetic wave are detected from the plurality of detection signals using an independent component analysis technique. Calculating a plurality of estimated signals estimated as noise signals, separating the plurality of estimated signals as a plurality of separated signals from the plurality of detection signals, the plurality of detection signals, and the plurality of separated signals; And storing the plurality of detection signals and the plurality of separated signals from the storage unit, and calculating the similarity between each detection signal and each separated signal, respectively. Similarity calculating means; and a signal applying means for determining a separation signal corresponding to each detection signal based on the similarity, and applying a separation signal corresponding to the detection signal corresponding to the communication signal to the communication line; It is summarized as having.

  In the present invention, the signal separation means detects the communication signal flowing through the communication line and the detection signal detected by the detection means as a plurality of detection signals, respectively, and uses the independent component analysis technique to obtain the original from the plurality of detection signals. In order to calculate a plurality of estimated signals estimated as signals and electromagnetic noise signals and separate the plurality of estimated signals from a plurality of detection signals as a plurality of separated signals, the superimposed distortion components are removed to remove the original signal and the electromagnetic signal It can be corrected to a noise signal. Further, the similarity calculation means calculates the similarity between each detection signal and each separation signal, and the signal application means determines the separation signal corresponding to each detection signal based on the similarity, respectively. The order of the separated signals can be ensured. From the above, it is possible to effectively correct the time waveform distortion generated in the signal flowing through the communication line due to external electromagnetic noise.

  According to a second aspect of the present invention, the similarity calculation unit processes the cross-correlation coefficient between the detection signal and the separated signal and the cross-correlation coefficient between the detection signal and the separated signal. Value, Euclidean distance between the detection signal and the separation signal, Minkowski distance between the detection signal and the separation signal, Mahalanobis distance between the detection signal and the separation signal, the detection signal and the separation Manhattan distance between signals, distance between the detection signals calculated by hierarchical clustering and the separated signals, distance between the detection signals calculated by the c-average method, and k average Based on either the distance between the detection signal calculated by the method and the separated signal, or the distance between the detection signal calculated by the fuzzy c-average method and the separated signal. And gist calculating a degree.

  According to a third aspect of the present invention, the frequency characteristic of the detection unit and the corresponding correction value are associated and stored in the storage unit, and the corresponding correction value corresponding to the frequency of the detection signal is stored from the storage unit. The gist of the invention is to further include a correction unit that reads and corrects the detection signal.

  According to a fourth aspect of the present invention, the signal separation means associates the frequency characteristic of the communication line with the corresponding correction value and stores it in the storage unit in advance, and is detected as a communication signal flowing through the communication line. The gist is to read a corresponding correction value corresponding to the frequency of the detected signal from the storage unit, correct the detected signal, and then calculate an estimated signal estimated as the original signal.

  According to a fifth aspect of the present invention, the signal applying means adjusts the magnitude of the separated signal applied to the communication line according to the magnitude of the amplitude of the detection signal corresponding to the communication signal. Is the gist.

  In the present invention according to claim 6, the communication line is a balanced line or a power line, a predetermined communication device is arranged at the end of the plurality of communication lines, and the communication signal is arranged by the communication device. The communication signal is terminated and output to the signal separation means when traveling in the forward direction, and further includes control means that does not terminate the communication signal when traveling in the reverse direction. .

  According to a seventh aspect of the present invention, in the communication line noise distortion correction method for correcting a communication signal on which a distortion component due to an electromagnetic noise signal is superimposed to an original signal, the electromagnetic noise signal that causes the distortion component to act is detected. The first step, the communication signal flowing through the communication line, and the detection signal detected in the first step are detected as a plurality of detection signals, respectively, and the original signal is detected from the plurality of detection signals using an independent component analysis technique. A second step of calculating a plurality of estimation signals estimated as a signal and the electromagnetic noise signal, and separating the plurality of estimation signals as a plurality of separation signals from the plurality of detection signals; the plurality of detection signals; A third step of storing a plurality of separated signals, and reading the plurality of detected signals and the plurality of separated signals from the storage means, and the similarity between each detected signal and each separated signal And a separation signal corresponding to each detection signal is determined based on the similarity, and a separation signal corresponding to the detection signal corresponding to the communication signal is applied to the communication line. And having 5 steps.

