JP2011047839A - Vector measuring device, vector measuring system, and vector measuring method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for measuring highly accurately at high speed, a frequency and a vector of an AC signal. <P>SOLUTION: Measurement data which are AC signals are inputted, and a fundamental wave component of discrete Fourier transform is operated by performing weighted addition with a complex constant determined by being repeated with a period which is a natural number. In the operation of the fundamental wave component, the operation is performed a plurality of times by changing the period of the discrete Fourier transform, and the frequency is operated from the ratio of each operation result of the discrete Fourier transform having each different period, and a phase detection result, by using a plurality of operation results of the discrete Fourier transform. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a vector measurement apparatus, a vector measurement system, and a vector measurement method, and in particular, a vector (amplitude, frequency, phase) of a variable frequency AC signal can be obtained at high speed, stability, and steady error regardless of industrial field or consumer field. The present invention relates to a vector measurement apparatus, a vector measurement system, and a vector measurement method that can be measured close to zero.

  Demand for AC signal vector (amplitude, frequency, phase) measuring devices is diverse. Typical applications include control signals for AC motors and generators, power system protection relays in the industrial field, power conversion devices for photovoltaic power generation devices, and image and audio noise canceling devices in the consumer field.

  In general, the fundamental component of a discrete Fourier transform (DFT) for an input signal can be obtained by non-recursive operation. Since the number of product-sum operations tends to be enormous when non-recursive operations are performed, for example, a method for reducing the number of product-sum operations by recursively computing the fundamental component of the discrete Fourier transform (DFT) as a vector measurement method for AC signals Has been published by Phadke et al. (Non-patent Document 1). Further improved, the technique described in Patent Document 1 is known. It is applied to AC signal measurement, such as a 400MW variable speed pumped storage power generation system using a secondary excitation system, and is useful for stable operation continuation when the AC system is abnormal.

  In order to reduce the number of multiply-accumulate operations, the coefficients are shifted simultaneously with the input data, and a recursive DFT is found. The calculation result is a vector viewed from the coordinates rotated counterclockwise at the fundamental frequency, and the input signal When the frequency is the same as the fundamental frequency, the calculation result is constant.

  Moreover, as a method for measuring the frequency of the AC signal, a method using the leakage effect of DFT has been published by Girgis et al. (Patent Document 2).

JP-A-01-283015 A. Girgis et al., “Frequency Measuring and Monitoring Apparatus, Methods and Systems”, US Pat. No. 4,319,329

A. G. Phadke et al., “A New Measurement Technique for Tracking Voltage Phasors, Local System Frequency, and Rate of Change of Frequency”, IEEE PAS-102, May 1983.

  Here, the computing ability of the microcomputer has been dramatically increased as compared with the time of publication of Non-Patent Document 1 and the like. Non-Patent Document 1 examines the power system frequency signal in 12 divisions per cycle. Today, more than 200 divisions can be easily realized with a general-purpose microcomputer. Assuming that the number of divisions of the reference frequency signal is N, the product-sum operation is reduced from N times of non-recursive operation to 2 times by recursive operation. Therefore, the effect of the recursive operation method of Non-Patent Document 1 increases as N increases.

  However, this method has a problem that if the measurement signal deviates by Δf from the reference frequency, not only steady amplitude and phase errors are generated due to the principle of DFT, but also a pulsation error of (2 · Δf) component occurs. For this reason, there is a drawback that it cannot be used practically unless two or more signals having the same frequency and different phases are input, and cannot be used for the most basic one-input signal.

  In addition, when the input is a three-phase AC system signal, the DFT positive phase vector calculation method has the advantage that it is difficult to be affected by noise and harmonics, but has a large error with respect to frequency change.

  On the other hand, the frequency measurement method according to Patent Document 2 does not specifically show a method for calculating the amplitude and phase of a vector from a frequency shift Δf. In addition, there are disadvantages that a special device is required because the number of product-sum operations is large and sampling needs to be started from the sign change point of the AC signal.

  An object of the present invention is to provide a vector measurement apparatus, a vector measurement system, and a vector measurement method that combine noise resistance / stability and response speed in view of the above-described problems of the prior art.

In order to achieve the above object, in the present invention, a data table for storing past N _w times of measurement data at regular intervals, and the N _w pieces of data are input in a fixed order and repeated with a period N _b which is a natural number. A first DFT computing means for outputting the result X _N obtained by weighted addition with complex constants as a fundamental wave component of discrete Fourier transform, and an output X _N of the first DFT computing means is outputted as a vector measurement result of measurement data. In this case, M _b is a natural number different from N _b, and the result X _M obtained by inputting the data for M _w times from the data table in a fixed order and weighting and adding with a complex constant repeated in the period M _b is used for the discrete Fourier transform. a second DFT calculation means for outputting a fundamental wave component, and the deflection angle calculating means for outputting a deflection angle theta _n of X _n, the real part or the imaginary part of the ratio of X _n and _{X M (X M / X n} ) Select And comparison means is provided, and configured to calculate output frequency f from the output of the comparison means and the deflection angle theta _n that.

In addition, a data table for storing past N _w times of measurement data at regular intervals, and a result X _N obtained by inputting the N _w data in a fixed order and weighting and adding with a complex constant repeated in a cycle N _b that is a natural number. The first DFT computing means that outputs as the fundamental wave component of the discrete Fourier transform and the output X _N of the first DFT computing means as the vector measurement result of the measurement data, the frequency f of the measurement data is detected A frequency detecting means for outputting, a declination calculating means for outputting the declination angle θ _n of X _N , and a correction calculating means for outputting a correction coefficient H from the declination angle θ _n and the frequency f. the result of multiplying the output X _N was configured to output as a vector measurement.

In addition, a data table for storing past N _w times of measurement data at regular intervals, a complex constant table that is repeated with a period N _b that is a natural number, and two or one complex constant are selectively output from the complex constant table. Constant selection means, weighted subtraction means for weighting and subtracting data to be removed from the data table and data added in place of the removed data with a complex constant from the constant selection means, and the output of the weighting subtraction means itself A recursive operation means for correcting the output of the constant, and a multiplying means for converting the complex constant from the constant selection means into a conjugate number and multiplying the output of the recursive operation means, and the output of the multiplication means is the basis of the discrete Fourier transform It was configured to output as a wave component.

In addition, a weighting of N _w data using a data table that stores measurement data at regular intervals and a complex constant that is repeated with a period N _b that is a natural number by inputting at least a part of N _w data in the data table. A first DFT computing means for computing X _N which is a fundamental wave component of a discrete Fourier transform corresponding to addition, and M _b is a natural number different from N _b, and at least a part of M _w pieces of the data table Second DFT computing means for computing X _M , which is a fundamental component of a discrete Fourier transform corresponding to weighted addition of M _w data, using complex constants repeated with a period M _b with data as input; a polarization angle calculating means for calculating a deflection angle theta _n of _n, provided selection means for selecting and outputting the real part or the imaginary part of the ratio of X _n and X _M, from the output of the polarization angle theta _n selection means To calculate the frequency f Configured.

In addition, a weighting of N _w data using a data table that stores measurement data at regular intervals and a complex constant that is repeated with a period N _b that is a natural number by inputting at least a part of N _w data in the data table. First DFT calculation means for calculating X _N which is a fundamental wave component of discrete Fourier transform corresponding to addition, frequency detection means for detecting frequency f of measurement data, and a deviation angle θ _n of X _N are output. A declination calculator and a correction calculation means for calculating the correction value H from the declination θ _n and the frequency f are provided, and the result of correcting the output X _N with this correction value H is output as a vector measurement value. did.

  In addition, a first DFT calculation means for obtaining a first calculation value by performing a discrete Fourier transform calculation using complex constants repeated at a first cycle from measurement data at a fixed interval, and a first cycle from the measurement data Second DFT operation means for obtaining a second operation value by performing a discrete Fourier transform operation using complex constants repeated at different second periods, and a division result of the first operation value and the second operation value Based on the above, the first calculation value is corrected, or the measurement data frequency is obtained.

Further, according to the present invention, if the same input data is calculated by two non-recursive DFTs having different periods N _b and M _b and the periods N _b and M _b and the data lengths N _w and M _w are appropriately selected, We paid attention to the fact that the real part or imaginary part of the ratio of the non-recursive DFT becomes a monotonic function of frequency using the phase of the detection vector obtained from the non-recursive DFT as a parameter. The frequency is detected using this characteristic.

