JP2018205152A - Signal processor - Google Patents

Abstract

To provide a signal processor that realizes high versatility and can improve reliability of a transmission line without interrupting operation of a system.SOLUTION: A signal processor 1 includes: an input unit 12 to which at least one type of random signal flowing in a transmission line 5 is input; and an operation unit 13 in which a state of degradation of the transmission line 5 is determined on the basis of spectral data generated by a Fourier transformation of the random signal input by the input unit 12. The operation unit 13 determines the state of degradation of the transmission line 5 on the basis of a normal spectrum in which the transmission line 5 is presumed to be normal and an abnormal spectrum used for determining the state of degradation of the transmission line.SELECTED DRAWING: Figure 2

Description

  The present disclosure relates to a signal processing device.

  2. Description of the Related Art Conventionally, various signals are transmitted and received through a transmission line conforming to the Ethernet (registered trademark) standard represented by a wired LAN (Local Area Network). Such a transmission line is comprised from the electric wire by a twisted pair wire, for example. Therefore, depending on the routing environment of the transmission path, in addition to deterioration over time due to humidity or temperature, etc., it may be exposed to loads such as bending, so communication quality may deteriorate due to problems such as stripping or untwisting of wires. There is a risk of lowering.

  Therefore, as a countermeasure for the trouble by determining the state of the transmission line, there is proposed one in which a ground fault current waveform of the transmission line is subjected to spectrum analysis when a ground fault of the transmission line occurs ( For example, see Patent Document 1). Moreover, what monitors the insulation degradation state of the transmission line at the time of a ground fault accident is proposed (for example, refer patent document 2). Further, there has been proposed an apparatus in which an AC voltage is applied to an insulator of a power cable and an insulation deterioration state of the power cable is determined based on a residual charge in the insulator (see, for example, Patent Document 3).

Japanese Patent Laid-Open No. 06-289086 JP, 2006-206035, A JP, 2006-217948, A

  However, the conventional techniques as described in Patent Documents 1 to 3 cannot stop the operation of the system because it is necessary to stop the function of the system in order to perform the determination. In addition, the conventional technique as described above is poor in versatility because it is necessary to set a dedicated threshold value for each system. Therefore, in the prior art, high versatility is realized, and the reliability of the transmission path cannot be improved without interrupting system operation.

  The present disclosure has been made in view of such a situation, and realizes high versatility and improves the reliability of a transmission path without interrupting the operation of the system.

  A signal processing device according to one aspect of the present disclosure is a signal processing device to which at least one type of random signal flowing in a transmission path is input, and is input by an input unit to which the random signal is input and an input unit An operation unit that determines a degradation state of the transmission line based on spectrum data generated by Fourier transforming the random signal, and the spectrum data includes a normal spectrum in which the transmission line is assumed to be in a normal state; An abnormal spectrum used for determination of the deterioration state of the transmission line, and the calculation unit determines the deterioration state of the transmission line based on the normal spectrum and the abnormal spectrum. And

  According to the signal processing device which is one aspect of the present disclosure, the degradation state of the transmission path is determined based on the normal spectrum and the abnormal spectrum. Like the normal spectrum and the abnormal spectrum, the spectrum of the random signal flowing through the transmission line varies depending on the change in the relative dielectric constant or dielectric loss tangent of the transmission line. The relative dielectric constant or dielectric loss tangent of the transmission line changes due to deterioration or deformation of the coating of the electric wires constituting the transmission line. Therefore, by determining the deterioration state of the transmission line based on the spectrum of the random signal, it is possible to determine whether or not the deterioration or deformation of the covering of the electric wire constituting the transmission line has occurred. In addition, the spectrum of the random signal is a frequency component of the transmission line, and can be easily obtained by Fourier transforming the random signal, so that high versatility is achieved and system operation is interrupted. It can be obtained without. Further, the fluctuation of the spectrum of the random signal can be detected before the transmission line completely fails. Therefore, high versatility can be realized, and the reliability of the transmission path can be improved without interrupting the operation of the system.

  Further, in the signal processing device according to one aspect of the present disclosure, the spectrum data further includes a reference spectrum obtained based on a design value of the transmission path, and the calculation unit includes the reference spectrum and the normal spectrum. Based on a normal product Mahalanobis distance value obtained from a normal difference spectrum which is a difference and an abnormal product Mahalanobis distance value obtained from an abnormal difference spectrum which is a difference between the reference spectrum and the abnormal spectrum, the deterioration state of the transmission path Is preferable.

