JP2006010432A - Leak current detector and leak current detecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To precisely and quickly detect a resistive leak current, using only a signal obtained by ZCTs, as to a leak current detector and a leak current detecting method in an electric power supply system. <P>SOLUTION: The signal from a ZCT part 1 including the ZCT 1-ZCT m for detecting a current flowing in a grounding line is input to extract frequency information of the signal, a basic parameter required for identification of a leak system is extracted by a basic parameter extracting means (basic parameter extracting part 2), an error between the basic parameter and an output parameter is extracted by an error extracting means (error extracting part 4), the output parameter is corrected to to make the extracted error get closely near to zero so as to identify the leak system including a leak route, the resistive leak current is detected by a system identifying circuit means (system identifying circuit part 3), and a caution or an alarm of leakage generation is output based on a level and a continuing time of the detected resistive leak current. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a leakage current detection device and a leakage current detection method for detecting a resistive leakage current in the field of electrical equipment management or electrical construction.

  As the connection configuration of the primary and secondary windings of the three-phase transformer in various power supply facilities, for example, there is a ΔΔ connection shown in FIG. 24, and VAB, VBC, and VCA are the A phase on the primary side, B-phase and C-phase line voltages, IA, IB, and IC are currents of each phase, EU, EV, and EW are phase voltages of each phase, Eu, Ev, and Ew are phase voltages of each phase on the secondary side, Ia , Ib, Ic are currents of each phase, and Vab, Vbc, Vca are line voltages of the respective phases. In this case, the relationship is (VAB, VBC, VCA) = (EU, EV, EW), (Vab, Vbc, Vca) = (Eu, Ev, Ew). The primary-side phase voltages VAB, VBC, and VCA are typically 6.6 kV, and the secondary-side phase voltages Vab, Vbc, and Vca are typically 200 V and 100 V, for example.

  FIG. 25 shows the connection configuration of the primary and secondary windings of the transformer in the case of YY connection. The same reference numerals as those in FIG. 24 indicate the same names. In this case, the line voltage and the phase voltage of each phase are different. FIG. 26 shows a case of a three-phase four-wire system with a secondary side neutral point drawing in a YY connection configuration, where the same reference numerals as those in FIGS. 24 and 25 indicate the same names, and Vag, Vbg, Vcg represents a voltage between each phase on the secondary side and the neutral point. In this case, the voltage between the neutral point and each phase may be supplied as a single phase.

  FIG. 27 shows a configuration of YΔ connection in which the primary side of the transformer is Y-connected and the secondary side is Δ-connected, and the same reference numerals as those in FIGS. 24 and 25 denote the same names. FIG. 28 shows a case where the primary side of the transformer is Δ-connected and the secondary side is Y-connected, contrary to FIG. FIG. 29 shows a configuration of VV connection in which the primary side and the secondary side of the transformer are V-connected. As a secondary side connection configuration, there is a single-phase three-wire configuration, and power can be supplied as 200V and 100V between phases.

  In addition to the above-described configuration, the transformer connection configuration includes a staggered connection, a Scott connection, a Woodbridge connection, a modified Woodbridge connection, and the like. In addition to the three-phase three-wire system, the secondary power supply configuration of the transformer includes a single-phase two-wire system, a single-phase three-wire system, etc., and if any leakage occurs in any connection system, It is desired to detect the leakage quickly. As a leakage detection means for that purpose, the method of measuring insulation resistance in a non-live line state does not necessarily reflect the insulation state of the live line state, so the leak current detection method in the live line state is suitable. Yes. The leakage current in the live line state includes a capacitive leakage current flowing through the ground capacitance and a resistive leakage current flowing through the leakage resistance due to insulation degradation or the like.

As the leakage current detection method in the live line state, there are mainly three types, ie, the I0 (zero phase current) method, the Igr (resistive leakage current) method 1 and the Igr method 2. The first I0 method is a method of detecting leakage current using a current transformer called ZCT, and the second Igr method 1 is a resistive leakage current detection method, but a special oscillator is prepared. In this system, a signal from the oscillator is injected into a live line via a signal injection transformer, and the signal is detected. The Igr method 2 is a method of detecting a resistive leakage current by preparing a reference voltage phase (see, for example, Patent Documents 1 to 16).
Japanese Utility Model Publication No. 6-105276 JP-A-6-43196 JP 2001-242205 A Japanese Patent Laid-Open No. 5-180885 Japanese Utility Model Publication No. 6-57037 JP 2000-74979 A JP 2002-131362 A JP 2001-215247 A JP 2001-165971 A JP-A-8-182179 JP-A-8-182180 Japanese Patent Laid-Open No. 10-10184 Japanese Patent Laid-Open No. 10-78462 JP-A-10-339757 Japanese Utility Model Publication No. 7-29477 JP 2002-125313 A

  In a power supply system, it is desired to quickly detect a leakage current due to an insulation failure or a ground fault to avoid a fire accident or the like due to the leakage current. To detect the leakage current, it is necessary to solve the following points.

  First, the first problem is that an accurate resistive leakage current cannot be detected due to an increase in capacitive leakage current due to an increase in harmonic current caused by inverter equipment or the like. In the I0 system, which is one of the conventional systems described above, the total leakage current from the ZCT is viewed. Specifically, the combined value of the capacitive leakage current and the resistive leakage current is observed. Therefore, in this I0 system, when the capacitive leakage current increases, it becomes impossible to measure the accurate resistive leakage current. As described above, as described above, there is an increase in inverter devices due to progress in energy saving. Inverter equipment generates considerable high frequency noise due to the nature of the device. These high-frequency noises often add an anti-noise capacitor between the device and the ground in order to minimize outward radiation (for control measures of VCCI; Voluntary Control Council for Information). For this reason, in the past, the capacitive leakage current, in which the slight capacitance between the power cable and the ground, was dominant, has tended to increase significantly due to the increase in the number of inverter devices and the like. As a result, in the conventional I0 system, it is difficult to accurately detect the resistive leakage current.

  The second problem is that a power failure occurs when the leakage current detector is installed. As an improvement over the conventional I0 system, there is an Igr system 1. In this Igr system 1, a dedicated oscillator is provided, and this signal is injected between the transformer and the ground. In the case of the resistive leakage current, the accurate resistive leakage current is detected with the point that the in-phase leakage current is detected. However, since this detector has a low oscillation frequency of several tens of hertz, not only has a drawback that the signal injection transformer becomes large, but also a signal must be injected into the ground line between the transformer and the ground. It is necessary to carry out the work with a power failure at the time of installation. Recently, there are many facilities that operate 24 hours a day, and power outage work is one of the items that I want to avoid very much.

  The third problem is prevention of electric shock accidents when installing equipment. There is an Igr method 2 as a method for avoiding power outage work at the time of equipment installation. This Igr method 2 does not require signal injection like the Igr method 1, but conversely needs to take in a reference voltage from the outside. The Igr method 2 extracts a component having the same phase as the reference voltage by taking in the reference voltage from the power supply facility that performs the detection of the resistive leakage current, thereby enabling the highly accurate detection of the resistive leakage current. . However, since it is necessary to take in this reference voltage from the outside to the device, there is a possibility of causing an electric shock accident if the power outage work is not performed at the time of equipment installation. Therefore, from the viewpoint of preventing an electric shock accident, it is desirable to avoid such an operation of taking in the reference voltage from outside.

  The fourth problem is the miniaturization of the detector and ensuring of simplicity. When a resistive leakage current occurs, it is desired to isolate each location, and to isolate an actual leakage location, and if possible, a handy type small detector that does not require connection to a simple power line is desired. Therefore, for example, it is desired to enable detection by simply connecting the ZCT or simply touching the ZCT at the corresponding location. That is, it is desirable that the detector operates with an operating power source such as a battery.

  The fifth problem is that, as shown in FIGS. 24 to 29, when the total number of target electric circuits connected to the ZCT is 6 types, 2 single-phase lines and 3 single-phase lines, a total of 8 types are connected. When Scott connection / Woodbridge connection / deformed Woodbridge connection, etc. are added, there are 12 types or more in total. It is desired to be applicable to such various configurations.

  The sixth problem is to increase the processing time. When detecting electric leakage, a time guard called a detection time limit is generally provided. This is to prevent inadvertently reporting the occurrence of unnecessary leakage to the monitoring center and to ensure that necessary information can be reported. Since these temporal guards differ depending on the customer system, the detection device is optimized by setting. Specific detection time limit setting values include, for example, 10 seconds / 30 seconds / 1 minute / 3 minutes / 5 minutes / 10 minutes. For example, when the detection time limit is set to 10 seconds, notification is not made unless the leakage current exceeds a certain value for 10 seconds continuously. Further, the detection time limit accuracy is about ± 10%. For this reason, as the processing time required for detection, for example, when the detection time limit setting is 10 seconds, high-speed leakage current detection processing of about ± 1 second (within 2 seconds in width) is required. For example, when signals from three ZCTs are processed by one MPU (microprocessor), a high speed and high speed of about 2 (seconds) / 3 (pieces) = 0.66 (seconds) per ZCT. Detection of accuracy is required. In general, when the target electric circuit system is expanded and high accuracy is further pursued, the processing time tends to increase. One of the problems to be solved is to increase the processing time.

  An object of the present invention is to solve the problems in leakage current detection in the power supply system as described above.