Further, it has been noted that the calculation result of the recursive DFT can be corrected using the phase obtained from the detection frequency and the non-recursive DFT. This correction is focused on that the relationship between the DFT cycle N _b and the number of data N _w is not limited to N _b = N _w . As a result, vector measurement is performed without steady-state error.

  According to the present invention, it is possible to provide a vector measuring device having noise resistance / stability and response speed. More specifically, by maintaining the advantages of DFT operations that are not easily affected by noise and harmonics, and by correcting non-recursive DFTs from recursive DFT operations that require a small amount of calculation, the amplitude and phase are corrected. The frequency, amplitude, and phase of a variable frequency signal can be measured even with one input signal, and the trade-off between noise characteristics / stability and response speed can be adjusted.

The block diagram of a vector measuring device. The block diagram of a data table. The block diagram of a DFT arithmetic unit. The block diagram of a DFT arithmetic unit. The block diagram of a DFT arithmetic unit. The block diagram of a DFT arithmetic unit. The block diagram of an output correction circuit. Polynomial product-sum calculator for output correction. Polynomial product-sum calculator for frequency calculation. The block diagram of a vector measuring device. The block diagram of a data table. The block diagram of a vector measuring device. The block diagram of a frequency discriminator. The block diagram of a vector measuring device. The block diagram of a data table. The block diagram of a vector measuring device. The block diagram of a vector measuring device. The block diagram of the application example of a vector measuring device. The block diagram of a vector measuring device. The figure which shows the locus | trajectory of ratio of 2 types of DFT calculation results. The figure which shows the range from which the ratio of 2 types of DFT calculation results becomes monotonous increase / phenomenon. The figure which shows the range from which the ratio of 2 types of DFT calculation results becomes monotonous increase / phenomenon. The figure which shows the range from which the ratio of 2 types of DFT calculation results becomes monotonous increase / phenomenon. The figure which shows the range from which the ratio of 2 types of DFT calculation results becomes monotonous increase / phenomenon. The figure which shows the relationship between the ratio of two types of DFT calculation results, a detection phase, and a frequency. The figure which shows the relationship between the ratio of two types of DFT calculation results, a detection phase, and a frequency. The figure which shows the relationship between the ratio of two types of DFT calculation results, a detection phase, and a frequency. The figure which shows the relationship of the ratio of two types of DFT calculation results, a detection phase, and frequency. The figure which shows the relationship between the ratio of two types of DFT calculation results, a detection phase, and a frequency. The figure which shows the relationship of the ratio of two types of DFT calculation results, a detection phase, and frequency. The figure which shows the weighting function of 2 types of DFT calculation results. The figure which shows the interpolation method of vector measurement. The figure which shows the example of a response of a vector measurement method. The figure which shows the example of a response of a vector measurement method. The figure which shows the absolute value of the correction coefficient of vector measurement. The figure which shows the absolute value of the correction coefficient of vector measurement. The figure which shows the example of a response of a vector measurement method. The figure which shows the example of a response of a vector measurement method. The figure which shows the example of a response of a vector measurement method. The figure which shows the example which applied the vector measurement method to internal induced voltage measurement.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. First, the basic concept of the present invention will be described in comparison with a reference example, and then a specific embodiment will be described.

In general, the fundamental wave component of the discrete Fourier transform (DFT) with respect to the input signal (X ₀ , X ₁ ,..., X _n-1 ) can be expressed by equation (1).

  If the non-recursive operation is performed as in equation (1), N product-sum operations are performed. Here, in order to clarify the end of the input signal and the end of the coefficient in the equation (1) and further indicate the current calculation result, it is re-represented in the equation (2).

Here, when the next latest input data x _N is sampled, the input signal is shifted to (X ₁ , X ₂ ,..., X _n ), and the new calculation result is given by equation (3).

  The calculation of equation (4) also requires N product-sum operations. In order to reduce the number of product-sum operations, the coefficients are shifted simultaneously with the input data, and in the case of a DFT that is recursively calculated, the calculation is performed using equation (4).

  From the difference between Equation (4) and Equation (2), Equation (5) is obtained.

  When formula (5) is arranged, formula (6) is obtained. However, the coefficient indicates that it holds for an arbitrary r.

  In the above equation (6), the calculation result is a vector viewed from coordinates rotating counterclockwise at the fundamental frequency. Therefore, when the frequency of the input signal is the same as the fundamental frequency, the calculation result is constant.

  This method has a problem that if the measurement signal deviates from the reference frequency by Δf, not only steady amplitude and phase errors are generated due to the principle of DFT, but also a pulsation error of (2 · Δf) component occurs. For this reason, there is a drawback that it cannot be used practically unless two or more signals having the same frequency and different phases are input, and cannot be used for the most basic one-input signal.

  In addition, when the input is a three-phase AC system signal, the DFT positive phase vector calculation method has the advantage that it is difficult to be affected by noise and harmonics, but has a large error with respect to frequency change.

  Here, it is used that the absolute value | Δf | of the deviation from the DFT reference frequency is substantially proportional to η in (7).

  However, a method for calculating the amplitude and phase of the vector from the frequency shift Δf is not specifically shown. In addition, there are drawbacks in that a special device is required because the number of product-sum operations is large and sampling must be started from the sign change point of the AC signal.

In the embodiment of the present invention, the period of the first non-recursive DFT is N _b , the number of input data is N _w, and it is not necessarily limited to N _b = N _w . In general, when N _w is relatively smaller than N _b , the response is excellent, and when N _w is relatively large, the stability is excellent.

Further, in the embodiment of the present invention, the second non-recursive DFT is calculated using the same input data and used for frequency calculation. The division number M _b is set to a value different from N _b, and the input data number M _w is set to N _w or less. The last input data of the second non-recursive DFT is the latest sample data in common with the first non-recursive DFT. Each calculation result can be expressed by (8) and (9).

  In the vector measurement device, the relational expressions (10) and (11) between the non-recursive DFT and the recursive DFT are used for the operations of (8) and (9), and the non-recursive is calculated after calculating the recursive DFT with few product-sum operations Convert to type DFT.

Here, the division numbers N _b and M _b and the data input numbers N _w and M _w of the DFT operation are selected so as to satisfy the relationship of (12) or (13).

  Here, the ratio of the DFT calculation result of Expression (9) and the DFT calculation result of Expression (8) is calculated by (14). Hereinafter, the frequency of the input signal is a single frequency in the range of (15), and the phase is arbitrarily in the range of (16).

(15) of f _b a DFT fundamental frequency of the periodic N _b, depends on the sampling period T and the period N _b. In principle, the fundamental wave component of the DFT calculation has a gain of 1 and an error of 0 at the frequency f = f _b .

An example of the locus of (14) is shown in FIG. In this example, (N _w : M _w : N _b : M _b = 3: 2: 3: 2) is selected. In the figure, the radius line drawn from the center of the circle constituting the locus is a vector when the phase of the input signal is θ = 0. In general, when the DFT calculation satisfies the conditions of (12) and (13), the locus of (14) moves with the frequency f, the center C moves and the radius r changes, and the detected phase θ _n detected in (10) is 2 It rotates counterclockwise on the circumference at double angle. What should be noted in FIG. 20 is that it is not the phase θ of the original input signal but the phase θ _n of the first DFT calculation result.

From this, the parameters (N _w , M _w , N _b , M _b ) are selected so that the movement of the center C of the circle is relatively large with respect to the radius r that varies with frequency, and the real part of the center coordinates and When the larger moving direction of the imaginary part is selected as D, D can be monotonously increased or monotonously decreased with respect to the frequency change.

Further, when the parameters (N _w , M _w , N _b , M _b ) satisfy the equation (12), the range shown in FIGS. 21 and 22 is selected, and when the equation (13) is satisfied, the parameters shown in FIGS. It was found that the selection of the real part and the imaginary part can be fixed when the selected range is selected, and there is no need to switch according to the detection phase.

In the frequency variation range (15), the real part D _r of D monotonously increases or monotonically decreases in the ranges of FIGS. 21 and 23, and the imaginary part D _i monotonously increases or monotonously decreases in the ranges of FIGS. all right.

21, FIG. 22, FIG. 23, and FIG. 24 show the results when _Nb is 200 or more, it was found that Nb is almost unchanged when _Nb is 50 or more, and the same tendency is observed when Nb is 6 or more.

  Among the ranges including FIGS. 21 and 22, it has been found that (17) and (18) are practically excellent, such as saving memory of complex coefficients for DFT calculation. (17) In both cases (18), the output D is fixed to the imaginary part.