  According to this signal processing device, the deterioration state of the transmission path is determined based on the normal product Mahalanobis distance value and the abnormal product Mahalanobis distance value. The Mahalanobis distance used in the normal product Mahalanobis distance value and the abnormal product Mahalanobis distance value is a determination index of the deterioration state of the transmission path that varies according to the distance from the group in the normal state. The group in the normal state can be obtained based on the normal difference spectrum which is the difference between the reference spectrum obtained based on the design value of the transmission line and the normal spectrum obtained when the transmission line is assumed to be in the normal state. It is. Therefore, it is possible to accurately determine not only the initial failure of the transmission line but also the life of the initially deteriorated product of the transmission line.

  Further, in the signal processing device according to the present disclosure, the calculation unit includes the normal product Mahalanobis using, as unit space data, feature item data corresponding to a specific frequency among the integration data obtained by integrating the normal difference spectrum after being converted into absolute values. A distance value is obtained, and the abnormal product Mahalanobis distance based on the feature item data corresponding to a specific frequency and the unit space data among the integrated data obtained by converting the abnormal difference spectrum into an absolute value It is preferable to obtain a value.

  According to this signal processing device, feature item data that is integration data corresponding to a specific frequency is used as unit space data out of integration data obtained by integrating difference spectra such as normal difference spectra and abnormal difference spectra as absolute values. The Mahalanobis distance is calculated. By making the difference spectrum an absolute value, the positive and negative fluctuations of the difference spectrum can be unified on the positive side. By converting the absolute value of the difference spectrum into integrated data, even if a random signal is used, a change in the transmission path can be made obvious. By extracting the feature item data corresponding to the specific frequency from the integrated data, the integrated data corresponding to the specific frequency can be extracted as unit space data. Therefore, for example, if a portion having a remarkable feature is specified with a specific frequency in the integrated data, a portion having a remarkable feature can be set as unit space data. If the Mahalanobis distance is obtained from such feature item data, the Mahalanobis distance including a portion where the feature is remarkable can be obtained, and therefore, a change in the transmission path can be made obvious.

  According to one aspect of the present disclosure, high versatility can be realized, and the reliability of the transmission path can be improved without interrupting the operation of the system.

It is a figure which shows an example of the communication system containing the signal processing apparatus 1 to which this indication is applied. It is a figure which shows an example of the function with which the signal processing apparatus 1 is provided. FIG. 3 is a diagram illustrating an example of a random signal that flows through a transmission path 5. It is a figure which shows an example of the spectrum of the normal transmission path. It is a figure which shows an example of the spectrum of the abnormal transmission path. It is a figure which shows an example of the difference of a spectrum with the same time-axis waveform in the transmission line 5 and a normal spectrum. It is a figure which shows an example of the difference of a spectrum with the normal time of the transmission line 5 in a different time-axis waveform, and an abnormal spectrum. 3 is a block diagram showing an example of a functional configuration executed inside a microcomputer 131. FIG. 5 is a flowchart for explaining deterioration determination processing of the signal processing device 1. It is a flowchart explaining a normal product MD value generation process. It is a flowchart explaining abnormal goods MD value generation processing. It is a figure which shows an example of the difference of the spectrum of the normal spectrum corresponding to a normal product, and the abnormal spectrum corresponding to a deteriorated product or a failure product. It is a figure which shows an example of a difference spectrum like the normal difference spectrum which is the difference of a normal spectrum and a reference spectrum, or the abnormal difference spectrum which is the difference of an abnormal spectrum and a reference spectrum. 3 is a diagram illustrating an example of a broken wire of a transmission line 5. FIG. It is a figure which shows an example of the stripping of the transmission line. It is a figure which shows an example of untwisting between the transmission lines. It is a figure which shows an example of the normal integration data which is integration data of a normal difference spectrum. It is a figure which shows an example of the abnormal integration data which are integration data of an abnormal difference spectrum. It is a figure which shows an example of the feature item of normal integration data. It is a figure which shows an example of the feature item of abnormality integration data. It is a figure which shows the example of determination of the degradation state of the transmission line 5 by the normal goods Mahalanobis distance value by RT method, and the abnormal goods Mahalanobis distance value. It is a figure which shows the example of determination of the degradation state of the transmission line 5 by the Mahalanobis distance value for every degradation or failure factor.

  Hereinafter, although embodiments of the present disclosure will be described based on the drawings, the present disclosure is not limited to the following embodiments. It should be noted that “consisting”, “consisting of”, “including”, “including”, “having”, “comprising”, or any other synonym thereof, covers non-exclusive inclusion relationships. Intended. For example, a process, method, article or device that includes an enumeration of elements is not necessarily limited to those elements, and other elements that are not explicitly enumerated or should be inherent It may be included in such a process, method, article or apparatus. Further, unless explicitly stated, “or” is inclusive and not an exclusive sum. For example, “condition A or B” is satisfied in any case where A exists and B does not exist, A does not exist and B exists, and both A and B exist.