  According to the present invention, it is possible to provide a communication line noise distortion correction filter and a communication line noise distortion correction method for effectively correcting time waveform distortion generated in a communication signal flowing through the communication line due to external electromagnetic noise. it can.

  First, in the case where a distortion component is superimposed on each communication signal by inducing an electromagnetic noise signal to one or more communication signals flowing in a balanced line or a power line used as one or more communication lines, Detects the time fluctuation of each communication signal having the distortion component and detects the detection signal (electromagnetic noise signal) detected by the sensor (hereinafter, the detected signal is referred to as “detection signal”) and is independent. While performing correction to remove distortion components using the component analysis method, the respective similarities between each corrected signal (hereinafter referred to as “separated signal”) and each detected signal are calculated, and this An object of the present invention is to provide a communication line noise distortion correction filter and a communication line noise distortion correction method for determining a separation signal corresponding to each detection signal based on the similarity. Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<Configuration of noise distortion correction filter for communication line>
FIG. 1 is a block diagram showing functional blocks of a communication line noise distortion correction filter according to the present embodiment. The communication line noise distortion correction filter 100 includes a first circulator unit 11, a signal separation unit 12, a storage unit 13, a similarity calculation unit 14, a signal application unit 15, and a second circulator unit 16. The sensor unit 17 is provided. Then, it is arranged immediately before the communication device 400 connected to one or more communication lines 200, in other words, installed in a state of being inserted between the one or more communication lines 200 and the communication device 400, It has a function of correcting the communication signal flowing through the communication line 200 on which the distortion component is superimposed by the action of the electromagnetic noise source 300 to the original signal and then outputting it to the communication device 400.

  The first circulator unit 11 terminates a communication signal that flows through one communication line 200 and has a distortion component superimposed by an electromagnetic noise signal generated by the electromagnetic noise source 300, and displays the terminated communication signal. A function of controlling the direction of signal transmission to be transmitted to the signal separation unit 12 is provided. When a communication signal is transmitted from the communication device 400, the communication signal is output to the communication line 2 as it is without terminating the communication signal. That is, when the communication signal proceeds in the forward direction in which the communication device 400 is arranged, the communication signal is terminated and output to the signal separation unit 12, and when the communication signal proceeds in the reverse direction, the communication signal is not terminated. It has a function.

  The sensor unit 17 is electrically connected to the outside of the communication line noise distortion correction filter 100 and has a function of detecting an electromagnetic noise signal that may have an influence of a distortion component on the communication line 200. In order to detect the same noise that has acted on the communication line 200, it is desirable to arrange it near the communication line 200. In addition, as such a sensor part 17, when detecting the signal radiated | emitted in space, when detecting the common mode noise of the communication line 200, it detects a signal fluctuation on a current probe and other conductors. In this case, a voltage probe or the like can be used.

  The signal separation unit 12 inputs one or more communication signals transmitted from the first circulator unit 11 and the detection signals detected by the sensor unit 17, and N (N is the number of communication lines 200 and the sensor unit). N detection signals corresponding to the number of 17 and a natural number equal to or greater than 2) are calculated as N detection signals that are estimated as original signals and electromagnetic noise signals from the N detection signals using an independent component analysis technique. In addition, it has a function of separating N estimated signals as N separated signals from N detected signals. The independent component analysis method will be described later.

  The storage unit 13 has a function of storing the N detection signals detected by the signal separation unit 12 and the separated N separation signals.

  The similarity calculation unit 14 has a function of reading out N detection signals and N separation signals from the storage unit 13 and calculating the similarity between each detection signal and each separation signal. The similarity calculation procedure will be described later.

  The signal application unit 15 determines a separation signal corresponding to each detection signal based on the similarity calculated by the similarity calculation unit 14, and supplies the separation signal corresponding to the detection signal corresponding to the communication signal to the communication line 200. It has a function to apply. The procedure for determining the separation signal will be described later.

<Independent component analysis method>
The independent component analysis method is a method for estimating an original signal and an electromagnetic noise signal from a detected detection signal. For example, a method based on maximization of non-Gaussianity, a method based on maximum likelihood estimation, and a mutual information amount It is possible to select and calculate at least one of a method, a method based on non-correlation with a nonlinear function, a method based on a tensor, and the like (see Non-Patent Documents 1 to 3). In this <Independent Component Analysis Method>, for convenience of explanation, “original signal and electromagnetic noise signal” are collectively referred to as “original signal”.