In the case of (17), the high speed property is excellent. This is because the response converges in 1 / n cycle of the fundamental frequency f _b because N _w = N _b / n. On the other hand, since the DC component gain is not 0, it is necessary to ensure resistance to drift noise by another means.

In the case of (18), when N _w = N _b , the response convergence requires one cycle of the fundamental frequency f _b , but the response stability until the convergence is excellent. Moreover, since the DC component gain is 0, it has excellent resistance to DC drift noise.

  In addition, it was found that the real part and the imaginary part can also be fixed in (19) and (20). In (19), the imaginary part is fixedly selected, and in (20), the real part is fixedly selected.

In the case of (20), the resistance to harmonics is excellent. This is because the DFT fundamental frequency of the second DFT computation result is n times the fundamental frequency f _b of the first DFT computation result.

Figure 25 shows the variation of the imaginary part D _i when the change in the imaginary part D _i when selecting the n = 2, Figure 26 is obtained by selecting the n = 2 in (18) (17).

In both figures, the horizontal axis represents the detection phase θ _n of the first DFT calculation. Thus, when the detection phase θ _n is used as a parameter, D _r or D _i is monotonically increasing or decreasing monotonously with the frequency f, so that the frequency f can be uniquely determined.

Furthermore, when the parameter _{(N w, M w, N} b, M b) satisfies (12), the orientation of the radius vector rotates with theta _n is determined only by theta _n, it was found that that does not depend on the frequency f . By utilizing this characteristic, the effect of expanding the selection range of the parameters (N _w , M _w , N _b , M _b ) and the frequency detection accuracy can be increased.

For example, a parameter when choosing the _{(N w: M w: N} b: M b = 3: 3:: 6 4), FIG. 27 is a real part D _r, Figure 28 shows the imaginary part D _i. The sign of the imaginary part D _i is inverted by θ _n = −22.5 ± 21 [degrees], and the sign of the real part D _r is shifted by 45 degrees and the sign is inverted by θ _n = 22.5 ± 21 [degrees]. Therefore, for example, if the sign of sin (2 · θ _n ) is positive, D _r is selected, and if it is 0 or less, D _i is selected and output, the frequency f can be detected at an arbitrary θ _n .

As another example, when the parameter is selected as (N _w : M _w : N _b : M _b = 1: 1: 4: 1), according to this parameter selection, in a quarter cycle of the basic period f _b Since it converges, vector measurement can be performed at high speed. In particular, it is suitable for measurement in a low frequency region of a device operating at a variable frequency.

In this case, FIG. 29 FIG. 30 shows the imaginary part D _i the real part D _r. The real part _Dr is monotonically decreasing and the imaginary part _Di is monotonically increasing. However, since the radius with respect to the movement of the center in the locus of D is relatively large, the frequency detection accuracy tends to decrease due to harmonic noise or the like. A method of weighting in accordance with the detected phase theta _n frequency f _i that detects the frequency f _r and the imaginary part D _i for detecting the real part D _r As a countermeasure. In this case, the real part D _r has the highest detection accuracy when θ _n = 22.5 [degrees], and the imaginary part D _i is shifted 45 degrees and has the highest detection accuracy at θ _n = 67.5 [degrees]. . By utilizing this, the weighting function ε is defined as shown in FIG. 31 and the frequency detection value f is set to (21), so that the detection accuracy and stability can be improved.

When the frequency is realized by the measuring apparatus by the above method, the relationship between the detection phase θ _n , the frequency f, the real part D _r , and the imaginary part D _i shown in FIGS. A method of calculating the frequency by calculation is conceivable. In this case, using the fact that the phase fluctuation fluctuates by (2 · θ _n ), the table data is stored in the range of 0 ≦ θ _n ≦ π, and if θ _n <0, 2 · π is added to θ _n And enter it into the table.

However, by adopting a calculation method that makes use of the characteristics of D _r or D _i as described below, it is possible to obtain a result with stable calculation accuracy and calculation amount.

First, the relationship between the center coordinates of the circle shown in FIG. 20 and the frequency f can be approximated by polynomial expression (22) within the range (15). Where a _n is a constant complex number.

Next, the relationship between the radius vector of the circle and the frequency f can be approximated by polynomial expression (23) within the range of (15). Here, b _n is a complex constant.

  The relationship with D is (24).

As described above, when parameters are selected so that (12) holds, the radius vector is in a constant direction regardless of the frequency f. Thereby, (23) becomes (25), and the constant memory can be reduced. Here, k _n and δ are real constants.

  When (25) is divided into a real part and an imaginary part, n-order equations (26) and (27) of f hold.

  Of the equation (26) for the real part and the equation (27) for the imaginary part, the monotonic increase or monotonic decrease is selected. Let u be the subscript of the selected person. In this case, the equation has only one solution in the range of (15).

  Because of these characteristics, Newton's method is suitable for finding f by solving equation (26) or (27). When the Newton method is used as a recursive arithmetic expression for real-time arithmetic, (28) is obtained.

  However, in (28), the derivative for f can be expressed by (29) when the real part is selected and (30) when the imaginary part is selected.

  In the normal Newton method, the calculation is repeated until the absolute value of the difference between the predicted value and the corrected value becomes a predetermined value. In the case of the present embodiment, it was found that the repeated calculation of (28) is unnecessary. Since the amount of computation is constant, it has excellent real-time performance and can realize highly accurate computation.

  Next, a method for correcting the vector calculation result (10) from the frequency f detected in (28) or the like will be described.

  In this embodiment, a non-recursive DFT that fixes the weighted complex coefficient of the product-sum operation is examined, and further, the error is a correction in which the denominator is inverted, the true value is used as the numerator, and the detection result is used as the denominator We considered approximating the coefficients.

  From the above, when the relationship between the original signal and the vector is (31), the correction coefficient is (32).

Correction coefficient (32) was found to be approximated by (28) of the frequency detection value f (10) phase detection value theta _n polynomial (33) of. Moreover, it has been found that the relationship between the data length N _w and the frequency N _b is an approximate expression even when N _w = N _b . Here, c _n and d _n are complex constants.

By the procedure shown above, even if the first signal input, even if the frequency is varied from the reference frequency f _b, vector measurement apparatus for detecting a frequency and amplitude, the phase without steady-state error can be achieved.

  Further, by utilizing the fact that the frequency can be detected simultaneously with the amplitude and phase of the vector, it is possible to output a vector measurement result by interpolating between DFT calculation cycles. As shown in FIG. 32, the simplest interpolation method is a method in which only the phase is interpolated while the amplitude and frequency are constant during the interpolation period.

For example, when interpolating between time t _n and t _{n + 1} with Δt in FIG. 32, the vector measurement value is interpolated with (34).

  On the other hand, when the vector measuring apparatus of the present embodiment is compared with the conventional filter, the following two points are different in the transient response to the input step change.

First, the transient response period is limited to a period of up to N _w pieces of data are updated after a step change, then is a point returning to a steady output. Second, the transient response waveform differs greatly depending on the initial phase even if the amplitude and frequency before and after the step are the same.

  Therefore, when evaluating the transient response of the vector measuring apparatus of the present embodiment, it is necessary to scan the initial phase as a parameter from 0 to 2π and confirm the transient response for all initial phases.

In the following (35), (36), and (37), an example is shown in which the response is evaluated when the input signal represented by the expression (31) changes to the next step at t = 0. The initial phase θ ₀ is scanned from 0 to 2π.

FIG. 33 shows the case where N _b , M _b , N _w , and M _w are selected by (17) and n = 2, and FIG. 34 shows the case where n is selected by (18) and n = 2. In each figure, the average value (ave), maximum value (max), and minimum value (min) indicate the average value, maximum value, and minimum value of the response at each time when the initial phase θ is scanned.

When FIG. 33 and FIG. 34 are compared, the former has N _w / N _b = 0.5, so the convergence time is short and the response is excellent, and the latter has N _w / N _b = 1.0, and the number of data used for the calculation. Since there is more than the former, it turns out that it is excellent in stability.

Although the input signal has been assumed to have a single frequency as described above, it often includes harmonic components and DC drift that cannot be ignored in practice. In the case of FIG. 33, the odd harmonic gain of the fundamental frequency f _b is zero. In the case of FIG. 34, the gain is 0 for both odd and even harmonics. This is an advantage of a calculation method using DFT.

However, in the case of the present embodiment, the original DFT calculation result X _N is also amplified together with the harmonic components of the input signal in (32). Therefore, the larger the absolute value of the correction coefficient H is larger deviation of the input frequency and the fundamental frequency f _b is the harmonic resistance is weakened. In particular, when the harmonic frequency moves together with the fundamental frequency of the input signal, there is a high risk that harmonic tolerance becomes a problem.