  The leakage current detection apparatus of the present invention is a leakage current detection apparatus that detects a resistive leakage current by processing a signal from a ZCT, and inputs a signal from the ZCT that detects a current flowing through a ground line. The basic parameter extracting means for extracting the basic information necessary for identifying the earth leakage system by extracting the frequency information, the error extracting means for extracting the error between the basic parameter and the output parameter, and the error extracting means System identification circuit means for correcting the output parameter so as to bring the error closer to zero, identifying a leakage system including a leakage path, and estimating and detecting a resistive leakage current.

  The basic parameter extracting means extracts the frequency information of the signal from the ZCT, demodulates the signal from the ZCT according to the frequency information, generates a baseband n-order harmonic vector signal, and generates the n-order harmonic. Means for phase normalizing a vector signal and a basic pattern including the n-order harmonic vector signal phase-normalized by the means are extracted and input to the system identification means.

  The phase normalizing means of the basic parameter extracting means includes means for normalizing the phase of the nth-order harmonic vector signal and performing control processing so that the combined power of the nth-order harmonic vector signal becomes a predetermined value. It has a configuration. Regarding the combined power, there are a method of normalizing the combined power including the fundamental wave to 1.0 and a method of normalizing the combined power to 1.0 with a component not including the fundamental wave. However, if the composite power of the nth harmonic is calculated in a form excluding the fundamental wave and normalized to 1.0, the subsequent calculation process becomes easy. .

  The system identification means inputs a reference voltage generator for outputting an nth-order harmonic signal and a reference voltage from the reference voltage generator, and inputs an output parameter based on the identification of the leakage system to the error extraction means. And a leakage detecting unit that performs correction processing of the output parameter so that an error from the basic parameter is minimized.

  Further, the system identification means minimizes an error between a dictionary reference section holding a parameter indicating a plurality of leakage patterns, a parameter indicating the leakage pattern from the dictionary reference section, and a basic parameter from the basic parameter extraction means. The parameter indicating the leakage pattern is configured as an initial parameter of the output parameter from the leakage identification unit. This dictionary reference parameter includes cases where the fundamental wave is included and not included. When the fundamental wave is included, it is preferable to use the one in which the combined power including the fundamental wave is normalized to 1.0. Further, although it is a case where the fundamental wave is not included, it is preferable to use one normalized so that the combined power not including the fundamental wave becomes 1.0. Or, for example, a case where a signal obtained by normalizing the fifth-order power to 1.0 is considered. In any case, what is important is to aggregate the essential parameters as much as possible to minimize the size of the dictionary.

  The dictionary reference unit stores a parameter indicating the leakage pattern stored in advance and a parameter identified and converged by the leakage identification unit.

  The leakage identifying unit of the system identifying unit includes a correcting unit that corrects the error by the error extracting unit in correspondence with the n-th harmonic, and the error corrected by the correcting unit and the reference voltage generating unit. A leakage simulation circuit for inputting the nth-order harmonic signal and a leakage simulation circuit normalization unit for inputting an output pattern obtained by normalizing the signal from the leakage simulation circuit to the error extraction unit. is there.

  The leakage identification unit of the system identification means includes a configuration in which the resistive leakage coefficient is multiplied by a coefficient that decreases in accordance with the order of the nth harmonic signal, and the capacitive leakage coefficient is multiplied by a constant 1.

  In addition, the leakage identification unit of the system identification unit randomly selects an n-order harmonic component corresponding to the error by the error extraction unit or selects a fixed pattern that is optimal for the leakage pattern, and determines the correlation. It has a correction means for minimizing.

  The leakage current detection method of the present invention is a leakage current detection method for detecting a resistive leakage current by processing a signal from the ZCT, and inputting a signal from the ZCT for detecting a current flowing in the ground line. Extracting frequency information of the signal, extracting basic parameters necessary for identifying the leakage system based on the frequency information, and correcting the output parameters so that an error between the basic parameters and the output parameters is minimized. And performing the process of identifying the leakage system including the leakage path and detecting the resistive leakage current by the identification convergence of the leakage system.

  Also, the signal from the ZCT is demodulated with a reference frequency and converted into a vector signal by complex conjugate calculation, the vector signal is converted to obtain frequency information of the signal from the ZCT, and the signal from the ZCT is demodulated according to the frequency information. The n-th harmonic signal component is obtained, and the n-order harmonic signal component is normalized and used as the basic parameter.

  In addition, the n-th harmonic signal from the reference voltage generation unit is multiplied by a coefficient, and an error between the output parameter that has been corrected and the basic parameter is extracted, and the error is minimized so that the error is minimized. It includes a process of performing the correction process of the output parameter and identifying the leakage system.

  Further, the present invention includes a step of identifying the leakage system as an initial value of a coefficient by which a parameter indicating a plurality of leakage patterns from the dictionary reference unit is multiplied by the nth-order harmonic signal from the reference voltage generation unit.

  In addition, the process includes a process of performing correction processing that minimizes the correlation by randomly selecting and synthesizing n-order harmonic components corresponding to the error between the basic parameter and the output parameter.

  A signal from the ZCT is input, frequency information of the signal is obtained, a basic pattern based on an nth-order harmonic signal included in the signal is obtained, and an output pattern that minimizes an error from the basic pattern can be obtained. Thus, the leakage system identification including the leakage path is performed to detect the resistive leakage current. Therefore, first, even in a power supply system with a large capacitive leakage current, the resistive leakage is accurately performed. Current can be detected. Second, no power outage work is required to detect leakage current. Thirdly, since it is not necessary to take in the reference voltage from the power supply system, there is no electric shock accident during installation work. Fourth, since it can be realized by an arithmetic processing function such as a microprocessor, it can be operated with a battery as an operation power source, and it can be easily carried by downsizing. Fifth, it automatically recognizes the frequency of the power supply system, and all power supplies such as three-phase three-wire, three-phase four-wire, single-phase three-wire, single-phase two-wire, etc. It can be applied to system resistive leakage current detection. Sixth, since it is possible to converge the leakage system identification at high speed, it is possible to detect resistive leakage current at high speed.

  Referring to FIG. 1, the leakage current detection apparatus of the present invention processes a signal from the ZCT unit 1 including ZCT1 to ZCTm and detects a resistive leakage current. A basic parameter extracting means (basic parameter extracting section 2) for inputting a signal from the ZCT for detecting the signal and extracting frequency information of this signal to extract a basic parameter necessary for identifying the leakage system, and this basic parameter extracting means The error extraction means (error extraction unit 4) for extracting the error between the basic parameter and the output parameter from the output parameter, and the output parameter is corrected so that the error extracted by the error extraction means approaches zero, and the leakage path System identification circuit means (system identification circuit unit 3) for estimating and detecting a resistive leakage current.

  The leakage current detection method of the present invention is a leakage current detection method for detecting a resistive leakage current by processing a signal from ZCT, and a signal from a ZCT (ZCT section 1) for detecting a current flowing through a ground line. Is input, the frequency information of this signal is extracted, the basic parameters necessary for identifying the leakage system are extracted by the basic parameter extraction means (basic parameter extraction unit 2) based on the frequency information, and the basic parameters and output are extracted. The parameters are compared by the error extraction means (error extraction section 4), and the output parameter correction processing is performed by the system identification means (system identification circuit section 3) so that the error of the comparison result is minimized. It includes a process of identifying a leakage system including a leakage path and detecting a resistive leakage current by identification convergence of the leakage system.

  FIG. 1 is a diagram for explaining the principle of the present invention. As basic parameter extraction means for inputting an output signal of a ZCT unit 1 including m-phase zero-phase current detection transformers (hereinafter referred to as ZCT) ZCT1 to ZCTm. Basic parameter extraction unit 2, system identification circuit unit 3 as a system identification unit for inputting the basic parameters extracted by basic parameter extraction unit 2, error extraction unit 4 as an error extraction unit, and various setting / display controls ZCT1 to ZCTm for detecting current flowing in the ground line by providing ZCT1 to ZCTm on the ground lines between the secondary windings of the plurality of transformers and the ground in the power supply system. Shows a configuration for estimating and detecting the resistive leakage current. Note that ZCT includes a single case. As shown in the figure, when signals obtained by a plurality of ZCTs are input, a process for estimating and detecting a resistive leakage current is performed by time-division processing. It is.

  The output signals from the m ZCT1 to ZCTm of the ZCT unit 1 are input to the basic parameter extraction unit 2, and the amplitude / nth (= 1 to n) harmonics included in the output signal of the ZCT unit 1 / From the phase / frequency information, basic parameters including the power corresponding to the n-th harmonic necessary for identifying the leakage system including the leakage path are extracted. On the other hand, the system identification circuit unit 3 simulates an actual leakage system, and includes output parameters including the extracted basic parameters and the power corresponding to the n-th harmonic by the leakage system estimated by the system identification circuit unit 3. Are matched (the output of the error extraction means is zero), the error extraction unit 4 feeds back to the system identification circuit unit 3 to correct the coefficient, and the leakage system identification is performed. Using the identified coefficient, the resistive leakage current actually generated in the system to which the ZCT is connected is estimated and detected. Note that the detection threshold value and detection time limit of the detection output are operated according to various settings by the various setting / display control unit 5. As described above, the final Igr output attention / warning information can be obtained.