  The present disclosure also relates to an apparatus for performing the processes described herein. The device may be specially constructed depending on the required purpose, or it may be constructed on a general purpose computer that is selectively activated or reconfigured by a computer program stored on the computer. Such a computer program may be stored in a computer-readable storage medium, which is not limited to a floppy disk, an optical disk, a CD-ROM, a magnetic optical disk, a read only memory, a random access memory, an EPROM. Any type of disk such as EEPROM, magnetic or optical card, etc. may be included, and any type of media indicated above for storing electronic instructions may be included. Are each coupled to a computer system bus.

  FIG. 1 is a diagram illustrating an example of a communication system including a signal processing device 1 to which the present disclosure is applied. As shown in FIG. 1, the signal processing apparatus 1 receives a signal flowing through the transmission line 5 through the data line 7 by the detection unit 6 or the like. The signal processing device 1 transmits and receives various signals to and from the communication device 3 via the communication line 8. The transmission line 5 is a channel through which at least one type of random signal flows, and is constituted by, for example, a single wire or a stranded wire. The communication system includes a transmission line 5 and a communication device 3 in addition to the signal processing device 1 and is mounted on a vehicle, for example. The transmission path 5 is installed and installed in the vehicle. Each of the signal processing device 1 and the communication device 3 is configured as an ECU in a vehicle, for example, and has different functions.

  FIG. 2 is a diagram illustrating an example of functions provided in the signal processing device 1. The signal processing device 1 includes an input unit 12, a calculation unit 13, an output unit 14, and a power supply 16. The input unit 12 is an interface through which various signals such as a digital input signal and an analog input signal can be input and transmitted to the calculation unit 13. The input unit 12 includes an input buffer 121 and an AD converter 122. The input buffer 121 converts a digital input signal into a signal level that can be input to the microcomputer 131, and temporarily stores the digital input signal when the digital input signal is input. The signal temporarily stored in the input buffer 121 can be read out by the arithmetic unit 13. The AD converter 122 converts an analog input signal into a digital value that can be input to the microcomputer 131. The AD converted signal can be read out by the arithmetic unit 13. The calculation unit 13 includes a microcomputer 131 and a semiconductor memory 132 and executes various processes on the signal input to the input unit 12. For example, when a random signal flowing through the transmission path 5 is input to the input section 12, the calculation section 13 is based on spectrum data generated by performing a Fourier transform on the random signal input to the input section 12. Determine the degradation state. The output unit 14 includes an output driver 141 and a communication unit 142, and outputs various signals based on the execution results of various processes by the calculation unit 13. The power supply 16 supplies a stable voltage to the input unit 12, the calculation unit 13, and the output unit 14. Note that the power source 16 is preferably highly accurate when it is also used as a reference voltage for the AD converter 122. The microcomputer 131 calculates a control amount from various input signals and outputs it. The semiconductor memory 132 is composed of, for example, an EEPROM, and stores various control amounts and data of the microcomputer 131. The output driver 141 converts the signal into a signal that can drive the actuator, such as an actuator drive signal, or amplifies the voltage in accordance with a signal output from the microcomputer 131. The communication unit 142 has driver and receiver functions. As a driver function, the communication unit 142 converts output data of the microcomputer 131 into a communication signal suitable for the communication standard. As a function of the receiver, the communication unit 142 converts a signal transmitted from another ECU into a level that can be input to the microcomputer 131.

  FIG. 3 is a diagram illustrating an example of a random signal flowing through the transmission line 5. Since the random signal as shown in FIG. 3 has been converted into a random signal by the scrambler, a waveform having the same pattern is not observed. Therefore, since the signal pattern, timing, and amplitude are completely random signals by simple waveform comparison, it is impossible to perform various evaluations using eye patterns. Accordingly, various fluctuation components with respect to time such as amplitude or jitter cannot be evaluated. Therefore, it is preferable to perform comparison by frequency components by converting a time axis waveform with amplitude change with respect to time into a frequency axis waveform with amplitude change with respect to frequency. For example, FFT (Fast Fourier Transform) is preferable as the process for converting the time axis waveform into the frequency axis waveform and generating the spectrum data.

  FIG. 4 is a diagram illustrating an example of the spectrum of the normal transmission path 5. FIG. 5 is a diagram illustrating an example of the spectrum of the abnormal transmission path 5. As shown in FIGS. 4 and 5, although the overall trend of the spectrum is approximated, the fine amplitudes are different and it is difficult to compare with only one spectrum. Further, for example, the coating 52 of the line constituting the transmission path 5 is worn away by rubbing, a part of the coating 52 is broken, the twisted pair line is untwisted, or the twist pitch is changed. As long as the transmission path 5 is changed, there is no problem in the communication function. However, if the symptom progresses as it is, a sign of a failure leading to a communication error, that is, the influence of the deterioration of the transmission line 5 only appears to be buried in the fine amplitude of the spectrum.