FIG. 35 shows the absolute value of the correction coefficient H corresponding to FIG. 33, and FIG. 36 shows the absolute value of the correction coefficient H corresponding to FIG. Although the absolute value of the correction coefficient H changes with the phase θ _n , only the maximum value and the minimum value are shown because error evaluation is intended. When both are close to direct current, the correction coefficient increases rapidly. In the case of FIG. 35, it becomes large rapidly even at 2 · f _b . From these facts, it can be seen from these figures that the frequency measurement range and the harmonic resistance are in a trade-off relationship.

As a method of achieving both the frequency measurement range and the harmonic tolerance, there is a method in which a plurality of vector measurement devices of the present embodiment having different fundamental frequencies are provided, and the calculation result of the vector measurement device closest to the detection frequency is selectively output. In particular, when the parameters (N _w , M _w , N _b , N _b ) are selected under the condition (13), one of the two types of DFT arithmetic units can be used continuously when the fundamental frequency is gradually changed.

  As described above, the vector detection apparatus of the present embodiment can perform vector measurement even with one input signal in the variable frequency range of (15), and can individually detect the frequency, amplitude, and phase. Furthermore, a vector calculation including a differential value and an integral value of the direct measurement signal can be realized in real time.

  Noise cancellation is the most direct application of the vector measurement method of the present invention. In this case, a signal having the same amplitude and the same frequency as that of the measurement signal may be output.

The simplest vector calculation application example is a reactive power control function by a single-phase power converter connected to the end of a power transmission and distribution network. Specifically, the SVC on the pole of the power distribution network and the single-phase power converter for a small-scale photovoltaic power generator for home use are applicable. In these single-phase power converters, the amplitude of the current command value in phase with the detected AC voltage vector V is calculated from the deviation between the DC voltage V _dc of the power converter and the set value. There is a method of receiving a current command value having a quadrature phase with respect to the vector V as a deviation between the amplitude of the AC voltage vector V and a set value, or a command value from an IT system linked to a power transmission and distribution network. The two-axis current command vector thus obtained is converted into an instantaneous value I _ref of the current command using the detection frequency f and phase θ of the AC voltage vector, and the current feedback signal from the current transformer for the instrument is converted into the current command I _ref. The target function can be realized by PWM control of the power converter so as to meet the above. With such a function, it is possible to provide a voltage stabilization function to a device connected to the end of the power transmission and distribution network, which has been a voltage destabilizing element.

  As described above, according to this embodiment, it is possible to measure an AC vector having a variable frequency even with one input signal.

  When measuring each vector from a plurality of input signals that can be expected to have the same frequency in a steady state as in a three-phase AC device, the transient response to a step change of the input signal can be further stabilized. This can be realized by correcting the vector detection signal using the average value of the frequency detection results of each signal.

  FIG. 37 and FIG. 38 show the responses when the three-phase alternating current signals have the same conditions as FIG. 33 and FIG. 34, respectively. The only difference is that the average value of the frequencies of the three phases is applied to all phases and input to (33) for vector correction calculation. Compared with the case of frequency correction for each phase, it can be seen that the response width (difference between the maximum value and the minimum value) is reduced and the stability is improved.

  As an application example of the vector measuring device of the present invention to a three-phase AC device, there is an effective magnetic flux measuring device for an electric device in addition to a protection relay of a power system assumed as an application destination.

  Vector control of AC devices converts terminal voltages and currents into biaxial DC values using the instantaneous symmetric coordinate method based on the instantaneous effective magnetic flux or the internal induced voltage vector. In this way, the AC machine is controlled as a pseudo DC machine.

  The vector control is premised on the vector measurement of the effective main magnetic flux. However, according to the present embodiment, a method of directly measuring the main magnetic flux can be put into practical use.

  Actually, the effective magnetic flux φ or the internal induced voltage E is obtained by calculation. In a steady state, the relationship between the terminal voltage vector V of the winding, the internal induced voltage vector E, the current vector I, and the frequency f can be expressed by (38) with the winding resistance R and leakage inductance L as parameters.

  The effective magnetic flux φ is obtained from the induced voltage vector E by (39).

However, since there was no conventional method for measuring a variable frequency AC signal vector at a response speed required for the current control system which is the innermost feedback loop, the frequency signal f of (38) and (39) is a voltage-current measurement. Actually, it is replaced with the frequency f _ref of the voltage or current command for convenience instead of the frequency detected from the signal.

According to this replacement, for example, when the command frequency f _ref is changed stepwise, both the internal induced voltage E and the effective magnetic flux φ calculated in (38) and (39) change stepwise. This is far from the physical phenomenon inside the AC device, and there is a problem of generating a transient phenomenon far from the vector control theory.

For convenience, if the frequency command value is used instead of the measurement frequency, the relationship (40) between the frequency and the phase is not established. For example, even if the phase changes stepwise, the frequency f _ref does not change in the calculation. Therefore, the vector calculation result of the effective magnetic flux φ is different from the physical phenomenon, and an unexpected transient phenomenon occurs.

  On the other hand, according to the vector measurement method of the present embodiment, it is possible to measure a transient vector change including the relationship (40).

FIG. 39 shows the phase and frequency response when the phase of the AC signal having the reference frequency f _b changes stepwise by 20 degrees. Here, the vector measurement parameter is selected as (18) and n = 2, and the vector of each phase is corrected with the frequency average value of the three-phase signal as in FIG.

Due to the one period of the fundamental frequency f _b a data input period N _w, the response frequency and phase are converged together in one cycle. Since the phase has advanced stepwise, the frequency during the response period of one cycle is higher than f _b . When the fluctuation of the frequency f during this period is integrated by (41), it corresponds to a phase step change of 20 degrees. From this, it can be seen that according to the vector measurement method of the present embodiment, transient vector measurement including the relationship between frequency and phase of (39) can also be realized. By calculating the effective main magnetic flux using this vector measurement method, vector control can be realized faithfully to the theory.

  FIG. 40 shows an example in which the effective magnetic flux vector measuring device is applied to a winding induction machine of a secondary excitation generator / motor. Here, the vector measurement method of FIG. 38 is used. FIG. 40 shows an operation during a transient phenomenon when a three-phase ground fault occurs on the AC system side. The positive phase voltage of the generator at this time and the detected value of the internal induced voltage are shown.

  In order to make the result easy to see, the value is converted into a vector value viewed from a coordinate system rotating at the AC system frequency, and is displayed with the amplitude and phase change of the generator voltage positive phase V and the internal induced voltage E.

  The change of the internal induced voltage phase is suppressed to about 2/3 of the positive phase voltage phase change, and the measurement conforms to the physical phenomenon that “the effective magnetic flux of the electrical equipment and the internal induced voltage do not change as rapidly as the terminal voltage”. I'm getting results.

  From the above, it can be seen that the effective magnetic flux vector / internally induced voltage measuring device of the present embodiment can achieve both the response speed and the stability required for the reference phase of the current control command signal calculation.

  An embodiment for specifically realizing the concept described above will be described below. Each of the following block diagrams has a function of embodying the contents of any of FIGS. 20 to 40 described above, equations (1) to (41), and the conceptual descriptions thereof. That is, in the following block diagrams, when a part of the explanation is omitted, the following explanation is supplemented by referring to the corresponding FIG. 20 to FIG. 40 or the expressions (1) to (41) and the conceptual explanation thereof. The explanation will proceed on the premise of what should be done.

  For example, in the case described in the above description of the concept based on the information of the predetermined diagram of the present case in order to obtain a predetermined numerical value, information corresponding to the diagram is stored in advance in the storage device, Indicates that the stored information is read to obtain a predetermined numerical value, or has an arithmetic function corresponding to an arithmetic expression for obtaining information corresponding to the figure, and executes the arithmetic To obtain the numerical value.

  Also, the following block diagrams are representative descriptions of each function, and each block diagram should not be limited to hardware. That is, for example, in an arithmetic unit having a CPU, Of course, the function may be realized by a program.

  An embodiment corresponding to the invention according to claims 1, 4 and 5 will be described with reference to FIG.

  The data input circuit 10 digitizes the data sampled by the sampler 11 in synchronization with the pulse command 30 output at regular intervals T from the timer circuit 20 via the hold circuit 12 and the A / D converter 13. Then, the digital data 14 is output.