  Therefore, in the present invention, the basic parameter extraction means (basic parameter extraction section 2), the system identification circuit means (system identification circuit section 3), and the error extraction means (error extraction section 4) are provided and input from the ZCT. Based on the obtained signal, the basic parameter including the n-th harmonic power extracted by the basic parameter extraction unit (basic parameter extraction unit 2) matches the output parameter identified by the system identification unit (system identification circuit unit 3). Thus, the leakage system identification is performed, and the resistive leakage current Igr is detected based on the coefficient at the time of convergence of the leakage system identification. The function of each means can be realized by a microprocessor having an arithmetic processing function. Therefore, it is possible not only to accurately measure the resistive leakage current, which is the first problem described above, but also to eliminate the power outage during the installation work, which is the second problem, and the reference voltage, which is the third problem. Phase capture is also unnecessary, that is, it is possible to avoid the occurrence of electric shock accidents, and furthermore, the fourth type of handy-type portable configuration can be realized. It can be flexibly applied to the system, and the processing time, which is the sixth problem, can be increased.

  FIG. 2 is an explanatory diagram of an alarm system to which the present invention can be applied. An alarm detection transmission device 12 is provided in a power receiving facility (cubicle) 11 of a power consumer 13 such as a factory, office building, school, hospital or the like. In this alarm detection transmission device 12, when leakage detection is performed as an abnormality detection, this detection signal is sent to a security room or office 15 alarm via, for example, a 100V light line (power line carrier) or communication line. The information is transmitted to the reception notification device 14, and a leakage notification is sent to the monitoring center 16 via the telephone network or the like automatically or by a security guard. In the case of a guard, it is also possible to send a leakage notice to the monitoring center 16 by facsimile machine FAX or telephone. In the monitoring center 16, maintenance personnel can be dispatched by the emergency dispatch vehicle 17 or the like to repair the location where the abnormality has occurred.

  FIG. 3 is an explanatory diagram of ZCT connection. For example, a ZCT for grounding the b phase among the a, b, and c phases on the secondary side of the ΔΔ connection transformer and detecting a current flowing through the ground line is shown. Shows when connected. Further, ra, rb, and rc denote ground equivalent resistances, and ca, cb, and cc denote ground equivalent capacitances. Under normal conditions, the ground-to-ground equivalent resistances ra, rb, and rc show large values. Therefore, the resistive leakage current is a negligible value. The ground-to-ground equivalent capacities ca, cb, and cc include power factor improving capacitors and the like that are not shown. Furthermore, since there are many harmonic components due to inverter devices, etc., the capacitive leakage current is relatively low. A large value will be shown. Then, since the current of the vector sum of the current obtained by vector synthesis of the resistive leakage current of each phase and the vector synthesis of the capacitive leakage current of each phase flows to the ground line, this is detected by ZCT. . By detecting by this ZCT and inputting to the leakage current detection device, the leakage system identification including the state of the leakage path is performed, and the resistive leakage current is estimated and detected with high accuracy and speed.

  FIG. 4 shows a leakage detection system in which the configuration of the present invention shown in FIG. 1 is applied to the configuration shown in FIG. 3. As an example, the b phase among the a, b, and c phases is grounded by a ground wire. The case where a ground fault occurs in the c-phase when ZCT is connected to the ground line is shown. Therefore, the c-phase is connected to the ground equivalent resistance through which the resistive leakage current Igr flows and the ground equivalent capacitance through which the capacitive leakage current Igc flows, and the a-phase has a negligible magnitude between the ground equivalent resistance. Therefore, the illustration is omitted, and only the equivalent ground-to-ground capacity through which the capacitive leakage current Igc flows is illustrated. Further, the leakage current detection device for inputting a signal from the ZCT for detecting the current flowing through the grounding wire is, as shown in FIG. 1, the basic parameter extraction unit 2 connected to the ZCT and the system identification circuit unit. 3 and the error extraction unit 4, and the system identification circuit unit 3 detects the resistive leakage current Igr by performing the leakage system identification including the leakage path.

  In addition, a case where a single ZCT is connected to the leakage current detection device is shown, but as shown in FIG. 1, it is also possible to connect a plurality of ZCTs corresponding to the power supply system. Although the case where the leakage current detection device is arranged at a position close to the ZCT is shown, this leakage current detection device may be arranged at a remote location and connected to the ZCT via networks known for various transmission methods. Is possible. It is also possible to connect via a wireless network as well as a wired network. When signals from ZCT are transmitted via each network, they are transmitted to the leakage current detection device according to the format according to the transmission method by ZCT compatible addressing, etc., and the leakage current detection device identifies each ZCT. Thus, it is possible to detect the resistive leakage current of the ZCT-compliant power supply system.

  FIG. 5 shows a waveform obtained by simulation of a leakage current waveform when the ratio of the resistive leakage current in the case of the single phase is 2 and the capacitive leakage current is 2, which corresponds to 3 cycles of 50 Hz. There are many components, and the waveform is greatly distorted from the sine wave of the fundamental frequency.

  FIG. 6 is an explanatory diagram of a leakage current spectrum, in which the horizontal axis indicates frequency [Hz], the vertical axis indicates power [dBm], and the spectrum for 0 to 500 Hz in a 50 Hz distribution area. That is, it can be seen that a harmonic component of an integral multiple of 50 Hz is included.

  FIG. 7 shows an example of the ground fault pattern corresponding to the spectrum of the first, third, fifth and seventh harmonic components. (A) ('0200) is a single-phase capacitive leakage current. (B) ('1200) is a single-phase capacitive leakage current plus a resistive leakage current, and (c) (' 0202) is a three-phase three-wire system with only a capacitive leakage current In the case of (d) ('1202), when a resistive leakage current is applied to the a phase of the three-phase three-wire system, (e) (' 0212) is a three-phase three-wire system and the capacitive leakage current is The cases where the a leakage current occurs in the a phase and the c phase and the resistive leakage current occurs in the c phase are respectively shown. The vertical axis represents the power of the primary / third / fifth / seventh harmonics with the combined power normalized to 1.0. That is, the leakage current waveform is slightly different depending on various leakage patterns, and the spectrum is slightly different. Therefore, it is also possible to determine whether or not a resistive leakage current is included based on the power pattern of the nth harmonic component included in the signal from the ZCT.

  FIG. 8 is an explanatory diagram of an embodiment of the present invention showing the internal configuration of each part in FIG. 1, and the same reference numerals as those in FIG. 1 denote the same parts. In FIG. 8, 21 is a filter unit, 22 is a data acquisition unit, 23 is a reference value extraction unit, 24 is a frequency information extraction unit, 25 is an I0 detection unit, 31 is a dictionary reference unit, 32 is a reference voltage generation unit, Reference numeral 33 denotes a leakage detection unit, and 34 denotes an Igr determination / timer monitoring unit. The ZCT unit 1 shows a case of m pieces of ZCT1 to ZCTm, and unnecessary components are removed from the respective output signals by the filter unit 21. Then, necessary data is captured by the data capturing unit 22. Since these data differ in frequency from 50 Hz / 60 Hz depending on the region, the frequency information extraction unit 24 automatically and accurately extracts the power supply frequency. The reference value extraction unit 23 generates a frequency of the nth harmonic using the power supply frequency extracted by the frequency information extraction unit 24 as a fundamental frequency, and generates a power corresponding to the nth harmonic included in the signal from the ZCT unit 1. This is used as a basic pattern. The LPF normalized output obtained by normalizing the nth harmonic power as the basic pattern is input to the system identification circuit unit 3 and the error extracting unit 4, and the inverse LPF normalized output indicating the nth harmonic power or the first order ( The fundamental wave power value is input to the Igr determination / timer monitoring unit 34 of the system identification circuit unit 3.

  The I0 detection unit 25 uses the signal in the frequency information extraction unit 24 to generate a I0 (capacitance, resistance and leakage current combined zero-phase current; the zero-phase current flows to the three-phase neutral point. This is a circuit that detects a current flowing between the power supply line and the ground (known as a current), and inputs a detection output signal I0DEC to the Igr determination / timer monitoring unit 34, and the detection output signal I0DEC is , I0 (zero phase current) indicates a predetermined value or less, the subsequent Igr determination is stopped to prevent the malfunction of the resistive leakage current detection process. The reference value extraction unit 23 extracts basic parameters necessary for identifying the leakage system, and outputs them as shown as LPF normalization output and LPF denormalization output. Although only these basic parameters can be used to identify the leakage system including the leakage path, the dictionary reference unit 31 is used to hold the dictionary reference unit 31 in order to speed up the identification convergence. It is determined which of the various leakage patterns is closest, and the identification parameters of the leakage identification unit 33 are initialized using the information of the closest leakage pattern as the coefficient of the initial value. Alternatively, it is possible to store an identification parameter when the leakage system identification was performed in the past and set the identification parameter as an initial value.

  Further, in the system identification circuit unit 3, a reference voltage generation unit 32 that generates a reference voltage including n-order harmonics is provided in the system identification circuit 3 in order to virtually simulate and construct an actual leakage system. The actual leakage system is identified by the reference voltage generation unit 32 and the leakage identification unit 33. Specifically, the leakage identifying unit 33 is identified so that the error (extracted by the error extracting unit 4) between the parameter identified by the leakage identifying unit 33 and the basic parameter extracted by the reference value extracting unit 23 approaches zero. Do. Based on the coefficient when the system is finally identified, the Igr component is calculated, the Igr determination / timer monitoring unit 34 determines with a predetermined threshold value, and then monitors whether or not the set time is continued. The Igr output is sent as a leakage detection signal.