  Specifically, when N original signals s before separation are expressed by equation (1), N detection signals x are expressed by equation (2), and N separated signals y are expressed by equation (3), The relationship between the detection signal x and the original signal s can be expressed by Equation (4), and the relationship between the separation signal y and the detection signal x can be expressed by Equation (5). However, A is an unknown matrix and W means a matrix to be estimated.

s = (s ₁ , s ₂ ,..., s _N ) ^T Expression (1)
_{However, s 1 = {s 1-1,} s 1-2, ..., s 1-m}, s 2 = {s 2-1, s 2-2, ..., s 2-m}, ..., s N = {S _N−1 , s _N−2 ,..., S _N−m }, where m represents the number of samples of each data constituting the original signal.

x = (x ₁ , x ₂ ,..., x _N ) ^T Expression (2)
_{However, x 1 = {x 1-1,} x 1-2, ..., x 1-m}, x 2 = {x 2-1, x 2-2, ..., x 2-m}, ..., x N = {X _N−1 , x _N−2 ,..., X _N−m }, and m means the number of samples of each data constituting the detection signal.

_{y = (y 1, y 2} , ..., y N) T ··· formula (3)
_{However, y 1 = {y 1-1,} y 1-2, ..., y 1-m}, y 2 = {y 2-1, y 2-2, ..., y 2-m}, ..., y N = {Y _N−1 , y _N−2 ,..., Y _N−m }, where m represents the number of samples of each data constituting the separated signal.

x = As (4)
y = Wx Formula (5)
Among the variables that constitute expressions (1) to _{(5), s, s 1} , s 2, ..., s N, x, x 1, x 2, ..., x N, y, y 1, y ₂ ,..., y _N , A, W are vectors (and so on).

The specific processing of the independent component analysis method first detects N detection signals x (step S101). Next, the matrix W is calculated using the equation (5) with the independence from the separated signal y as an evaluation measure (step S102). Thereafter, W≈A− ¹ is set, and the separation signal y is calculated by substituting the calculated matrix W into the equation (5) (step S103). Since the separated signal y approximates the original signal s, the relationship of s = A ⁻¹ x≈Wx = y is established. In particular, for the specific processing algorithm in step S102, the above-described method based on maximization of non-Gaussianity, the method based on maximum likelihood estimation, the method based on mutual information, the method based on non-linear function non-correlation, the method based on tensor, and the like. Will be used.

For example, the data constituting the detection signal x ₁ is described as {x _1-1 , x _1-2 ,..., X _1-m }, but this is indicated by x _1-1 or x _1-2. This means that the individual data such as are not detected at intervals of time, but m data detected at the sampling interval τ are continuous signals obtained at mτ time. The same applies to the original signal s and the separated signal y, and the signal that is finally output to the communication device 400 is a continuous signal.

<Similarity calculation procedure>
As a method for calculating the similarity, a cross-correlation coefficient between each detection signal and each separation signal can be used. Here, the N detection signals are expressed as x ₁ = {x _1-1 , x _1-2 ,..., X _1-m }, x ₂ = {x _2-1 , x _{2-2 ,. m}, ..., x N =} {x N-1, x N-2, ..., x N-m} ( where, m is the number of samples of each data constituting the detection signal), and the N number of separation signals _{y 1 = {y 1-1, y} 1-2, ..., y 1-m}, y 2 = {y 2-1, y 2-2, ..., y 2-m}, ..., y N = { y _N−1 , y _N−2 ,..., y _N−m } (where m is the number of samples of each data constituting the separated signal), the cross-correlation coefficient ρ is defined by equation (6) can do.

_{ρ (x, y) = E} [(x i -E [x i]) (y i -E [y i])] / σ xi σ yi ··· formula (6)
However, i = 1, 2,..., N, E [] represents the average value of each data, and σ represents the standard deviation of each data. By using Equation (6), it is possible to calculate the similarity between each detection signal x _i and each separation signal y _i .