The data table 100 outputs the latest sampling data 110 (x _Nw ) and data 120 (x ₀ ) before N _w samples, and inputs them to the first DFT calculator 200 and the second DFT calculator 300.

The first DFT arithmetic unit 200 and the second DFT arithmetic unit 300 are synchronized with the pulse command 30 and perform non-recursive DFT operations of the basic periods N _b and M _b , respectively, and the results 230 (X _N ) and 330 ( X _M ) is output.

With the above configuration, in this embodiment, the two DFT computing units 200 and 300 share a data string (x ₀ , x ₁ , x ₂ ,..., X _Nw ).

The output correction circuit 400 outputs a vector measurement result 410 (X) from the first DFT calculation result 230 (X _N ) and the frequency detection result 610 (f). Further, a double angle 420 (2 · θ _n ) of the deflection angle θ _n is output. In this embodiment, [exp (j · 2 · θ _n )] is output as the double angle 420 of the deflection angle θ _n .

The comparator 500 calculates the ratio (X _M / X _N ) of the first DFT calculation result 230 (X _N ) and the second DFT calculation result 330 (X _M ) by the divider 501 and doubles the declination angle The comparison part 520 (D) is output by selecting the real part or imaginary part of the ratio according to 420 (2 · θ _n ). Specific examples of the selection method include the previous equation (21) and FIG.

The frequency detection circuit 600 inputs the comparison signal 520 (D) and the double angle 420 (2 · θ _n ) of the declination, and outputs a frequency detection result 610 (f). In this embodiment, a recursive operation is employed to increase the operation efficiency, and the previous value fold (630) of the frequency detection value is calculated using a polynomial product sum operation using the data memories 601 and 602 and the shift register 603 synchronized with the pulse command 30. The input is made to the device 620.

FIG. 2 is a diagram showing the configuration of the data table 100, which includes (N _{w + 1} ) data memories 101 and N _w shift registers 102. The shift register 102 synchronizes with the pulse command 30 from the timer circuit 20 and sequentially inputs the input data (x ₀ , x ₁ , x ₂ ,..., X from address # (N _w ) to # (0). _Nw ) is recorded. Specifically, the memory address # from (N _w-1) # to (N _w), followed by # shifted # from _{(N w-2) (N} w-1), from the last # (0) There is a method of shifting data in # (1). The latest input data 110 (X _Nw ) and data 120 (x ₀ ) before N _w samples are output from this data table.

FIG. 3 is a diagram showing an embodiment of the first DFT calculator 200, which outputs the nonrecursive DFT calculation result 230 (X _N ) of (8). The N _b data memories 201 store the complex coefficients of the DFT operation, and the N _b shift registers 202 synchronized with the pulse command 30 constitute a ring memory. Data memories 203 and 204 are shift memories for recursive DFT operations that are shifted by a register 205 synchronized with the pulse command 30.

With the above configuration, the ring memory outputs the data output 210 at address # (0) and the data output 220 at address # (N _w ), and _{adds /} subtracts the data table outputs 110 (x _Nw ) and 120 (x ₀ ). The result of the recursive DFT operation is stored in the data memory 203 using the multiplier 206 and the multiplier 207. In order to convert the recursive DFT operation result to the non-recursive DFT operation result, the conjugate number of the data table output 210 is calculated by 208, the amplitude is corrected by the coefficient unit 209, and the recursive operation result 203 is multiplied to be non-recursive. An operation result 230 (X _N ) is output.

FIG. 4 is a diagram showing an example of the second DFT calculator 300, which outputs the non-recursive DFT calculation result 330 (X _M ) of (9). The M _b data memories 301 store the complex coefficients of the DFT operation, and the M _b shift registers 302 synchronized with the pulse command 30 constitute a ring memory. Data memories 303 and 304 are shift memories for recursive DFT operations that are shifted by a register 305 synchronized with the pulse command 30.

With the above configuration, the ring memory outputs the data output 310 at address # (0) and the data output 320 at address # (M _w ). In this embodiment, since M _w = N _w is set, the data table output 120 (x _Nw−Mw ) is x ₀ . This and 110 (x _Nw ) are _subjected to a recursive DFT operation using an adder / subtractor 306 and a multiplier 307, and the result is stored in the data memory 303. In order to convert the recursive DFT operation result to the non-recursive DFT operation result, the conjugate number of the data table output 310 is calculated by 308, the amplitude is adjusted by the coefficient unit 309 to obtain a correction coefficient, and the recursive operation result is multiplied. To output a non-recursive operation result 330 (X _M ).

  As described above, according to the embodiments of FIGS. 3 and 4, by converting the recursive DFT operation result to the non-recursive DFT operation result, there is an effect of reducing the number of product-sum operations from N times to 3 times.

FIG. 5 is a diagram showing another embodiment of FIG. 3 and can be applied to the case where N _w = N _b . In this case, since the data memory 220 (#N _w ) and the data memory 210 (# 0) in FIG. 3 rotate once to have the same value, as shown in FIG. 5, a subtracter 2061 and a multiplier 2071 are used. The same result as in FIG. 3 can be obtained. The embodiment of FIG. 5 has the effect of reducing the multiplication once.

FIG. 6 is a diagram showing another embodiment of FIG. 3 and can be applied to the case of N _w = (1/2) × N _b . In this case, since the data memory 220 (#N _w ) and the data memory 210 (# 0) in FIG. 3 are inverted in sign and have the same absolute value, an adder 2062 and a multiplier are used as shown in FIG. The same result as in FIG. 3 can be obtained using 2072. The embodiment of FIG. 6 has the effect of reducing the multiplication once.

Figure 7 is a diagram showing an embodiment of an output correction circuit, the output of the first DFT calculator 230 inputs the (X _N), 401 is the absolute value | X _N | normalized by dividing by deflection angle theta _n Is output as exp (j · θ _n ). This is squared by the multiplier 402 and the double angle 420 is output as exp (j · 2 · θ _n ). From the double angle and the frequency detection value 610 (f), a polynomial product-sum calculator 430 calculates a correction coefficient 440 (H), and a multiplier 403 corrects the first DFT calculation result 230 (X _N ) to perform vector measurement. The result 410 (X = H · X _N ) is output.

FIG. 8 is a diagram showing an embodiment of the polynomial product-sum operation unit 430, which is a circuit for performing a polynomial operation on the correction coefficient H of (33). The polynomial arithmetic circuit is composed of (n + 1) × 4 real type or integer type constant memory 411, multiplier 412 and adder / subtractor 413, and a real number part 421 [cos (2 · θ _n )] of a double angle output 420 and an imaginary number. The real part 441 and the imaginary part 442 of the correction coefficient 440 (H) obtained from the part 422 [sin (2 · θ _n )] and the frequency detection values 610 (f) to (33) are output.

FIG. 9 is a diagram showing an embodiment of the polynomial product-sum operation unit 620. From the circuit for calculating the polynomial of (26), (29) or (27) and (30), and the circuit for calculating the recursive operation expression of (28). Become. The polynomial arithmetic circuit is composed of (n + 1) × 3 real type or integer type constant memory 611, multiplier 612, and adder / subtractor 613, and a real part 421 [cos (2 · θ _n )] of double angle output 420 and an imaginary number. Unit 422 [sin (2 · θ _n )], the previous value 630 (fold) of the frequency detection result, the calculation result 614 of (26) or (27) from the comparator output 520 (D), and (29) or (30 ) Is output, and the frequency detection value 610 (f) is output by the divider 615 and the subtractor 616.

With the above configuration, a variable frequency vector measurement signal can be obtained from a single input signal. According to this embodiment, since the second DFT calculator also uses the same N _w data strings as the first DFT calculator, it is possible to obtain a calculation result with a small harmonic error.

An embodiment corresponding to the invention according to claim 9 will be described below with reference to the drawings. FIG. 10 shows an embodiment when M _w is set in (13). The data table 130 outputs the data 140 (x _Nw−Mw ) before M _w samples in addition to the latest input data 110 (x _Nw ) and data 120 (x ₀ ) before N _w samples.

The first DFT computing unit operates with the same data string (x ₀ , x ₁ , x ₂ ,..., X _Nw ) as in the embodiment of FIG. 1, but the second DFT computing unit 300 is not the data 120. 140 is entered. As a result, the second DFT computing unit 300 is different from FIG. 1 in that the second DFT computing unit 300 computes with a data string (x _Nw−Mw , x _{Nw−Mw + 1} ,..., X _Nw ). Others are the same as those in FIG. 1, and a description thereof is omitted to avoid duplication.