  FIG. 9 is a detailed block diagram of an embodiment of the present invention, and the same reference numerals as those in FIG. 8 denote the same parts. The filter unit 21 of the basic parameter extraction unit 2 includes a termination circuit 1 to a termination circuit m, a selection circuit SEL1, a gain switching unit GSW, and a low-pass filter LPF. Each ZCT1 to ZCTm of the ZCT unit 1 is terminated. The circuit 1 to the termination circuit m are terminated with a predetermined impedance, and are input to the selection circuit SEL1 as a voltage value. The selection circuit SEL1 selects a ZCT that is a leakage current detection target. For example, when selecting three ZCTs, the selection circuit SEL1 can be sequentially selected in units of 0.6 seconds. In the case of one ZCT, the selection circuit SEL1 can be omitted or fixedly selected.

  The gain switching unit GSW corresponds to a variable gain amplifier that switches the amplification gain for the ZCT output signal selected by the selection circuit SEL1. The ZCT output signal corresponds to a wide range of about 50 mA to 800 mA, for example, and the amplification gain for the ZCT output signal sequentially selected by the selection circuit SEL1 is switched in order to ensure a predetermined dynamic range. The selection circuit SEL1 and the gain switching unit GSW are controlled by the analog control unit of the data capturing unit 22.

  The data acquisition unit 22 has a configuration including an AD converter A / D, an offset removal circuit Offset, a buffer BUFF, and an analog control unit, and the frequency information extraction unit 24 includes a carrier generation unit CRR0, The demodulator includes a demodulator DEMO, a low-pass filter LPF0, a complex conjugate circuit, a frequency converter, and a carrier generator CRRn. The I0 detection unit 25 includes a power calculation unit PWR and a determination unit. The reference value extraction unit 23 includes demodulation units DEM1 to demn, low pass filters LPF1 to LPFn, and an LPF normalization unit. It has a configuration.

  The offset removal circuit Offset of the data capturing unit 22 calculates the DC offset component because each of the digital signals corresponding to ZCT1 to ZCTm converted by the AD converter A / D includes a DC offset component. Processing is performed so that the offset component becomes zero, and the offset component is temporarily stored in the buffer BUFF. Thereby, it is possible to further improve the accuracy of detecting the resistive leakage current by digital arithmetic processing. Then, the data is read from the buffer BUFF and input to the analog control unit, the demodulation units DEM1 to DEMn of the reference value extraction unit 23, and the demodulation unit DEM0 of the frequency information extraction unit 25.

  The analog control unit raises the gain of the gain switching unit GSW when the signal level input to the buffer BUFF is lower than the lower limit value of the predetermined range with respect to the signal level from the ZCT selected by the selection circuit SEL1, and conversely When the upper limit of the predetermined range is exceeded, gain switching control is performed so as to decrease the gain of the gain switching unit GSW. When the signal level input to the buffer BUFF is always within a predetermined range, the control structure for automatic gain switching can be omitted by setting the gain of the gain switching unit GSW as a preset value. The same applies when the AD converter A / D has a wide dynamic range.

  FIG. 10 is an explanatory diagram of the carrier generating unit CRR0 and the demodulating units DEM0 to DEMn. The carrier generating unit CRR0 of the frequency information extracting unit 24 has an unknown frequency of the fundamental wave of the signal from the ZCT, but the power supply system Since the fundamental frequency is generally 50 Hz or 60 Hz, for example, data for generating a carrier having an intermediate frequency of 55 Hz can be stored in the ROM. The ROM stored data in this case can be expressed as (cos (2π55 (Hz) t) −jsin (2π55 (Hz) t)) indicating vector rotation at a radius of 55 Hz.

  Further, the demodulating units DEM0 to DEMn of the frequency information extracting unit 24 and the reference value extracting unit 23 multiply the input signal by n and the cos (nω) as the harmonic order, and obtain the demodulated output (DEM output) of the real part. And a multiplier for multiplying -sin (nω) to obtain an imaginary part demodulated output (DEM output). In the frequency information extracting unit 24, the signal from the buffer BUFF is demodulated. It becomes a DEM input to DEM0, and carrier cos (2π55 (Hz) t) of carrier cos (2π55 (Hz) t) −jsin (2π55 (Hz) t) and sin (2π55 (Hz) t) from carrier generation unit CRR0 Are input to each multiplier through a path indicated by a dotted arrow, and the demodulated output signal is passed through the low-pass filter LPF0 to the complex conjugate circuit and the I0 detector 25. To enter into a power calculating unit PWR.

  For the demodulating units DEM1 to DEMn of the reference value extracting unit 23, signals from the buffer BUFF become respective DEM inputs, and carriers corresponding to orders from the carrier generating unit CRRn of the frequency information extracting unit 24 (as will be described later) An n-order harmonic signal synchronized with the extracted power supply frequency; cos (nω), −sin (nω)) is input to the multiplier for demodulation. As a result, the demodulated output signal corresponding to the nth-order harmonic signal indicates the nth-order harmonic component included in the signal from the ZCT, and the LPF normalization is performed via the low-pass filters LPF1 to LPFn, respectively. Enter in the department.

FIG. 11 shows an example of the configuration of the above-described low-pass filters LPF0 to LPFn. Multiplication for multiplying a unit time delay circuit T and an input signal and a signal via the delay circuit T by coefficients ω ₀ to ω _k , respectively. And the adder Σ, the configuration of a transversal FIR filter is shown. For example, since the low-pass filter LPF0 of the frequency information extraction unit 24 wants to remove unnecessary components other than the vicinity of 55 Hz, the bandwidth is about 55 Hz ± 25 Hz. Similarly, the other low-pass filters LPF1 to LPFn are set to a band corresponding to the nth-order harmonic, and the cutoff frequency is set to about ± 25 Hz with respect to the nth-order harmonic frequency, and unnecessary harmonic signals are removed. By using the carrier generation unit CRR0, the demodulation unit DEM0, and the low-pass filter LPF0 in the frequency information extraction unit 24, it is possible to extract a frequency between 30 Hz and 80 Hz. Therefore, the leakage signal can be extracted even before the frequency synchronization of 50 Hz / 60 Hz is established.

  FIG. 12 is an explanatory diagram of the I0 detection circuit 25. The I0 detection circuit 25 includes a power calculation unit PWR and a determination circuit. A demodulated signal is input through the low-pass filter LPF0, and the demodulated signal is squared by the power calculation unit PWR. The power is obtained, and the difference between the power and the preset value REF is obtained by the calculation unit of the judgment circuit. If the polarity is negative in the polarity judgment unit, there is a leakage current (I0DEC = 1.0). If the polarity is positive, a detection signal I0DEC indicating no leakage current (I0DEC = 0.0) is output and input to an Igr calculation LPF inverse normalization unit (see FIG. 9) of the Igr determination / timer monitoring unit 34 To do. In this case, in the case of the detection signal I0DEC having no leakage current, the resistance leakage current detection process after the Igr determination / timer monitoring unit 34 is not performed in order to prevent malfunction.

FIG. 13 is an explanatory diagram of the complex conjugate circuit and the frequency conversion unit in the frequency information extraction unit 24. The output signal of the low-pass filter LPF0 is input as a complex conjugate input to the multiplier and the delay circuit. In this case, the sampling interval signal converted by the AD converter A / D is input to the complex conjugate circuit via the buffer BUFF and further to the complex conjugate circuit via the demodulator DEM0 and the low-pass filter LPF0. input. When signals having this time difference are LPF01 and LPF02, respectively, the complex conjugate output Y is
Y = LPF02 × (LPF01) * (1)
It can be expressed as. Note that * indicates a complex conjugate operation.

  The complex conjugate output Y becomes a vector that shifts counterclockwise when the frequency of the signal input to the AD converter A / D is higher than 55 Hz, and conversely, when the frequency is lower than 55 Hz, the complex conjugate output Y becomes a vector that shifts clockwise. Become. As indicated by the dotted arrow, this vector signal is input to the frequency conversion unit as frequency information extraction input.

Specifically, the captured leakage current is detected at an arbitrary demodulation frequency f _DEM at an arbitrary time interval, and two phase vectors are extracted. Further, by extracting the phase difference at an arbitrary time interval from the two phase vectors obtained in the preprocessing, the difference between the demodulation frequency and the frequency of the capture current is extracted.

If the frequency of the captured leakage current is f _AC , the arbitrary demodulation frequency is f _DEM , and the deviation between the captured leakage current frequency and the arbitrary demodulation frequency is Δf,
f _AC = f _DEM + Δf (2)
It becomes.

A plurality of frequencies for demodulation are generated based on two pieces of frequency information of an arbitrarily determined demodulation frequency f _DEM and Δf obtained by preprocessing. When the primary frequency f _DEM1 , the tertiary frequency f _DEM3 , the quintic frequency f _DEM5 , and the seventh frequency f _DEM7 are obtained, the following equation is obtained.
f _DEM1 = f _DEM + Δf
f _DEM3 = 3 × (f _DEM + Δf)
f _DEM5 = 5 × (f _DEM + Δf)
f _DEM7 = 7 × (f _DEM + Δf)
Therefore, the general formula for the frequency of the nth harmonic is:
_{f DEMn = n × (f DEM} + Δf) ··· (3)
It can be expressed as.