  Note that, instead of the cross-correlation coefficient, a processed value of the cross-correlation coefficient between the detection signal and the separated signal (for example, an addition value obtained by adding a predetermined value, a subtraction value, a multiplication value, a division value, a square value, Reciprocal, etc.), distance between detection signal and separated signal (Euclidean distance, Minkowski distance, Mahalanobis distance, Manhattan distance), cluster type analysis method (hierarchical clustering, c-average method, k-average method, fuzzy c-average method, etc.) It is also possible to set the similarity by using any one or more of the distances between the detection signal and the separation signal calculated by (1). A technique using such a cross-correlation coefficient is disclosed in Non-Patent Document 4, and a distance between a detection signal and a separation signal (Euclidean distance, Minkowski distance, Mahalanobis distance, Manhattan distance) or a cluster method analysis method ( Patent Document 8 and Non-Patent Document 5 disclose a technique using a distance between a detection signal and a separation signal calculated by hierarchical clustering, c-average method, k-average method, fuzzy c-average method, and the like. .

<Procedure for determining separation signal>
As described above, when the independent component analysis method is applied, the relevance (detection order of detection signals and estimation order of separation signals) between each detection signal and each separation signal is not guaranteed. It is not guaranteed that the “separation signal order” that corresponds to which detection signal corresponds to. Therefore, in order to appropriately filter out a plurality of separated signals, the order of the separated signals y corresponding to the detected signals x is determined using the similarity calculation result between the detected signals x and the separated signals y.

Specifically, when N detection signals x are set to the above-described equation (2) and N separation signals y are set to the above-described equation (3), as shown in FIG. Each similarity calculated with each separated signal y is compared (step S201). More specifically, the similarity 1 between the detection signals x ₁ detected by the communication line 200 of the first run and the separation signals y _1, similarity 2 between the detection signal x ₁ and the separation signals y ₂ When, ... and compares the similarity N between the detection signal x ₁ and the separation signal y _N.

  Next, as a result of the comparison in step S202, the separation signal y having the highest similarity to the detection signal x is determined as the separation signal corresponding to the detection signal x (step S202). When the similarity is calculated based on the distance, the separation signal with the shortest distance may be determined.

  Subsequently, the magnitude (amplitude level) of the determined separated signal y is adjusted according to the magnitude (amplitude level) of the detection signal x (step S203).

  Finally, the adjusted separation signal y is applied (filter output) to the communication line 200 where the detection signal x is detected (step S204).

  By performing steps S201 to S204 for all the detection signals, it is possible to determine a separation signal corresponding to each detection signal.

  According to the present embodiment, the signal separation unit 12 detects the communication signal flowing through the communication line 200 and the detection signal detected by the sensor unit 17 as a plurality of detection signals, respectively, and uses the independent component analysis technique to detect the plurality of detection signals. In order to calculate a plurality of estimated signals estimated as the original signal and the electromagnetic noise signal from the detection signal, and to separate the plurality of estimated signals as a plurality of separated signals from the plurality of detected signals, the superimposed distortion component is removed. It can correct | amend to an original signal and an electromagnetic noise signal. The similarity calculation unit 14 calculates the similarity between each detection signal and each separation signal, and the signal application unit 15 determines the separation signal corresponding to each detection signal based on the similarity. Therefore, the order of the separated signals can be ensured. From the above, it is possible to effectively correct the time waveform distortion generated in the signal flowing through the communication line due to external electromagnetic noise.

  In the present embodiment, the configuration using the two circulators of the first circulator unit 11 and the second circulator unit 16 has been described. However, the number of circulators is not limited to two. Further, in the present embodiment, the description has been made using one sensor unit 17, but the same effect can be obtained even when a plurality of sensor units 17 are used.

  FIG. 3 is a graph showing the waveforms of the detection signal and the separation signal corresponding to the detection signal in association with each other. That is, conventionally, with respect to a plurality of separated signals obtained by independent component analysis, the detection order with the detected detection signals has not been guaranteed, but according to the present embodiment, the separated signals are based on the similarity. Therefore, it is possible to output an appropriate separation signal to the communication line 200.