FIG. 11 is a diagram showing an embodiment of the data table 130, and is different from FIG. 2 in that the data (x _Nw−Mw ) at the address # (M _w ) is added as the output signal 140. Others are the same as those in FIG. 2, and a description thereof is omitted to avoid duplication.

According to this embodiment, since the second DFT calculator operates with a shorter data string than in the case of FIG. 1, the degree of freedom in selecting parameters (N _w , M _w , N _b , M _b ) is increased. However, since the latest data M _w are selected, there is an effect that the calculation response is realized without delay.

  An embodiment corresponding to the second aspect of the present invention will be described below with reference to FIG. The same reference numerals as those in FIG. 1 indicate the same contents, and thus description thereof is omitted to avoid duplication.

The part that outputs the vector measurement result 410 and the frequency detection result 610 from the data input circuit 10 is the same as the vector measurement circuit of FIG. The data table 131 has the same configuration as the data table 130, and in this embodiment, the setting value is M _w = (1/2) N _w . As the result data 140, data (x _{Nw / 2} ) is output.

Reference numeral 21 denotes a pulse multiplier which outputs a double frequency pulse command 41 in synchronization with the pulse command 30 and inputs it to the DFT calculators 260 and 360. The DFT calculator 260 has the same configuration as the first DFT calculator 200, except that the pulse command to the shift register 202 is 41. With this configuration, the DFT calculator 260 outputs the DFT calculation output 240 of the basic period (N _b / 2). The DFT calculator 360 has the same configuration as the second DFT calculator 300, except that the pulse command to the shift register 302 is 41. With this configuration, the DFT calculator 360 outputs a DFT calculation output 340 having a basic period (M _b / 2).

Since the output correction circuit 440 has the same configuration as that of 400 and the fundamental frequency is changed from f _b to 2 · f _b , only the value of the constant memory 411 of the polynomial sum-of-products calculator 430 constituting the 440 is different from 400.

The comparator 540 has the same configuration as 500 and outputs the comparison result 550 to the frequency detection circuit 640. Since the frequency detection circuit 640 has the same configuration as 600 and the fundamental frequency is changed from f _b to 2 · f _b , only the value of the constant memory 611 of the polynomial sum-of-products calculator 620 constituting the 640 is different from 600.

  With the above configuration, the second vector measurement result 450 and the second frequency detection result 650 are output and input to the output switching device 700 together with the vector measurement result 410 and the frequency detection result 610.

  The input switching device 700 includes a frequency discriminator 730 and changeover switches 750 and 760. The change-over switches 750 and 760 select 410 and 610 when the frequency discrimination signal 740 (SW) is level 0, and select 450 and 650 when the level determination signal 740 (SW) is level 1.

FIG. 13 is a diagram showing the configuration of the frequency discriminator 730, which includes constant generators 731, 732, changeover switches 733, 734, comparators 735, 736, and a flip-flop 737. The changeover switch 733 selects the frequency detection value 610 when the flip-flop output 740 (SW) is level 0, and the changeover switch 734 selects the second frequency detection value 650 when the flip-flop output 740 (SW) is level 1. . Comparator 735 outputs level 1 when the input exceeds 1.6 × f _b , and comparator 736 outputs level 1 when the input is less than 1.4 × f _b . The outputs of the comparators 735 and 736 are used as the set-side reset input of the flip-flop 737, and the output of the flip-flop 737 is output as the frequency discrimination signal 740 (SW). With this configuration, when 740 (SW) is level 0, the detection frequency 610 can have a token, and when level 740, the second detection frequency 650 can have a token.

As described above, according to the embodiments of FIGS. 12 and 13, the data memories 301 of the DFT computing units 200 and 260 and the DFT computing units 300 and 360 can be shared. In addition, since the basic frequency ratio between the two vector measurement results 410 and 450 is constant at (N _b / M _b ), there is an effect of obtaining an output with stable characteristics.

  FIG. 14 is a diagram showing another embodiment corresponding to the second aspect of the present invention. The same reference numerals as those in FIG.

  The part that outputs the vector measurement result 410 and the frequency detection result 610 from the data input circuit 10 is the same as the vector measurement circuit of FIG. The configuration of the data table 150 will be described later with reference to FIG.

  The DFT calculator 370 is a DFT calculator with a basic period Kb, and has the same configuration as that of the first DFT calculator 200 except for the data memory 201 and the shift register 202 that constitute a ring memory.

Here, DFT calculator 370 sets the Kw and Kb so as to satisfy the relation of _{K w / K b = N w} / N b = M w / M b.

Since the output correction circuit 441 has the same configuration as 400 and the basic period changes from N _b to M _b , only the value of the constant memory 411 of the polynomial sum-of-products calculator 430 constituting the 441 is different from 400.

The comparator 541 has the same configuration as 500 and outputs the comparison result 551 to the frequency detection circuit 641. Since the frequency detection circuit 641 has the same configuration as 600 and the basic period changes from N _b to M _b , only the value of the constant memory 611 of the polynomial product-sum operation unit 620 constituting 641 is different from 600.

  With the above configuration, the second vector measurement result 451 and the second frequency detection result 651 are output and input to the output switching device 701 together with the vector measurement result 410 and the frequency detection result 610.

  The input switching device 701 has the same configuration as the input switching device 700 in FIG. 13, and only the threshold setting of the comparator 731 is different.

Here, if the fundamental frequency corresponding to the fundamental period M _b and f _m, the relationship of f _b <f _m is satisfied. When the intermediate value between f _b and f _m is f _ave , the threshold f _{1 of the} comparator 735 is set to f _ave <f ₁ <f _m, and the threshold f _{2 of the} comparator 736 is set to f _b <f ₂ <f _ave . To do.

FIG. 15 shows an example of the table 150. In addition to the latest input data 110 (x _Nw ), data 120 (x ₀ ) before N _w samples, and data 140 (x _Nw-Mw ) before M _w samples , Data 170 (x _Nw−Kw ) before Kw samples are output.

14 and 15, the number of DFT calculators can be reduced by one as compared with the case of FIG. In addition, since the fundamental frequencies N _b and M _b can be set finely, the phenomenon described with reference to FIGS. 35 and 36, that is, an effect of suppressing a decrease in harmonic tolerance that occurs when the fundamental frequencies deviate greatly from the respective fundamental frequencies is obtained.

  As described above, FIGS. 12 and 14 show an example in which two vector measurement results and two frequency detection results are switched. It can be extended to more output switching in the same way, thereby realizing vector detection with higher stability and higher harmonic tolerance over a wider frequency range.

An embodiment corresponding to the invention according to claim 3 will be described below with reference to FIG. This embodiment shows a case where n = 3. Reference numeral 800 denotes a three-phase vector measurement device, and three input signals 801 (x _a ), 802 (x _b ), and 803 (x _c ) are input, and three vector measurement signals 811 (X _a ) and 812 (X _b ) are input. , 813 (X _c ) and the frequency detection result 820 (f). The vector measuring device 800 inputs the pulse command 30 of the timer circuit 20 to three sets of vector measuring circuits 801, 802, and 803 having the same configuration. The numbers assigned to the devices constituting the vector measuring circuit 801 are the same as those in FIG. 1, and the description thereof is omitted to avoid duplication. The only difference from FIG. 1 is that the frequency input signal 820 of the output correction circuit 400 is input. The vector measurement circuits 801, 802, and 803 output frequency detection signals 611, 612, and 613, respectively, and an adder 831 and a coefficient multiplier 832 output an average value 820. This configuration has the effect of increasing the stability of the detection result as shown in the comparison between FIGS. 37 and 38 and FIGS. 33 and 34.

  An embodiment corresponding to the invention according to claim 6 will be described below with reference to FIG.

A data input circuit 10, a data table 100, a first DFT calculator 200, and a second DFT calculator constituting a measuring apparatus that inputs an input signal (x _in ) and outputs a vector measurement result 410 and a frequency detection result 610. 300, the output correction circuit 400, the comparator 500, and the frequency detection circuit 600 have the same products and configurations as those in FIG.

  In an output interpolation circuit 840 for interpolating and outputting the vector measurement result 410, a frequency multiplication circuit 850 synchronized with the pulse command 30 output from the timer circuit 20 outputs a multiplication pulse command 851 at intervals of Δt.

  The cos function generator 841 outputs cos (2πf · Δt), and the sin function generator 842 outputs sin (2πf · Δt). Since the correction is in a range where the absolute value of (2πf · Δt) is small, the function generators 841 and 842 can be approximated by a quadratic function and a linear function of f, respectively. A correction coefficient 846 [K = exp (j · 2πf · Δt)] is output from these, the unit imaginary number generator 843, the multiplier 844 and the adder 845.