The frequency information extraction circuit using the complex conjugate output as the frequency information extraction input includes a vector information → Δθ conversion circuit, a frequency information extraction circuit including a Δθ + 55 Hz addition MOD circuit, and a carrier generation unit CRRn. Vector information (X + jY) is input to the vector information → Δθ conversion circuit and converted into angle information Δθ (proportional to the frequency information of Δf) using a tan ⁻¹ function. Then, it is added (frequency addition) to the 55 Hz angle information generated by the carrier generation unit CRR0, and rotation angle information for each sample is obtained (± 180 degree mod addition). The carrier from the carrier generation unit CRRn A frequency information extraction output (rotation vector signal of CRRn = (cos (nωt) −jsin (nωt))) can be obtained.

  The frequency information extraction circuit may have other configurations, for example, a ROM or the like in which the complex conjugate signal (X + jY) is address-converted (upper X + lower Y) and the relationship between the address and the frequency is stored in advance. It is also possible to obtain a desired frequency information (for example, 51.1 Hz, 60.3 Hz, etc.) by accessing the above memory. Alternatively, the frequency information conversion may be configured such that the output of the complex conjugate circuit is converted into vector information having a radius of 1.0, and only the imaginary component is used to convert the output into desired frequency information using a ROM. Furthermore, the frequency may be calculated suddenly using a PLL circuit.

  FIG. 14 is an explanatory diagram of the LPF normalization unit of the reference value extraction unit 23, where 101 is a power synthesis unit, 102 is an inverse transform unit, 103 is a square unit, 104 is a division unit, 105 is a square root unit, and 106-109. Indicates a multiplier. The output signals of the low-pass filters LPF1 to LPFn are input to the power combining unit 101, squared, and added (Σ) to obtain the total power a, which is output as SYNLPFP, and at the inverse conversion unit 102 Then, by the 1 / √a process, the multiplier 106 multiplies the output signals of the low-pass filters LPF1 to LPFn as a level before obtaining the power, and normalizes them. Note that LPF (n) input to the multiplier 106 is shown as a representative for the sake of simplicity, and the illustrated configuration is provided for the low-pass filters LPF1 to LPFn.

Next, in order to normalize the phase, the square unit 103 squares it to R ² and outputs it as LPFA (n), and the division unit 104 sets 1 / R ² = b. The signal is input to the multiplier 109, multiplied by the control force α, and output as LPFAI (n). Further, in the square root part 105, √b = 1 / R = LPFAIS (n) is inputted to the multiplier 107, and this multiplication output is inputted to the multiplier 108 to perform a complex conjugate operation, and the real part Re is obtained. The signal LPFN (n) (LPF normalization output in FIG. 8) obtained by normalizing the level and phase is input to the error extraction unit 4, the dictionary reference unit 31, and the leakage identification unit 33.

  The basic parameters important for leakage signal identification will be explained. As shown in FIG. 5, the leakage waveform is a very complicated waveform, but it is clear that it is a repetitive waveform at the power supply frequency. The reference voltage phase is extremely important for resistive leakage current detection, but the waveform of the leakage current input to the AD converter A / D is complex as described above, and the reference phase cannot be specified. It is. Although it is extremely easy to identify a leakage system when it is assumed that the reference phase can be specified, it is not practical. For this reason, it is indispensable to create an algorithm that does not depend on the reference phase in the leakage system identification. Therefore, an autocorrelation is one answer to the question of what is the basic parameter for the leakage system identification from the signal captured by the AD converter A / D and what is the parameter that is not swung to the reference phase. It is a series.

  When the input signal is f (t), the autocorrelation sequence A (τ) is A (τ) = Σ (f (t) × (f (t + τ)) *) (* indicates a complex conjugate operation). For example, the autocorrelation series of composite leakage (combination of resistive leakage and capacitive leakage), capacitive leakage, and resistive leakage (when resistive leakage and capacitive leakage occur) are shown in FIGS. As shown in B) and (C), autocorrelation sequences differ for each leakage condition, but the values are constant regardless of the extraction phase of the extracted signal f (t). Therefore, if system identification is performed using an autocorrelation sequence, system identification independent of the reference phase is possible.

  The problem is the integration period of the autocorrelation calculation of the waveform, but it is desirable that it is one cycle of the power supply frequency. In the waveform converted by the AD converter A / D, it is difficult to cut out one period from the sampling quantization. However, in synchronization with the input signal, the demodulating units DEM1 to demn demodulate the low-pass filter. If it is a baseband signal from which unnecessary bands are removed by LPF1 to LPFn, integration for one period is extremely easy and can be realized accurately.

  The levels of signals input to the AD converter A / D are various, and it is difficult to stably identify the system based on these levels. Therefore, as shown in FIG. 14 described above, normalization of the power of the leakage current is executed with the combined power. Further, as shown in FIG. 30 to be described later, the power may be obtained by synthesizing the powers of major harmonics that have a great influence on identifying the leakage system excluding the fundamental wave.

  In addition, the autocorrelation sequence calculation has an extremely large amount of calculation, and when such calculation processing is introduced, the amount of processing increases dramatically and the processing time becomes long. On the other hand, it is well known that the autocorrelation sequence represents the power spectrum of the input waveform. In fact, in the case of the power spectrum, it is not affected by the phase change. Actually, the autocorrelation sequence before the phase rotation is matched with the autocorrelation sequence after the phase rotation is performed even if arbitrary phase rotation is performed on the vector signals output from the low-pass filters LPF1 to LPFn. This is because the autocorrelation calculation itself calculates a complex conjugate vector, and the phase component disappears. Therefore, if the power of the output signals of the low-pass filters LPF1 to LPFn is viewed, it indicates that the autocorrelation calculation need not be performed.

  That is, the outputs of the low-pass filters LPF1 to LPFn are multiplied by each n-order harmonic vector signal by a complex conjugate vector having a radius of 1.0 to obtain a signal unrelated to the phase, and this is used as a reference signal for leakage system identification. . At the same time, by introducing a similar phase normalization circuit to the output of the leakage current simulation circuit EQL in the leakage identification unit 33 to obtain an output independent of the phase, this is used as the output (output parameter) of the leakage identification circuit. Then, the error extraction unit 4 obtains an error from the reference value (basic parameter), and the leakage system is identified so that this error is minimized. By executing such processing, the amount of calculation can be drastically reduced.

  FIG. 16 is an explanatory diagram of the dictionary reference unit 31. The error calculation unit 111, the power calculation unit 112, the power value integration unit 113, the selectors 114 and 120 (SEL), the minimum value determination unit 115, and the EQL. (Leakage current simulation circuit) It has a configuration including a coefficient unit 116, a leakage pattern storage unit 117, an initialization coefficient storage unit 118, and a control coefficient storage unit 119. In FIG. 9, error calculation, PWR The functions shown as value integration minimum value determination, initialization / control, SEL2, EQL coefficient, and various leakage pattern dictionaries are shown as specific configurations.

  For example, even in the case of a three-phase three-wire system, the leakage parameter includes a-phase leakage, b-phase leakage, c-phase leakage, capacitive leakage current, resistive leakage current, and harmonics as the first harmonic, secondary There are many types of harmonics, third harmonics, etc., and it is extremely difficult to perform system identification in a short time with such a large number of parameters. For this reason, various leakage patterns are stored in the leakage pattern storage unit 117 as a dictionary, and sequentially read leakage patterns are input to the error calculation unit 111 via the selector 114, and the LPF normalization from the reference value extraction unit 23 is input. Error, that is, an error from the reference value LPFN (n) is calculated, the power for the error is calculated by the power calculator 112, integrated by the power value integrator 113, and the error is minimized by the minimum value determiner 115. Select the earth leakage pattern. Although it is an error calculation part in this case, it is better to use only the nth-order harmonic effective for identifying the leakage system without using all the harmonics including the fundamental wave. However, when used, for the purpose of minimizing the dictionary pattern, it is preferable to calculate the error by normalizing the power of the harmonic group in the range to be used to 1.0 or the like. Alternatively, the minimum value may be obtained so that the combined power of the nth-order harmonics excluding the fundamental wave is 1.0.

  The minimum value determination unit 115 is for selecting a leakage pattern that minimizes an integration error from a plurality of dictionary reference results. The selector 114 (SEL) is a selection circuit for a plurality of fixed dictionary patterns and a leakage pattern (leakage pattern stored in a memory not shown) obtained in the past, and the coefficient from the EQL coefficient unit 116 and the leakage pattern The coefficient from the storage unit 117 is switched and input to the error calculation unit 111, and the selector 120 (SEL) switches the coefficient from the EQL coefficient unit 116 and the coefficient from the initialization coefficient control unit 118 to EQLRa, EQLCa, EQLRb, EQLCb, EQLRc, and EQLCc are input to leakage detection unit 33. Further, regarding the control force of the leakage current simulation circuit EQL, for example, in the three-phase three-wire system, there are a-phase resistive leakage path and capacitive leakage path, and further c-phase resistive leakage path and capacitive leakage path. For this reason, control power that matches each leakage pattern is provided to speed up convergence.