FIG. 4 shows the waveforms of the original signal (including the electromagnetic noise signal), the detection signal, and the separated signal in the case where induction due to external electromagnetic noise occurs in the communication signal flowing through the communication cable lines arranged in parallel. It is a graph of an example of an application result. s ₁ is an original signal waveform, s ₂ is an electromagnetic noise signal waveform, x ₁ is a detection signal waveform in which a communication signal is detected, x ₂ is a detection signal waveform in which a detection signal is detected, and y ₁ is a separated signal waveform corresponding to the communication signal. (corresponding to the original signal), y ₂ represents the separation signal waveform corresponding to the detection signal (corresponding to the electromagnetic noise signal). According to the graph shown in FIG. 4B, the waveform distortion occurs in the communication signal flowing through the communication cable due to electromagnetic noise from the outside, but at the same time the original signal is restored efficiently, Since the separation signal is output in accordance with the detection order, the original signal waveform can be restored (improvement of waveform distortion) as shown in FIG. 4C.

  FIG. 5 is a configuration diagram showing another functional block of the noise distortion correction filter for communication line according to the present embodiment. The communication line noise distortion correction filter 100 has a configuration in which an isolator 18 is inserted between the first circulator unit 11 and the signal separation unit 12 of the communication line noise distortion correction filter shown in FIG.

  The isolator 18 has an effect of suppressing signal reflection that may occur when a detected signal flows from the first circulator unit 11 to the signal separation unit 12. Thereby, it becomes possible to prevent the operation | movement stabilization and destruction in the 1st circulator part 11. FIG.

  FIG. 6 is a block diagram showing another functional block of the noise distortion correction filter for communication line according to the present embodiment. The communication line noise distortion correction filter 100 has a configuration in which a data correction unit 19 is inserted between the sensor unit 17 and the signal separation unit 12 of the communication line noise distortion correction filter shown in FIG.

  The sensor unit 17 that detects an electromagnetic wave or the like has a certain frequency characteristic, and it is generally known that the stability of the frequency characteristic decreases as the target frequency bandwidth increases. As a result, the detection signal detected by the signal separation unit 12 may include a certain waveform response distortion corresponding to the frequency characteristic of the sensor unit 17. Therefore, the data correction unit 19 associates the frequency characteristic of the sensor unit 17 with the corresponding correction value and stores it in advance in a storage unit (not shown), and corresponds to the frequency of the detection signal detected by the sensor unit 17. The corresponding correction value is read from the storage unit, and the detection signal is corrected. That is, the detection signal detected in the time domain is converted into the frequency domain, and the frequency characteristic of the sensor unit 17 that is grasped in advance with an arbitrary frequency resolution is used for the frequency domain data on the converted frequency domain. Can be corrected. For example, when the sensitivity of the sensor unit 17 decreases in an arbitrary frequency section, a process such as increasing the level of frequency domain data can be given as an example. Therefore, even if the fluctuation range of the frequency characteristic of the sensor unit 17 is large and a distortion component is included in the detection signal, it is possible to effectively distort the detection signal and accurately estimate the original electromagnetic noise signal. It becomes.

  In the signal separation unit 12, the frequency characteristic of the communication line 200 and the corresponding correction value are associated with each other and stored in advance in the storage unit, and correspond to the frequency of the detection signal detected as the communication signal flowing through the communication line 200. The same effect can be obtained by reading the corresponding correction value from the storage unit, correcting the detection signal, and calculating the estimated signal estimated as the original signal.

It is a block diagram which shows the functional block of the noise distortion correction filter for communication lines which concerns on this Embodiment. It is a flow which shows the determination procedure of a separated signal. It is a graph which correlates and shows a waveform of a detection signal and a separation signal corresponding to the detection signal. In the example of the application result shown in association with the waveforms of the original signal (including the electromagnetic noise signal), the detection signal, and the separation signal when induction due to external electromagnetic noise occurs in the communication signal flowing through the communication cable line arranged in parallel It is a graph. It is a block diagram which shows the other functional block of the noise distortion correction filter for communication lines which concerns on this Embodiment. It is a block diagram which shows the other functional block of the noise distortion correction filter for communication lines which concerns on this Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 ... 1st circulator part 12 ... Signal separation part 13 ... Storage part 14 ... Similarity calculation part 15 ... Signal application part 16 ... 2nd circulator part 17 ... Sensor part 18 ... Isolator 19 ... Data correction part 100 ... Communication line Noise distortion correction filter 200 ... communication line 300 ... electromagnetic noise source 400 ... communication equipment S101-S103, S201-S204 ... step