  The data memories 848 and 849 are controlled by the shift register 852 with a multiplication pulse command 851. The result 861 corrected by the multiplier 847 switches the result 861 corrected by the switch circuit 860 and the original vector measurement result 410 and outputs the vector detection result 870.

  The switch circuit 860 outputs the vector measurement result 410 side only during the output of the pulse command 30, and outputs the corrected result 861 for the others.

  As a result, the vector measurement result 410 is output for each calculation cycle of the DFT calculators 200 and 300, and the corrected result 861 is output for each cycle Δt of the multiplication pulse command 851.

  According to the embodiment of the present invention, there is an effect of obtaining an output that is smoothly corrected at a timing obtained by dividing the calculation cycle.

  An embodiment of claim 8 corresponding to the invention of claim 7 will be described below with reference to FIGS.

  Here, an embodiment will be described using an example in which the internal induced voltage / effective magnetic flux vector of a three-phase permanent magnet generator is measured and the internal induced voltage / effective magnetic flux vector measurement result is applied to permanent magnet generator control. The calculation method is the method shown in the previous FIG. 38 and the embodiment shown in the previous FIG.

In FIG. 18, the permanent magnet generator 1010 has an instrument transformer 1011 for the voltage (V _ab , V _bc , V _ca ) between the terminals, and the stator side armature current (I _a , I _b , I _c ). It is connected to the AC side of the first power converter 1013 via the instrument current transformer 1012. The self-extinguishing power element constituting the first power converter 1013 is subjected to pulse width modulation control by a pulse controller 1014. The DC side of first power converter 1013 is connected to the DC side of second power converter 1016 via capacitor 1015, and the AC side is connected to three-phase AC power supply 1020 via harmonic suppression reactor 1017. The

In the permanent magnet generator driving apparatus having the above configuration, the output signals 901 (V _ab ), 902 (V _bc ), 903 (V _ca ) of the instrument transformer 1011 and the output signal 904 of the instrument current transformer 1012 ( I _a ), 905 (I _b ), and 906 (I _c ) are input to the effective magnetic flux vector measuring device 900 of this embodiment. The effective magnetic flux vector measuring device 900 outputs an effective magnetic flux vector 910 (φ) and a frequency 820 (f). The effective magnetic flux vector measuring device 900 will be described later with reference to FIG.

A vector amplitude 931 is output from the effective magnetic flux vector 910 by the absolute value calculator 930, and a deflection angle 941 (θ _φ ) is output from the deflection angle calculator 940. In the case of a permanent magnet generator, the frequency 920 (f) is converted into a speed signal 984 (n) by a constant (Kn) multiplier 983 and output to the system controller 970.

On the other hand, the deflection angle 941 (θ _φ ) is input to the three-phase two-phase converter 950 and the two-phase three-phase converter 960. The three-phase to two-phase converter 950 outputs two-phase current values 951 (I _d ) and 952 (I _q ) according to (42). Here, the current value 951 (I _d ) increases or decreases the effective magnetic flux, and the current value 952 (I _q ) is proportional to the torque of the permanent magnet generator 10.

The effective magnetic flux amplitude 931 and the current value 952 (I _q ) are output to the system controller 970 by the multiplier 981 and the constant (K _φ ) multiplier 982 as the torque feedback value 984 (τ _fB ). φ

The system controller 970 outputs a magnetic flux command 971 (φ _ref ) and a torque command 972 (τ _ref ), and compares them with feedback values 931 and 984 by subtracters 973 and 974, respectively, and sends them to the magnetic flux adjuster 975 and the torque adjuster 977. Output.

The command 977 from the magnetic flux regulator 975 compares the current value 951 (I _d ) with the subtractor 953 and inputs it to the current regulator 954. The current regulator 954 outputs the modulation factor command 956 (α _d ) to the two-phase / three-phase converter 960.

The command 978 from the torque adjuster 977 is compared with the current value 952 (I _q ) by the subtractor 954 and input to the current adjuster 955. The current regulator 957 outputs the modulation factor command 958 (α _q ) to the two-phase / three-phase converter 960.

The two-phase three-phase converter 960 outputs the three-phase modulation rate commands 961 (α _a ), 961 (α _b ), and 963 (α _c ) to the pulse controller 1014 according to (43).

  FIG. 19 shows an embodiment of an effective magnetic flux vector measuring apparatus 900.

The output signals 901 (V _ab ), 902 (V _bc ), and 903 (V _ca ) of the instrument transformer 1011 are converted into phase voltages 804 (V _a ), 805 (V _b ), 806 by a subtracter 910 and a coefficient multiplier 911. (V _c ) and input to the three-phase vector measuring device 800 controlled by the pulse command 30 from the timer circuit 20. The three-phase vector measuring apparatus 800 has the same configuration as that shown in FIG. 16, and the input / output signals having the same numbers represent signals of the same part, and voltage vectors 811 (V _a ), 812 (V _b ), 813 (V _c ) and frequency Output 820 (f) is output.

Armature currents 904 (i _a ), 905 (i _b ), and 906 (i _c ) from the current transformer 1012 are input to vector calculators 907, 908, and 909 having the same configuration, and current vectors 911 ( _{I a), 912 (I b} ), and outputs a 913 (I _c). The vector calculator 907 includes the same data table 100 as in FIG. 1, a first DFT calculator 200, and an output correction circuit 400. The only difference from FIG. 1 is that the output 920 of the vector measuring device 800 is used as a frequency detection value to the output correction circuit 400. The rest of the configuration and operation are the same as in FIG. Description is omitted.

The voltage vectors 811, 812, 813 and current vectors 911, 912, 913 calculated in the above configuration are input to the magnetic flux vector calculators 931, 932, 933. The magnetic flux vector calculators 931, 932, and 933 calculate the magnetic flux vector at (38) and (39), and output the magnetic flux vectors 934 (φ _a ), 935 (φ _b ), and 936 (φ _c ).

  This is input to constant multipliers 941, 942, 943 and an adder 950 to output an effective magnetic flux vector 910 (φ).

  According to the embodiment of the present invention, effective magnetic flux vector measurement is realized by using only the signals of the instrument transformer 1011 and the instrument current transformer 1012 without being affected by the control system command value. For this reason, there exists an effect which implement | achieves the stable magnetic flux measurement.

  According to the embodiment of the present invention, since the frequency detection result obtained by the voltage vector calculation is used at the time of the current vector calculation, the calculation amount can be reduced.

  Although the present embodiment is a three-phase device, since the internal induced voltage and the effective magnetic flux are measured for each of the three phases, it goes without saying that it can be applied to a single-phase device with the same configuration.

10 data input circuit 20 timer circuit 30 pulse command 40 DFT calculation coefficient for pulse command 100,130,150 data table 110 N _w sample before the input data x _Nw
120 Latest input data x ₀
200 First DFT calculator 230 Non-recursive DFT calculation result X _N
300 Second DFT calculator 330 Non-recursive DFT calculation result X _M
400 Output correction circuit 410 Vector calculation result X
430, 620 Polynomial product-sum calculator 500 comparator 510 comparator output D
600 Frequency detection circuit 610 Frequency output f
700 Input switching device 730 Frequency discriminator 800 Three-phase vector measurement device 840 Output interpolation circuit 850 Frequency multiplication circuit
900 Effective magnetic flux vector measurement device 910 Effective magnetic flux vector 950 Two-phase three-phase conversion circuit 960 Three-phase two-phase conversion circuit 970 System control device 1010 Permanent magnet generator 1011 Instrument transformer 1012 Instrument current transformer 1013 First power conversion Unit 1014 Pulse Controller 1015 DC Capacitor 1016 Second Power Converter 1017 AC Reactor 1020 AC System

Claims (21)