本発明では、基準電圧の振幅の大きさは１．０としてある。これは、一般的に波形が歪むと高調波を発生するが、この高調波は次数に応じてレベルが１／ｎに減少することが知られている（数学的にも証明可能）。一方、後述するＥＱＬ部では、容量性漏洩電流は、周波数に比例（次数に比例）して増大していくが、抵抗性漏洩電流は、次数に反比例して減少していく。これらを考慮し、基準電圧発生部３２では、半径１．０のベクトル電圧を基準電圧として出力する。一般に、ｎ次高調波は、時間軸波形がどの程度歪みを受けているかで決定され、最悪状態の歪みでは、基本波を基準として１／ｎに低下することになるが、最悪でない場合には、数１０ｄＢ減衰した形で順次低下する現象となる。この場合、高調波は１／ｎよりゆっくりした減衰カーブで、基本波を除いた形で現れる。従って、より正確な漏電検出をしようとした場合、基本波を除く高調波のパワーが、基本波のパワーと比較してどの程度減少しているかを算出し、この算出度合いに応じて、図１７の基準電圧発生部からのｎ次高調波の基準電圧ＶＧａ（１），ＶＧｂ（１），ＶＧｃ（１）〜ＶＧａ（ｎ），ＶＧｂ（ｎ），ＶＧｃ（ｎ）に、周波数特性上の補正を行うこともできる。但し、この差は微々たるものなので、通常は省略しても問題はない。 FIG. 17 is an explanatory diagram of the reference voltage generation unit 32 (see FIG. 9). The three-phase primary to n-order reference voltages VGa (1), VGb (1), VGc (1) to VGa (n), The case where it implement | achieves by ROM for generating VGb (n) and VGc (n) is shown. Assuming that the primary a-phase is the reference voltage phase 0 degree, the primary voltage phase is as shown in the equations (4) to (6), and the n-order voltage phase is as shown in the equations (7) to (9). Become.

In the present invention, the amplitude of the reference voltage is 1.0. This is generally known to generate a harmonic when the waveform is distorted, and the level of this harmonic is known to decrease to 1 / n depending on the order (which can also be mathematically proved). On the other hand, in the EQL section described later, the capacitive leakage current increases in proportion to the frequency (proportional to the order), but the resistive leakage current decreases in inverse proportion to the order. Considering these, the reference voltage generation unit 32 outputs a vector voltage having a radius of 1.0 as a reference voltage. In general, the nth-order harmonic is determined by how much the time-axis waveform is distorted. In the worst-case distortion, the nth-order harmonic is reduced to 1 / n with respect to the fundamental wave. , It becomes a phenomenon of decreasing gradually in a form attenuated by several tens dB. In this case, the harmonic appears as an attenuation curve slower than 1 / n, excluding the fundamental wave. Therefore, when more accurate leakage detection is attempted, the degree to which the power of the harmonics excluding the fundamental wave is reduced as compared with the power of the fundamental wave is calculated. N-order harmonic reference voltages VGa (1), VGb (1), VGc (1) to VGa (n), VGb (n), and VGc (n) from the reference voltage generator of FIG. Can also be done. However, since this difference is slight, there is usually no problem even if it is omitted.

  FIG. 18 is an explanatory diagram of an EQL (leakage current simulation circuit) unit of the leakage identification unit 33 (see FIG. 9), and the primary to n-th order reference voltages VGa (1), VGb from the reference voltage generation unit 32 described above. (1), VGc (1) to VGa (n), VGb (n), VGc (n) are input, and current parameters EQLRa (1), EQLRb (1), EQLRc (1) to EQLCa (n), EQLCb (N) and EQLCc (n) are subjected to complex multiplication and addition (Σ) to obtain final desired EQL (1) to EQL (n) outputs. However, here, when the order of the nth-order harmonic increases, the capacitive leakage increases in proportion to the frequency. On the other hand, with respect to resistive leakage, there is no change depending on the frequency and is constant. Further, since the harmonic signal is generally reduced to 1 / n according to the order, the coefficient is multiplied by 1 / n for EQLRa (n). And calculate the output. Nothing is done about the capacitive leakage current. By doing so, a more practical leakage system identification becomes possible.

Assuming that the voltage source of the nth harmonic is V, the impedance of the load existing between each phase and the ground is Z, and the current flowing through ZCT is I, I = V / Z. On the other hand, the load impedance Z is a parallel connection of a capacitive load existing between each phase and the ground and a resistive load. When the frequency is ω, the capacitance is C, and the resistance is r, the capacitive load The current Ic flowing through
Ic = V / (1 / jωC) = jωCV (10)
The current Ir flowing through the resistive load is
Ir = V / r (11)
It becomes. Therefore, the current I finally flowing through the ZCT is
I = V (1 / r + jωC) (12)
It becomes. Actually, since the voltage source V is an nth order harmonic reference voltage source and each phase (a / b / c phase) exists, all these combined vectors are obtained.

  In order to obtain the resistive leakage current that is the extraction target, it is necessary to measure the resistive leakage current component by separating the captured zero-phase current component into a resistive leakage current component and a capacitive leakage current component. . The capacitive leakage current is a phase advanced by 90 degrees of the resistive leakage current.

  The capacitive load and the resistive load that are actually constructed in the leakage system are defined not by impedance but by admittance from the viewpoint of ease of system identification. The zero-phase current input from the ZCT is estimated by multiplying and combining the admittance vector and the reference n-order harmonic reference voltage source.

  However, the zero-phase current estimated above is based solely on the nth-order harmonic reference voltage from the reference voltage generator 32. On the other hand, the reference voltage phase and amplitude of the leakage system actually connected to the ZCT are different from the above reference phase and amplitude, and it is necessary to correct this amplitude difference / phase difference. For this reason, a phase normalization circuit / amplitude correction circuit is provided in the leakage system to perform final system identification.

  FIG. 19 is an explanatory diagram of an EQL (leakage current simulation circuit) normalization unit 131 and an AMP (n) unit 132 of the leakage identification unit 33, wherein 133 to 134 are multipliers, 136 is a square circuit, 137 is a division circuit, and Square root circuit, 138 and 139 are adders, 140 is a multiplier, 141 is a delay circuit (T), 142 and 143 are limiters (LM), and 144 is a division circuit. Similar to the normalization of LPFn, the EQL normalization unit 131 obtains a signal normalized to a vector having a radius of 1.0 by EQLn complex conjugate calculation, and multiplies the EQLn output to output EQLN (n).

  The AMP (n) unit 132 performs correction so that the output of each EQLNn matches the LPFAn that is the reference value. This is output as a vector value that rotates with a radius of 1.0 from the reference voltage generation source, and the harmonic power attenuation is taken into account so that it decreases to 1 / n as the order n of the nth harmonic increases in the EQL section. However, in practice, this correction is performed when the power attenuation is smaller than the assumed power attenuation. Therefore, the correction circuit is the same as the normal AGC circuit. Note that a configuration in which the output A of the division circuit 144 is input to the multiplier 133 and the EQL (n) is multiplied by a plurality of stages may be provided to perform coarse adjustment and fine adjustment. Or, depending on the degree of distortion of the nth-order harmonic, the power ratio of the fundamental wave and the non-fundamental wave is greatly different, and the power values other than the fundamental wave are not particularly problematic even when collectively controlled. Therefore, AMP (1) can be made independent and AMP (2) to AMP (n) can be collectively controlled.

  FIG. 20 is an explanatory diagram of the error extraction unit 4 (see FIG. 9). LPF normalization output (n) (basic pattern) from the LPF normalization unit (see FIG. 14) of the reference value extraction unit 23, and AMP The difference from the output AMP (n) (output pattern) from the (n) unit 132 (see FIG. 19) is obtained and input to the EQL correction 1 (see FIG. 9) of the leakage identifying unit 33. System identification is performed so that the error extraction output (n) from the error extraction unit 4 is minimized. As a basic algorithm in this case, the least square method can be applied. In addition, (n) shows 1 to n of primary to n-order.

  FIG. 21 is an explanatory diagram of correction means shown as EQL correction 1, SEL3 / PN generation, and EQL correction 2 of the leakage detection unit 33 (see FIG. 9). An EQL (leakage current simulation circuit) correction unit 151 and a selection unit 152 are illustrated. And an EQL (leakage current simulation circuit) correction unit 153 are shown. In addition, it has the structure corresponding to three phases a phase, b phase, and c phase, shows only the configuration corresponding to the a phase, and outputs EQLRa and EQLCa from the EQL correction unit 153. This outputs EQLRb, EQLCb, EQLRc, and EQLCc corresponding to c phase. In addition, CNTRa and CNTCa are input for the a phase, but CNTRb, CNTCb, CNTRc and CNTCc are input for the b and c phases. Reference numeral 154 denotes a complex conjugate calculation unit, 155 and 156 denote multipliers, 157 denotes a selection unit, 158 denotes a selection pattern generation unit, 159 denotes an adder (Σ), 160 denotes a multiplier, 161 denotes an adder, and 162 denotes a limiter (LM). ) And 163 denote delay circuits (T).

  The LPFn output (nth-order harmonic signal) is a line spectrum and becomes a fixed pattern after extraction. In general, system identification is generally performed so that an error between a reference value and an estimated value is minimized, specifically, an error correlation is eliminated. However, when an LPFn signal (nth-order harmonic signal) is used, it becomes a fixed pattern, and therefore has a very high correlation with the reference signal itself. Therefore, when performing system identification, it is necessary to make these signals uncorrelated. Specifically, a selection circuit capable of randomly selecting an nth-order error among errors of each order is provided. Thereby, the spectrum of LPFn (nth harmonic signal) serving as a reference becomes random and uncorrelated. Therefore, stable system identification is possible. For this purpose, a selection circuit 157 and a selection pattern generation unit 158 are provided to randomly select an error corresponding to the nth harmonic of each phase.