Claims (7)

In a communication line noise distortion correction filter that corrects a communication signal in which a distortion component due to an electromagnetic noise signal is superimposed on an original signal,
Detecting means for detecting the electromagnetic noise signal that causes the distortion component to act;
The communication signal flowing through the communication line and the detection signal detected by the detection means are detected as a plurality of detection signals, respectively, and the original signal and the electromagnetic noise signal are estimated from the plurality of detection signals using an independent component analysis technique. Signal separating means for calculating a plurality of estimated signals and separating the plurality of estimated signals as a plurality of separated signals from the plurality of detection signals;
Storage means for storing the plurality of detection signals and the plurality of separated signals;
The similarity calculation means for reading the plurality of detection signals and the plurality of separation signals from the storage means and calculating the similarity between each detection signal and each separation signal,
A signal applying unit that determines a separation signal corresponding to each detection signal based on the similarity, and applies a separation signal corresponding to the detection signal corresponding to the communication signal to the communication line;
A noise distortion correction filter for a communication line, comprising: The similarity calculation means includes:
A cross-correlation coefficient between the detection signal and the separated signal, a processed value of a cross-correlation coefficient between the detection signal and the separated signal, a Euclidean distance between the detection signal and the separated signal, The Minkowski distance between the detection signal and the separation signal, the Mahalanobis distance between the detection signal and the separation signal, the Manhattan distance between the detection signal and the separation signal, and the detection calculated by hierarchical clustering The distance between the signal and the separated signal, the distance between the detected signal and the separated signal calculated by the c-average method, and the distance between the detected signal and the separated signal calculated by the k-average method The similarity is calculated based on any one of the distances between the detection signal and the separation signal calculated by a fuzzy c averaging method. Line for noise distortion correction filter.   Correction that correlates the frequency characteristic of the detection means and the corresponding correction value in the storage means, reads out the corresponding correction value corresponding to the frequency of the detection signal from the storage means, and corrects the detection signal The communication line noise distortion correction filter according to claim 1 or 2, further comprising means. The signal separating means includes
The frequency characteristic of the communication line and the corresponding correction value are associated with each other and stored in advance in the storage unit, and the corresponding correction value corresponding to the frequency of the detection signal detected as the communication signal flowing through the communication line is read from the storage unit. 4. The communication line noise distortion correction filter according to claim 1, wherein an estimated signal estimated as the original signal is calculated after correcting the detection signal. 5. The signal applying means includes
5. The magnitude of the separation signal applied to the communication line is adjusted according to the magnitude of the amplitude of the detection signal corresponding to the communication signal. 6. Noise distortion correction filter for communication lines. The communication line is a balanced line or a power line, and a predetermined communication device is arranged at the end of the plurality of communication lines,
When the communication signal proceeds in the forward direction in which the communication device is arranged, the communication signal is terminated and output to the signal separation unit, and when the communication signal proceeds in the reverse direction, the communication signal is not terminated. The communication line noise distortion correction filter according to any one of claims 1 to 5, further comprising means. In a communication line noise distortion correction method for correcting a communication signal in which a distortion component due to an electromagnetic noise signal is superimposed to an original signal,
A first step of detecting the electromagnetic noise signal that causes the distortion component to act;
The communication signal flowing through the communication line and the detection signal detected in the first step are detected as a plurality of detection signals, respectively, and the original signal and the electromagnetic noise signal are detected from the plurality of detection signals using an independent component analysis technique. A second step of calculating a plurality of estimated signals estimated as follows, and separating the plurality of estimated signals as a plurality of separated signals from the plurality of detection signals;
A third step of storing the plurality of detection signals and the plurality of separated signals;
A fourth step of reading the plurality of detection signals and the plurality of separation signals from the storage means, and calculating the similarity between each detection signal and each separation signal;
A fifth step of determining a separation signal corresponding to each detection signal based on the similarity, and applying a separation signal corresponding to the detection signal corresponding to the communication signal to the communication line;
A noise distortion correction method for a communication line, comprising:

Images