  FIG. 6 is a diagram illustrating an example of a difference between a spectrum in which the transmission line 5 is normal and an abnormal spectrum with the same time-axis waveform. FIG. 7 is a diagram illustrating an example of a difference between a spectrum in which the transmission path 5 is normal and an abnormal spectrum with different time-axis waveforms. 6 and 7 show the absolute values of the difference in spectrum obtained by Fourier transforming the time axis waveform that has passed through the normal transmission line 5 and the time axis waveform that has passed through the transmission line 5 in which the coating 52 of the line is peeled off by 2 mm. An example of a graph in which is plotted is shown. Specifically, FIG. 6 shows the difference in spectrum obtained by flowing the same time-axis waveform through the normal transmission line 5 and the abnormal transmission line 5, respectively. As shown in FIG. 6, since a difference of several dB appears at 66 MHz and 80 MHz or more, it can be understood that this is an influence of 2 mm of coating peeling. On the other hand, FIG. 7 shows the difference in spectrum obtained by flowing different time axis waveforms through the normal transmission line 5 and the abnormal transmission line 5, respectively. As shown in FIG. 7, there is a spectral difference across the frequency from 0 to 125 MHz. Therefore, the characteristics due to the difference in the transmission path 5 cannot be grasped at all.

  Therefore, it may be difficult to read a minute state change of the transmission path 5 on the frequency axis. However, the above description is a result of trying to read a minute change of the transmission line 5 with one spectrum. Therefore, a description will be given of reading minute changes in the transmission path 5 using a plurality of spectra in an arbitrary frequency bandwidth.

  FIG. 8 is a block diagram illustrating an example of a functional configuration executed inside the microcomputer 131. As shown in FIG. 8, a microcomputer 131 causes a signal observation processing unit 1301A, a waveform extraction processing unit 1302A, an FFT processing unit 1303A, an averaging processing unit 1304A, a measurement data selection processing unit 1305A, a difference processing unit 1306A, and an absolute value processing unit. Functions of 1307A, integration processing unit 1308A, feature item processing unit 1309A, sensitivity and standard S / N ratio processing unit 1310A, two-variable processing unit 1311A, unit space center calculation processing unit 1313, and MD value processing unit 1312A are realized. In addition, the microcomputer 131 causes the signal observation processing unit 1301B, the waveform extraction processing unit 1302B, the FFT processing unit 1303B, the averaging processing unit 1304B, the measurement data selection processing unit 1305B, the difference processing unit 1306B, the absolute value processing unit 1307B, and the integration processing unit. 1308B, the feature item processing unit 1309B, the sensitivity and standard SN ratio processing unit 1310B, the two-variable processing unit 1311B, and the MD value processing unit 1312B are realized. Further, the microcomputer 131 implements the function of the deterioration determination processing unit 1314.

  In addition, when any one of the signal observation processing unit 1301A and the signal observation processing unit 1301B is not particularly limited, it is referred to as a signal observation processing unit 1301. Further, when any one of the waveform extraction processing unit 1302A and the waveform extraction processing unit 1302B is not particularly limited, it is referred to as a waveform extraction processing unit 1302. Further, when any one of the FFT processing unit 1303A and the FFT processing unit 1303B is not particularly limited, it is referred to as an FFT processing unit 1303. Further, when any one of the averaging processing unit 1304A and the averaging processing unit 1304B is not particularly limited, it is referred to as an averaging processing unit 1304. Further, when any one of the measurement data selection processing unit 1305A and the measurement data selection processing unit 1305B is not particularly limited, it is referred to as a measurement data selection processing unit 1305. Further, when any one of the difference processing unit 1306A and the difference processing unit 1306B is not particularly limited, it is referred to as a difference processing unit 1306. Further, when any one of the absolute value processing unit 1307A and the absolute value processing unit 1307B is not particularly limited, it is referred to as an absolute value processing unit 1307. Further, when any one of the integration processing unit 1308A and the integration processing unit 1308B is not particularly limited, the integration processing unit 1308 will be referred to as an integration processing unit 1308. Further, when any one of the feature item processing unit 1309A and the feature item processing unit 1309B is not particularly limited, it is referred to as a feature item processing unit 1309. When any one of the sensitivity and standard S / N ratio processing unit 1310A and the sensitivity and standard S / N ratio processing unit 1310B is not particularly limited, the sensitivity and standard S / N ratio processing unit 1310 is referred to as a sensitivity and standard S / N ratio processing unit 1310. When any of the two-variable processing unit 1311A and the two-variable processing unit 1311B is not particularly limited, it is referred to as a two-variable processing unit 1311. Further, when any one of the MD value processing unit 1312A and the MD value processing unit 1312B is not particularly limited, it is referred to as an MD value processing unit 1312.