  That is, the selection unit 157 selects and outputs an error corresponding to the nth harmonic in accordance with a random selection signal from the selection pattern generation unit 158, and inputs the error to the adder 159 of the EQL correction unit 153. Alternatively, all the switches can be selected in a fixed pattern and input to the adder 159 of the EQL correction unit 153 for addition. As a result, the signal can be made uncorrelated with the reference signal. Accordingly, the selection pattern generation unit 158 is configured to output a random pattern selection signal or to output a predetermined fixed pattern selection signal.

Further, the output signal of the reference voltage generating unit 32 is input to the above-described EQL unit, and the capacitive leakage current vector identified by the system identification and the resistive resistance leakage current vector are respectively multiplied, and the combined estimated nth-order harmonic signal EQL ( 1) to EQL (n) are output. As an equalization coefficient parameter of EQL, for each phase (PH = a, b, c),
_{EQL (PH) = EQL R (} PH) + jEQL C (PH) ··· (13)
It can be expressed as.

  The above-mentioned EQL correction means can apply an automatic equalization technique applied to a modem or the like. That is, the error signal of the reference value and the estimated value is calculated by the complex conjugate of the phase normalization signal, further multiplied by the inverse of the amplitude correction, and multiplied by a certain control force. The resistive component is multiplied by the complex conjugate of the error and the reference voltage (acts so as to eliminate the correlation between the reference voltage and the error), and the system identification is corrected for the resistive component using the least square method.

に示すものとなる。又抵抗性成分については、基準電圧発生部３２からの基準電圧の次数に応じて１／ｎに低下する為，ＥＱＬ部（図１８参照）に示すように、係数のＥＱＬＲａ，ＥＱＬＲｂ，ＥＱＬＲｃを、１／ｎの定数を乗算して対応する。容量性成分については、次数の増加に伴う振幅の低下に対して、インピーダンスの低下により相殺されるので、そのままとする。 For capacitive components,

It will be shown in Further, since the resistance component is reduced to 1 / n according to the order of the reference voltage from the reference voltage generation unit 32, the coefficients EQLRa, EQLRb, EQLRc are set as shown in the EQL unit (see FIG. 18). Corresponds by multiplying by a 1 / n constant. The capacitive component is left as it is because the decrease in the amplitude accompanying the increase in the order is offset by the decrease in the impedance.

The detailed EQL correction algorithm will be described based on the following equations (15) to (17).

  Since the EQL output is subjected to phase normalization and amplitude correction by the AMPn circuit, and finally output to the error extraction circuit, the reverse is performed in the correction circuit. That is, the error signal from the error extraction unit 4 is multiplied by AMPI (n) (inverse value of AMP (n)) by the multiplier 155 of the EQL correction unit 151, and then the CJE (n ) (* Of CJE *), complex conjugate reverse rotation, synthesis of error and multiplication of control force to perform tap coefficient correction. Further, CNTRa and CNTCa are control powers for correcting tap coefficients of various leakage patterns.

  When correcting the tap coefficient, for example, in the three-phase three-wire system, EQLRa, EQLCa, EQLRc, and EQLCc are final tap coefficients. For example, if the correction order is only the first order, third order, fifth order, and seventh order, it is conceivable to synthesize all these correlation errors to correct the tap coefficient, or to perform correction in individual fine units. This is a point of conflict between ensuring convergence accuracy and ensuring convergence time. For this reason, it is possible to improve the convergence speed and the accuracy by optimizing the error synthesis.

  The selection pattern generation unit 158 that performs selection control of the selection circuit 157 has a function of generating a PN pattern or a fixed pattern as described above. The generation of the PN pattern is, for example, pseudo-random with 15 bits as one cycle. It can be a random pattern by a signal. This random pattern is easy to generate in M series or the like. In this case, if 1 is a switch on and 0 is a switch off, 1111 (for example, in the order of primary, tertiary, fifth order, and seventh order) turns all the switches on, synthesizes errors, and performs tap updating. If it is 1000 (for example, in the order of primary, tertiary, fifth order, and seventh order), tap correction is performed using only the primary error path. In the earth leakage identification system, since the correlation between phases is high, the control power for error correction becomes a control power with a large correlation. The optimum values of these control forces can be obtained by simulation and stored in the dictionary (ROM) of the dictionary reference unit 31 described above.

  As described above, the leakage waveform output from the ZCT is an aggregate of single tone waveforms, and is a signal having extremely high correlation. When performing system identification of such a signal, a circuit for randomization is required somewhere, and it is also necessary for flattening the spectrum. In this way, by flattening the spectrum, the stability of system identification is increased and high-speed convergence is possible. Further, fixed pattern correction (for example, all n-th order error synthesis) may be performed so that average high-speed pull-in is possible at the first pull-in.

  FIG. 22 is an explanatory diagram of EQL coefficient convergence characteristics, where the vertical axis of (A) and (B) indicates the EQL coefficient, the horizontal axis indicates the number of corrections, and (A) is capacitive in a phase and c phase. When there is a leakage and there is a resistive leakage in the c-phase, (B) shows a case where there is a single-phase leakage in both resistance and capacitance. In any case, it can be seen that convergence is achieved with about six corrections. By using the leakage pattern held in the dictionary reference section 31, the EQL coefficient starts from about 1 instead of 0 in the Ra and Ca in FIG. In FIG. 22B, Rc indicates that the convergence can be speeded up starting from about 0.5.

  23 is an explanatory diagram of the Igr determination / timer monitoring unit 34 (see FIG. 9), in which 171 is an IGR calculation unit, 172 is an IGR determination unit, 173 is a timer monitoring unit, 174 is a setting unit, 181, 182, and 188. , 189 are multipliers, 184 are adders, 185 and 186 are square circuits, 187 are adders, 190 and 191 are adders, and 192 and 193 are continuous time monitoring units. The setting unit 174 indicates a part of the functions of the various setting / display control units 5 in FIG.

  When the detection signal I0DEC indicating the presence or absence of the leakage current from the I0 detection unit 25 is 0 indicating the absence, the output signal of the multiplier 189 is 0 and the determination in the IGR determination unit 172 is not performed. That is, as described above, malfunction can be prevented. If the detection signal I0DEC is 1 and indicates that there is a leakage current, the corrected coefficients EQLRa, EQLRb, and EQLRc from the EQL correction unit 153 of the leakage detection unit 33 are input to the IGR calculation unit 171 and their phases are different. Then, they are vector-synthesized and added by the adder 187 as power by the square circuits 185 and 186 to obtain an Igr component. Further, since the combined power is normalized in the LPFn normalization circuit, in order to restore this, the SYNLPFP from the LPF normalization unit (see FIG. 14) is multiplied in the multiplier 188, and the original power is restored. The level is returned to the IGR determination unit 172 via the multiplier 189.

  The IGR determination unit 172 inputs the caution threshold value from the setting unit 174 to the adder 190 and the warning threshold value to the adder 191, and determines the level of the output signal of the IGR calculation unit 171. When the threshold value is exceeded, the continuous time is monitored by the continuous time monitoring units 192 and 193 of the timer monitoring unit 173. The time limit set by the setting unit 174 is, for example, 10 seconds. When the continuous time exceeds the set time, a caution signal or a warning signal is transmitted. The set time limit in this case may be a longer time period according to the configuration of the power supply system.

  FIG. 34 is an explanatory diagram of the combined waveform of the nth-order harmonic component, and shows examples of leakage waveforms when the nth-order harmonic component is large and small. A three-phase, three-wire system, capacitive leakage current is generated in both a phase and c phase, and resistive leakage current is also generated in both a phase and c phase. Is extremely large and shows a case where the waveform is greatly distorted, and the other shows a case where the n-order harmonic component is relatively small and the waveform is not so distorted. When this one waveform is compared with the other waveform, the two waveforms are greatly different, but the resistive leakage current has the same value. FIG. 6 described above shows an actual measurement waveform in the actual field, and the level of the nth harmonic is a waveform that is reduced by about 18 dB from the fundamental wave.

  FIG. 32 is an explanatory diagram of amplitude distribution of the third and seventh harmonics when the fifth harmonic is normalized. For example, in the case where the level of the fifth harmonic is normalized to 1.0, The power values of the third and seventh orders are shown in a two-dimensional graph, and various types of assumed leakage patterns are plotted. This indicates that the third and seventh amplitude coordinate positions can be specified by the leakage patterns. ing. The same result was obtained in the simulation for one and the other waveform shown in FIG. Although it is the power of the n-th harmonic, even if the amount of leakage is the same, the combined power value is greatly different for each leakage system. However, the same deterioration pattern is observed when the power is normalized. Therefore, it can be understood that some normalization means is effective in order to perform stable leakage detection.

  FIG. 33 is an explanatory diagram of amplitude distribution obtained by normalizing the third-order, fifth-order, and seventh-order harmonic powers, normalized so that the combined power of n-order harmonics excluding the fundamental wave is 1.0, and the third-order, fifth-order harmonics are normalized. The next and seventh power patterns are shown, and the power values of various leakage patterns are shown at the same level as in FIG. Such a pattern can also be stored in the dictionary as a basic parameter.