A data table for storing past N _w times of measurement data at a fixed interval, and a result X _N obtained by inputting the N _w data in a fixed order and weighting and adding with a complex constant that is repeated in a cycle N _b that is a natural number. In the first DFT calculation means that outputs the fundamental wave component of the Fourier transform and the vector measurement device that outputs the output X _N of the first DFT calculation means as the vector measurement result of the measurement data,
M _b is a natural number different from N _b, and M _w times of data from the data table are input in a fixed order, and the result X _M obtained by weighted addition with a complex constant repeated in the period M _b is the fundamental wave of the discrete Fourier transform. a second DFT calculation means for outputting as a component, a polarization angle calculating means for outputting a deflection angle theta _n of the X _n, the real part or the imaginary ratio of the X _n and the _{X M (X M / X n} ) And a comparator for selecting and outputting a portion, and calculating and outputting the frequency f from the deviation angle θ _n and the output of the comparator. Has a plurality of vector measurement apparatus according to claim 1, unlike the period N _b from each other, it is configured to share the data table, representing the output frequency f out of the plurality of vector measurement device A vector measurement system having a token function and configured to determine whether or not a token is to be delivered and to select a measurement device as a delivery destination according to a detection frequency of a vector measurement device having the token.   The average value of the frequency output from the plurality of vector measuring devices provided for each discrete data string that has the plurality of vector measuring devices according to claim 1 and is expected to have the same frequency is represented by a frequency f. A vector measurement system characterized by output as A data table for storing past N _w times of measurement data at a fixed interval, and a result X _N obtained by inputting the N _w data in a fixed order and weighting and adding with a complex constant that is repeated in a cycle N _b that is a natural number. In the first DFT calculation means that outputs the fundamental wave component of the Fourier transform and the vector measurement device that outputs the output X _N of the first DFT calculation means as the vector measurement result of the measurement data,
Frequency detecting means for detecting the frequency f of the measurement data, declination calculating means for outputting the declination angle θ _n of the X _N , and correction calculation for outputting a correction coefficient H from the declination angle θ _n and the frequency f and means is provided, vector measuring device and outputs the correction result to the coefficient H is obtained by multiplying the output X _N as a vector measurement.   5. The vector measurement device according to claim 4, wherein the frequency is detected using any one of the functions according to claims 1 to 3. 6. The vector measuring device according to claim 4, wherein an interpolation coefficient K is calculated from the frequency f and an elapsed time or moving distance Δt from the latest data measurement value, and the interpolation coefficient K and the correction coefficient are calculated. vector measurement apparatus and outputting a result of multiplying the H on the X _N as a vector measurement.   7. A voltage / current measuring device for measuring a terminal voltage and a coil current of an electric machine applying an electromagnetic induction phenomenon caused by a coil current, and the vector measurement according to claim 1, wherein the terminal voltage measurement signal and the coil current signal are respectively input. In the apparatus, a voltage vector V, a current vector I, and a frequency f are output from the vector measuring device, and an internal induced voltage vector E is calculated and output from the voltage vector V, the current vector I, and the frequency f. Electromechanical vector detection device.   8. The electric machine vector detection device according to claim 7, wherein an effective magnetic flux vector is calculated and output from the frequency f and the internal induced voltage vector E. 2. The vector measurement device according to claim 1, wherein the M _b is a natural number smaller than the N _b , and the M _w satisfies an expression represented by M _w = (M _b / N _b ) × N _w. Characteristic vector measuring device. 2. The vector measurement apparatus according to claim 1, wherein the N _w is a natural number equal to or greater than 4, and n is a natural number from 2 to 4, where the N _w : the M _w : the N _b : the M _b = 1: 1: n: 1, and the comparison means outputs an imaginary part of (X _M / X _N ) as a fixed output. 2. The vector measurement apparatus according to claim 1, wherein N _w is a natural number of 6 or more, and n is a natural number from 2 to 5, where N _w : M _w : N _b : M _b = (n + 1) : (N + 1) :( n + 1): n, and in the comparison means, the imaginary part of (X _M / X _N ), which is the ratio of X _N to X _M , is fixedly output. A vector measuring device. 2. The vector measurement apparatus according to claim 1, wherein the N _w is a natural number of 6 or more, n is 2 or 3, and the N _w : M _w : N _b : M _b = 1: 1: 1. _N is a vector measuring device, wherein the comparison means outputs, as a fixed output, an imaginary part of (X _M / X _N ), which is the ratio of X _N to X _M. 2. The vector measurement device according to claim 1, wherein the N _w is a natural number of 6 or more, and n is a natural number from 2 to 5, where the N _w : the M _w : the N _b : the M _b = n: n: n: 1, and in the comparison means, the real part of (X _M / X _N ), which is the ratio of X _N to X _M , is output as a fixed output. . A data table for storing past N _w times of measurement data at regular intervals, a complex constant table that is repeated with a period N _b that is a natural number, and a constant selection that selects and outputs two or one complex constant from this complex constant table Means, weighted subtracting means for weighting and subtracting data removed from the data table and data added in place of the removed data with a complex constant from the constant selecting means, and the output of the weighted subtracting means itself And a recursive operation means for correcting the output of the constant selection means, and a multiplying means for converting the complex constant from the constant selection means to a conjugate number and multiplying the output of the recursive operation means by means of discrete Fourier transform. A vector measuring device that outputs as a fundamental wave component. A data table for storing measurement data at regular intervals, and a weighted addition of N _w data using complex constants that are repeated with a natural number of periods N _b using at least a part of the N _w data in the data table as input. First DFT computing means for computing X _N which is a fundamental wave component of the discrete Fourier transform corresponding to, and M _b is a natural number different from N _b, and at least a part of M _{w in the} data table Second DFT calculation means for calculating X _M , which is a fundamental component of a discrete Fourier transform corresponding to weighted addition of M _w data, using a complex constant that is repeated at a period M _b with the data of a polarization angle calculating means for calculating a deflection angle theta _n of the X _n, wherein X _n and provided the real part or the selection means for selecting and outputting the imaginary part of the ratio of the X _M, and the deflection angle theta _n Output of the selection means A vector measuring device that calculates a frequency f from A data table for storing measurement data at regular intervals, and a weighted addition of N _w data using complex constants that are repeated with a natural number of periods N _b using at least a part of the N _w data in the data table as input. Outputs a first DFT calculation means for calculating X _N which is a fundamental wave component of discrete Fourier transform corresponding to the above, a frequency detection means for detecting the frequency f of the measurement data, and a declination angle θ _n of the X _N And a correction calculation means for calculating a correction value H from the deviation angle θ _n and the frequency f, and a result obtained by correcting the output X _N with the correction value H is output as a vector measurement value. A vector measuring device characterized by: This corresponds to a weighted addition of N _w data using a complex constant that is repeated with a period N _b that is a natural number, with at least a part of the N _w data from a data table storing measurement data at regular intervals as an input. X _N which is a fundamental wave component of discrete Fourier transform is calculated, and M _b is a natural number different from N _b, and is repeated at a period M _b with at least a part of M _w data in the data table as input. using the complex constants calculates the X _M is a fundamental component of the discrete Fourier transform corresponding to the weighted sum of data M _w of which calculates a deflection angle theta _n of the X _n, wherein said X _n X A vector measurement method in which a real part or an imaginary part of the ratio of _M is selected, and a frequency f is calculated from the deviation angle θ _n and the selection result. Data to be removed from the data table for selecting and outputting two or one complex constants from the complex constant table that is repeated with a period N _b that is a natural number, and storing the past N _w times of measurement data at regular intervals; The data added in place of the data to be removed is subjected to weighted subtraction with the complex constant selected by the constant, and a recursive operation is performed to correct its output based on the result of the weighted subtraction. Is a vector measurement method in which multiplication is performed to convert the result into a conjugate number and the result of the recursive operation is multiplied, and this multiplication is output as a fundamental wave component of discrete Fourier transform. An operation result as a discrete Fourier transform corresponding to N _w pieces of data calculated using a complex constant that is repeated with a natural number N _b as a period from a data table that stores measurement data at regular intervals is stored in the data table. A vector measurement method for correcting as a recursive operation based on data removed from N _w and data replaced with the removed data, and correcting a result of the recursive operation based on a conjugate value of the complex constant.   A first DFT operation means for obtaining a first operation value by performing a discrete Fourier transform operation using a complex constant repeated from the measurement data at a constant interval in a first cycle; and the first cycle from the measurement data; A second DFT operation means for obtaining a second operation value by performing a discrete Fourier transform operation using complex constants repeated at different second periods; and the first operation value and the second operation value. A vector measurement apparatus comprising means for correcting the first calculation value or obtaining a frequency of the measurement data based on a division result.   A discrete Fourier transform operation is performed from complex measurement data repeated at a first cycle from measurement data at a fixed interval to obtain a first calculation value, and a second cycle different from the first cycle is obtained from the measurement data. A discrete Fourier transform operation is performed using a repeated complex constant to obtain a second operation value, and the first operation value is obtained based on a division result of the first operation value and the second operation value. Or a vector measurement method for obtaining the frequency of the measurement data.

Images