  The signal observation processing unit 1301 acquires a random signal flowing through the transmission path 5. The waveform extraction processing unit 1302 extracts a part of the waveform of the random signal acquired by the signal observation processing unit 1301. The FFT processing unit 1303 executes a Fourier transform of a part of the waveform of the random signal extracted by the waveform extraction processing unit 1302. The averaging processing unit 1304 performs averaging processing on the spectrum with a specific frequency bandwidth in the range of several KHz to several MHz in the spectrum subjected to Fourier transform by the FFT processing unit 1303, and reduces the influence of external noise superimposed on the spectrum. To do. The measurement data selection processing unit 1305 performs data selection processing on a plurality of averaged spectra averaged by the averaging processing unit 1304. Specifically, the measurement data selection processing unit 1305 selects an averaged spectrum to be stored for each same frequency among a plurality of averaged spectra. Note that it is preferable to increase the number of types of normal spectra by repeatedly executing the signal observation processing unit 1301A to the measurement data selection processing unit 1305A in different environments such as various temperatures, humidity, or vibrations. The difference processing unit 1306A is a normal spectrum if a normal spectrum is generated by the signal observation processing unit 1301A, the waveform extraction processing unit 1302A, the FFT processing unit 1303A, the averaging processing unit 1304A, and the measurement data selection processing unit 1305A. And a normal difference spectrum which is a difference from the reference spectrum. The difference processing unit 1306B is an abnormal spectrum if an abnormal spectrum is generated by the signal observation processing unit 1301B, the waveform extraction processing unit 1302B, the FFT processing unit 1303B, the averaging processing unit 1304B, and the measurement data selection processing unit 1305B. And an abnormal difference spectrum which is a difference from the reference spectrum. The reference spectrum is obtained based on the design value of the transmission path 5.

  The absolute value processing unit 1307A converts the normal difference spectrum into an absolute value. The integration processing unit 1308A obtains integrated data of the normal difference spectrum that has been converted to an absolute value. The feature item processing unit 1309A extracts data of a specific frequency from the integrated data as feature item data. The sensitivity and standard S / N ratio processing unit 1310A obtains the sensitivity and standard S / N ratio using the feature item data as unit space data. The two-variable processing unit 1311A converts all members of the unit space data into two variables based on the sensitivity and the standard SN ratio. The unit space center calculation processing unit 1313 obtains the center of the unit space from two variables. The MD value processing unit 1312A obtains a normal Mahalanobis distance value that is a distance from the center of the unit space for each member of the unit space.

  The absolute value processing unit 1307B converts the abnormal difference spectrum into an absolute value. The integration processing unit 1308B obtains integration data of the abnormal difference spectrum that has been converted to an absolute value. The feature item processing unit 1309B extracts data of a specific frequency from the integrated data as feature item data. The sensitivity and standard S / N ratio processing unit 1310B obtains the sensitivity and standard S / N ratio using the feature item data as signal data. The two-variable processing unit 1311B converts all members of the signal data into two variables based on the sensitivity and the standard S / N ratio. The MD value processing unit 1312B obtains an abnormal product Mahalanobis distance value that is a distance from the center of the unit space for each member of the signal data.

  The deterioration determination processing unit 1314 performs deterioration determination based on the normal product Mahalanobis distance value, the abnormal product Mahalanobis distance value, and the threshold value. Specifically, the deterioration determination processing unit 1314 determines the deterioration state of the transmission line 5 according to whether or not the abnormal product Mahalanobis distance value is in a normal range specified by the normal product Mahalanobis distance value, and determines the deterioration state. The process ends. The normal range is determined by the upper limit value of the normal product Mahalanobis distance value. In addition, if the abnormal product Mahalanobis distance value does not fall within the normal range by defining the deterioration range, abnormal range, and failure range in the abnormal range, the abnormal product Mahalanobis distance value may be degraded, abnormal, or faulty. Can be determined. Therefore, if the first threshold value is set as the threshold value between the normal range and the abnormal range, it can be determined whether or not the abnormal product Mahalanobis distance value belongs to the normal range. Furthermore, if the second threshold value is set as a threshold value between the deterioration range and the abnormal range in the abnormal range, it is determined whether the abnormal product Mahalanobis distance value belongs to either the deterioration range or the abnormal range. can do. Furthermore, if the third threshold value is set as a threshold value between the abnormal range and the failure range in the abnormal range, it is determined whether the abnormal product Mahalanobis distance value belongs to either the abnormal range or the failure range. can do.