  In FIG. 30, focusing on the fact that power values other than the fundamental wave and the fundamental wave fluctuate greatly, refer to the power value excluding the fundamental wave, and LPF (n ) Is normalized to obtain LPFN (n). Since the leakage identification unit requires the same processing as the LPF normalization unit, processing with the configuration shown in FIG. 31 is performed.

  In FIG. 30, LPF (1) is subjected to complex conjugate arithmetic processing to normalize the phase, and then output as LPFN (1) as it is. SYNLPFP representing the power value of (1) is output and input to the IGR detection unit 171 of the Igr determination / timer monitoring unit shown in FIG. In addition, LPF (n) other than the fundamental wave normalizes the phase in the same manner as LPF (1), but at the same time the power is set so that the combined value of the powers of n-th harmonics other than the fundamental wave is 1.0. Normalize. That is, a value a obtained by synthesizing the square values of the harmonics LPF (k) to LPF (n) excluding the fundamental wave LPL (1) in the synthesis PWR by the synthesis unit Σ is set to 1 / √a, and other than the fundamental wave Multiply the phase normalized output of LPF (n) to normalize the power and output as LPFN (n). The leakage identifying unit performs identification using this LPFN (n).

  FIG. 31 is an explanatory diagram of the EQL normalization unit and the AMPn unit. The same processing as in FIG. 30 is performed, and EQL (n) is after the phase is normalized so that it is independent of the phase. , The amplitude is corrected so that the combined power value excluding the fundamental wave is 1.0, and then output. At the same time, a correction signal AMPI (n) (for the purpose of making the loop gain constant) for amplitude reverse correction is output. EQL (1) is phase-normalized and outputs its real part Re as EQLN (1). At the same time, the power value is obtained and the reciprocal operation is performed to obtain an output signal of SYNQLP. The EQL output is input to the error extraction unit 4 with a combined power value of 1.0, and the coefficient is corrected so that the error becomes zero, and the leakage system is identified. Based on the identified result, the output of the fundamental wave (first harmonic) can be calculated, and a desired resistive leakage current Igr can be obtained. Specifically, it is obtained by the configuration shown in FIG.

  Also, as equivalent to ZCT, single-phase two-wire or three-phase three-wire are bundled in the same direction to detect common mode current, and the detected common mode current is detected by the leakage current detecting means described above. The resistive leakage current can be detected by the same processing as in the example.

It is a principle explanatory view of the present invention. It is explanatory drawing of an alarm system. It is connection explanatory drawing of ZCT. It is explanatory drawing of a leak detection system. It is a simulation waveform explanatory drawing of the leakage current of a single phase. It is explanatory drawing of a leakage current spectrum. It is a pattern explanatory drawing of a spectrum. It is explanatory drawing of the Example of this invention. It is a detailed block diagram of the Example of this invention. It is explanatory drawing of a carrier generation part and a demodulation part. It is explanatory drawing of a low-pass filter. It is explanatory drawing of an I0 detection circuit. It is explanatory drawing of a complex conjugate circuit and a frequency conversion part. It is explanatory drawing of a LPF normalization part. It is explanatory drawing of an autocorrelation calculation result. It is explanatory drawing of a dictionary reference part. It is explanatory drawing of a reference voltage generation part. It is explanatory drawing of an EQL part. It is explanatory drawing of an EQL normalization part and an AMP (n) part. It is explanatory drawing of an error extraction part. It is explanatory drawing of a correction | amendment means. It is EQL coefficient convergence characteristic explanatory drawing. It is explanatory drawing of an Igr determination / timer monitoring part. It is explanatory drawing of (DELTA) (DELTA) connection. It is explanatory drawing of a YY connection. It is explanatory drawing of YY connection (neutral point extraction). It is explanatory drawing of a Ydelta connection. It is explanatory drawing of (DELTA) Y connection. It is explanatory drawing of VV connection. It is explanatory drawing of a LPF normalization part. It is explanatory drawing of an EQL normalization part and an AMPn part. It is amplitude explanatory drawing of the 3rd and 7th harmonic when the 5th harmonic is normalized. It is amplitude distribution explanatory drawing which normalized the 3rd, 5th, 7th harmonic power. It is a synthetic waveform explanatory view of the nth harmonic component.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ZCT part 2 Basic parameter extraction part 3 System identification circuit part 4 Error extraction part 5 Various setting / display control part 21 Filter part 22 Data acquisition part 23 Reference value extraction part 24 Frequency information extraction part 25 I0 detection part 31 Dictionary reference part 32 Reference voltage generator 33 Leakage identification unit 34 Igr judgment / timer monitoring unit

Claims (14)

In a leakage current detection device that detects a resistive leakage current by processing a signal from ZCT,
A basic parameter extracting means for inputting a signal from the ZCT for detecting a current flowing in the ground line, extracting frequency information of the signal, and extracting a basic parameter necessary for identification of the leakage system;
An error extracting means for extracting an error between the basic parameter and the output parameter;
System identification circuit means for correcting the output parameter so that the error extracted by the error extraction means approaches zero, identifying a leakage system including a leakage path, and estimating and detecting a resistive leakage current. Leakage current detector characterized by.   The basic parameter extracting means extracts frequency information of the signal from the ZCT, demodulates the signal from the ZCT according to the frequency information, generates a baseband nth-order harmonic vector signal, and removes the fundamental wave Means for normalizing the second harmonic vector signal, and extracting a basic pattern including the power of the nth harmonic vector signal phase-normalized by the means and inputting the basic pattern to the system identification means; The leakage current detection device according to claim 1.   The means for normalizing the phase of the basic parameter extracting means includes: phase normalization of the nth-order harmonic vector signal; and control means for controlling the combined power of the n-order harmonic vector signal to be a predetermined value. The leakage current detection device according to claim 1, further comprising:   The system identification unit inputs a reference voltage generation unit that outputs an nth-order harmonic signal, and a reference voltage from the reference voltage generation unit, and inputs an output parameter based on identification of the leakage system to the error extraction unit. The leakage current detection apparatus according to claim 1, further comprising: a leakage identification unit that performs a correction process on the output parameter so that an error from the basic parameter is minimized.   The system identification unit minimizes an error between a dictionary reference unit holding a parameter indicating a plurality of leakage patterns, a parameter indicating the leakage pattern from the dictionary reference unit, and a basic parameter from the basic parameter extraction unit. The leakage current detection device according to claim 1, wherein a parameter indicating the leakage pattern is configured to be set as an initial parameter of the output parameter from the leakage identification unit.   6. The leakage current detection apparatus according to claim 5, wherein the dictionary reference unit stores a parameter that is identified and converged by the leakage identification unit together with a parameter indicating the leakage pattern stored in advance.   The leakage identification unit of the system identification unit includes a correction unit that corrects the error by the error extraction unit in correspondence with the n-th harmonic, the error corrected by the correction unit, and the reference voltage generation unit. A leakage simulation circuit for inputting the nth-order harmonic signal and a leakage simulation circuit normalization unit for inputting an output pattern obtained by normalizing the signal from the leakage simulation circuit to the error extraction unit. The leakage current detection apparatus according to claim 1 or 4, characterized in that:   The leakage identification unit of the system identification unit includes a configuration in which a resistive leakage coefficient is multiplied by a coefficient that decreases in accordance with the order of the nth harmonic signal, and a capacitive leakage coefficient is multiplied by a constant 1. Item 5. The leakage current detection device according to Item 1 or 4.   The leakage detection unit of the system identification unit includes a correction unit that selects an n-th harmonic component corresponding to an error by the error extraction unit randomly or in a fixed pattern to minimize the correlation. The leakage current detection device according to claim 1 or 4. In a leakage current detection method for detecting a resistive leakage current by processing a signal from ZCT,
Input a signal from the ZCT that detects the current flowing in the ground line, extract frequency information of the signal, extract basic parameters necessary for identifying the leakage system based on the frequency information,
Correct the output parameter so that the error between the basic parameter and the output parameter is minimized, identify the leakage system including the leakage path, and detect the resistive leakage current by the convergence of the leakage system identification. A leakage current detection method characterized by comprising a process of:   The signal from the ZCT is demodulated with a reference frequency and converted into a vector signal by complex conjugate calculation. The vector signal is converted to obtain frequency information of the signal from the ZCT, and the signal from the ZCT is demodulated according to the frequency information. The leakage current detection method according to claim 10, further comprising: obtaining an n-order harmonic signal component and normalizing the n-order harmonic signal component as the basic parameter.   An error is extracted between the output parameter obtained by multiplying the nth-order harmonic signal from the reference voltage generator by a coefficient and subjected to correction processing and the basic parameter, and the output is set so that the error is minimized. The leakage current detection method according to claim 10, further comprising: performing the parameter correction process to identify the leakage system.   A step of identifying the leakage system as an initial value of a coefficient by which a parameter indicating a plurality of leakage patterns from the dictionary reference unit is multiplied by the n-th harmonic signal from the reference voltage generation unit. The leakage current detection method according to claim 10 or 12.   11. A process of performing correction processing for minimizing correlation by randomly selecting and combining n-order harmonic components corresponding to an error between the basic parameter and the output parameter. 14. The leakage current detection method according to any one of items 13.

Images