  FIG. 9 is a flowchart for explaining the deterioration determination process of the signal processing apparatus 1. FIG. 10 is a flowchart for explaining normal product MD value generation processing. FIG. 11 is a flowchart for explaining abnormal product MD value generation processing. FIG. 12 is a diagram illustrating an example of a spectral difference between a normal spectrum corresponding to a normal product and an abnormal spectrum corresponding to a deteriorated product or a failed product. FIG. 13 is a diagram illustrating an example of a difference spectrum such as a normal difference spectrum that is a difference between a normal spectrum and a reference spectrum or an abnormal difference spectrum that is a difference between an abnormal spectrum and a reference spectrum. FIG. 14 is a diagram illustrating an example of a broken wire in the transmission path 5. FIG. 15 is a diagram illustrating an example of the stripping of the transmission path 5. FIG. 16 is a diagram illustrating an example of untwisting between transmission lines 5. FIG. 17 is a diagram illustrating an example of normal integration data that is integration data of normal difference spectra. FIG. 18 is a diagram illustrating an example of abnormality integration data that is integration data of an abnormality difference spectrum. FIG. 19 is a diagram illustrating an example of characteristic items of normal integration data. FIG. 20 is a diagram illustrating an example of the feature item of the abnormality accumulation data. FIG. 21 is a diagram illustrating a determination example of the deterioration state of the transmission path 5 based on the normal product Mahalanobis distance value and the abnormal product Mahalanobis distance value by the RT method. FIG. 22 is a diagram illustrating a determination example of the deterioration state of the transmission path 5 based on the Mahalanobis distance value for each cause of deterioration or failure.

  The process of step S11 to step S13 is a deterioration determination process. The process of step S31 to step S44 is a normal product MD value generation process. The process of step S61 to step S73 is an abnormal product MD value generation process. In addition, the process of step S31-step S35 and the process of step S61-step S65 are the same processes. Therefore, the description about the process of step S61-step S65 is abbreviate | omitted.

  In step S <b> 11, the calculation unit 13 executes normal product MD value generation processing. Details of the normal MD value generation process will be described later. In step S <b> 12, the calculation unit 13 executes abnormal product MD value generation processing. Details of the abnormal product MD value generation processing will be described later. In step S <b> 13, the calculation unit 13 performs deterioration determination based on a normal product Mahalanobis distance value that is a normal product MD value, an abnormal product Mahalanobis distance value that is an abnormal product MD value, and a threshold value.

  In step S <b> 31, the calculation unit 13 acquires a random signal that flows through the transmission path 5. In step S32, the calculation unit 13 extracts a part of the waveform of the random signal. In step S33, the calculating part 13 performs a Fourier transform. In step S34, the calculation unit 13 performs an averaging process on the spectrum with the specific frequency bandwidth. In step S35, the calculating part 13 performs a data selection process with respect to several averaged spectrum. In step S36, the calculation unit 13 stores the execution result of the data selection process in the normal database as a normal spectrum. In step S <b> 37, the calculation unit 13 obtains a normal difference spectrum that is a difference between the normal spectrum stored in the normal database and the reference spectrum.

  In step S38, the calculation unit 13 converts the normal difference spectrum into an absolute value. In step S39, the calculating part 13 calculates | requires the integration data of the normal difference spectrum made into the absolute value. In step S40, the calculating part 13 extracts the thing of a specific frequency among integrated data as feature item data. In step S41, the calculating part 13 calculates | requires a sensitivity and a standard S / N ratio by making feature item data into data of unit space. In step S42, the calculation unit 13 converts all members of the unit space data into two variables based on the sensitivity and the standard SN ratio. In step S43, the calculation unit 13 obtains the center of the unit space from two variables. In step S44, the calculation unit 13 obtains a normal product Mahalanobis distance value that is a distance from the center of the unit space for each member of the unit space, and ends the normal product MD value generation process.

  In step S66, the calculation unit 13 sets the execution result of the data selection process as an abnormal spectrum. In step S67, the calculation unit 13 obtains an abnormal difference spectrum that is a difference between the abnormal spectrum and the reference spectrum.

  In step S68, the calculation unit 13 converts the abnormal difference spectrum into an absolute value. In step S69, the calculating part 13 calculates | requires the integration data of the abnormal difference spectrum made into the absolute value. In step S70, the calculating part 13 extracts the thing of a specific frequency among integrated data as feature item data. In step S <b> 71, the calculation unit 13 obtains sensitivity and standard S / N ratio using the feature item data as signal data. In step S72, the calculation unit 13 converts all members of the signal data into two variables based on the sensitivity and the standard S / N ratio. In step S73, the calculation unit 13 obtains an abnormal product Mahalanobis distance value that is a distance from the center of the unit space for each member of the signal data, and ends the abnormal product MD value generation process.

  In other words, as shown in FIG. 12, the spectral difference between the normal spectrum and the abnormal spectrum is very small. Therefore, the calculation unit 13 uses the reference spectrum obtained based on the design value of the transmission path 5 for determining the deterioration state of the transmission path 5. The calculation unit 13 obtains a normal product Mahalanobis distance value from the normal difference spectrum that is the difference between the reference spectrum and the normal spectrum, and obtains an abnormal product Mahalanobis distance value from the abnormal difference spectrum that is the difference between the reference spectrum and the abnormal spectrum. As shown in FIG. 13, a difference spectrum such as a normal difference spectrum and an abnormal difference spectrum shows a difference, but since the difference is 3 dB or less at the maximum, the difference is very small as it is, and the difference can be shaken positively or negatively. In addition to appearing, there is a difference in phase or amplitude with respect to degradation or failure scale. Therefore, it is not possible to evaluate uniformly as it is. Therefore, as shown in FIG. 17 and FIG. 18, difference data such as a normal difference spectrum and an abnormal difference spectrum are converted into absolute values and then integrated to generate integrated data. AWG_Err_Leng in FIG. 20 shows a case where a part of the copper wire 511 included in the strand 51 such as the strands 51A, 51B, 51C is disconnected as shown in FIG. In Rubber_Gap of FIG. 20, as shown in FIG. 15, the case where the coating is peeled off due to the change in the distance between the conductors 53 covered with the coating 52 is shown. As shown in FIG. 16, “Fray_Length” in FIG. 20 shows a case where untwisting occurs due to the distance between the conductors 53. In this way, by integrating the normal difference spectrum and the abnormal difference spectrum after making them absolute values, the normal difference spectrum is integrated after converting the absolute difference values to absolute values, and then integrating them. A minute change in the transmission path 5 appears remarkably. Further, among the integrated data, the normal item Mahalanobis distance value is obtained using the feature item data corresponding to the specific frequency as shown in FIG. 19 as unit space data, and the feature item data corresponding to the specific frequency as shown in FIG. 20 is obtained. An abnormal product Mahalanobis distance value is obtained as signal data. As a result, as shown in FIG. 21, the distributions of the unit space data in the normal population and the signal data corresponding to deterioration or failure are clearly identified. Therefore, the deterioration state of the transmission line 5 can be determined according to whether or not the Mahalanobis distance is in the normal range. Further, as shown in FIG. 22, the abnormal product Mahalanobis distance value increases as the length of the degradation or failure scale increases. For example, even if the length of the degradation or failure scale is 1 mm, the abnormal product Mahalanobis distance value is 10 or more, so that it is possible to detect the degradation or failure even if the transmission path 5 is minutely changed. That is, the arithmetic unit 13 can determine the deterioration state of the transmission path 5 based on spectrum data generated by Fourier transforming a random signal received by the communication unit 142.

  From the above description, the signal processing device 1 to which the present disclosure is applied determines the degradation state of the transmission path 5 based on the normal spectrum and the abnormal spectrum. Like the normal spectrum and the abnormal spectrum, the spectrum of the random signal flowing through the transmission line 5 varies due to a change in the relative dielectric constant or dielectric loss tangent of the transmission line 5. The relative dielectric constant or dielectric loss tangent of the transmission line 5 changes due to deterioration or deformation of the line coating 52 constituting the transmission line 5. Therefore, by determining the deterioration state of the transmission line 5 based on the spectrum of the random signal, it is possible to determine whether the line covering 52 constituting the transmission line 5 has deteriorated or deformed. Further, the spectrum of the random signal is a frequency component of the transmission path 5 and can be easily obtained by Fourier transforming the random signal, so that high versatility is realized and the operation of the system is interrupted. It can be obtained without. Further, the fluctuation of the spectrum of the random signal can be detected before the transmission line 5 completely fails. Therefore, high versatility can be realized, and the reliability of the transmission path 5 can be improved without interrupting system operation.

  The signal processing apparatus 1 also detects the abnormal product Mahalanobis obtained from the normal product Mahalanobis distance value obtained from the normal difference spectrum that is the difference between the reference spectrum and the normal spectrum and the abnormal difference spectrum that is the difference between the reference spectrum and the abnormal spectrum. Based on the distance value, the deterioration state of the transmission path 5 is determined. The Mahalanobis distance used in the normal product Mahalanobis distance value and the abnormal product Mahalanobis distance value is a determination index of the deterioration state of the transmission line 5 that varies according to the distance from the group in the normal state. The group in the normal state is obtained based on the normal difference spectrum that is the difference between the reference spectrum obtained based on the design value of the transmission line 5 and the normal spectrum obtained when the transmission line 5 is assumed to be in the normal state. It can be done. Therefore, not only the initial failure of the transmission line 5 but also the life of the initially deteriorated product of the transmission line 5 can be accurately determined.

  Further, the signal processing device 1 obtains a normal product Mahalanobis distance value using characteristic item data corresponding to a specific frequency as unit space data among the integrated data obtained by converting the normal difference spectrum into absolute values, The abnormal product Mahalanobis distance value is obtained based on the feature item data corresponding to the specific frequency and the unit space data among the integrated data obtained by converting the difference spectrum into absolute values. By converting the difference spectra such as the normal difference spectrum and the abnormal difference spectrum into absolute values, the positive and negative fluctuations of the difference spectrum can be unified on the positive side. By converting the absolute value of the difference spectrum into integrated data, even if a random signal is used, a change in the transmission path 5 can be made obvious. By extracting the feature item data corresponding to the specific frequency from the integrated data, the integrated data corresponding to the specific frequency can be extracted as unit space data. Therefore, for example, if a portion having a remarkable feature is specified with a specific frequency in the integrated data, a portion having a remarkable feature can be set as unit space data. If the Mahalanobis distance is obtained from such feature item data, the Mahalanobis distance including a portion where the feature is remarkable can be obtained, so that the change in the transmission path 5 can be made obvious.

  The signal processing apparatus 1 to which the present disclosure has been applied has been described based on the embodiments. However, the present disclosure is not limited thereto, and modifications may be made without departing from the spirit of the present disclosure.

  For example, the Mahalanobis distance is described as an example of calculation using the RT method, but the present invention is not limited to this, and may be calculated using the APRT (Accumulated Principal component RT) method.

  Moreover, although the transmission line 5 is an electric wire such as a stranded wire or a single line and an example in which a signal flowing through the transmission line 5 is input from the data line 7 to the signal processing device 1 is described, the present invention is not limited to this. For example, the transmission path 5 may be a printed wiring in the device, and a signal flowing in the transmission path 5 may be extracted from the printed wiring in the device.

DESCRIPTION OF SYMBOLS 1 Signal processor 12 Input part, 121 Input buffer, 122 AD converter 13 Calculation part, 131 Microcomputer 1301,1301A, 1301B Signal observation processing part 1302,1302A, 1302B Waveform extraction processing part 1303,1303A, 1303B FFT processing part 1304,1303A , 1304B Averaging processing unit 1305, 1305A, 1305B Measurement data selection processing unit 1306, 1306A, 1306B Difference processing unit 1307, 1307A, 1307B Absolute value processing unit 1308, 1308A, 1308B Integration processing unit 1309, 1309A, 1309B Feature item processing unit 1310, 1310A, 1310B Sensitivity and standard S / N ratio processing unit 1311, 1311A, 1311B Two-variable processing unit 1312, 1312A, 1312B MD value processing unit 313 Unit space center arithmetic processing unit 1314 Degradation determination processing unit 132 Semiconductor memory 14 Output unit, 141 Output driver, 142 Communication unit 16 Power source 3 Communication device 5 Transmission path, 51, 51A, 51B, 51C Wire, 511 Copper wire 52 Coating , 53 Conductor 6 Detector, 7 Data line, 8 Communication line

Claims (3)

A signal processing device to which at least one type of random signal flowing in a transmission line is input,
An input unit to which the random signal is input;
Based on spectrum data generated by Fourier transforming the random signal input by the input unit, an arithmetic unit that determines the degradation state of the transmission path;
With
The spectral data is
A normal spectrum in which the transmission path is assumed to be in a normal state;
An abnormal spectrum used to determine the deterioration state of the transmission path;
Including
The computing unit is
Based on the normal spectrum and the abnormal spectrum, the deterioration state of the transmission path is determined.
A signal processing apparatus. The spectral data is
A reference spectrum obtained based on a design value of the transmission line,
Further including
The computing unit is
A normal product Mahalanobis distance value obtained from a normal difference spectrum which is a difference between the reference spectrum and the normal spectrum; and an abnormal product Mahalanobis distance value obtained from an abnormal difference spectrum which is a difference between the reference spectrum and the abnormal spectrum; Based on the above, the deterioration state of the transmission path is determined.
The signal processing apparatus according to claim 1. The computing unit is
Among the integrated data obtained by integrating the normal difference spectrum after making it an absolute value, the normal item Mahalanobis distance value is obtained using feature item data corresponding to a specific frequency as unit space data,
Among the accumulated data obtained by integrating the abnormal difference spectrum after making it an absolute value, the abnormal item Mahalanobis distance value is obtained based on the feature item data corresponding to a specific frequency and the unit space data.
The signal processing apparatus according to claim 2.

Images