A transient homologous waveform analysis method

Abstract

The present invention discloses a transient homologous waveform analysis method, which can resample and synchronously fuse multiple transient homologous recording files to generate a Comtrade waveform file, and use cubic spline interpolation, multi-stage digital filtering and least square curve fitting multiple digital signal processing methods to realize coaxial analysis of multiple homologous waveforms. The present invention solves the problem that multiple homologous waveforms cannot be analyzed on the same time axis and scale due to different sampling rates and asynchronous sampling, and the problem that the fundamental signal comparison analysis accuracy is not high due to differences in sampling circuits and CT characteristics. By comparing and analyzing the fault recording files of the relay protection device based on the transient sampling value of the intelligent recorder, it provides a basis for the intelligent recorder to perform real-time monitoring and early warning of the sampling value of the relay protection device.

Description

Technical Field

[0001] The invention relates to a transient homologous waveform analysis method and belongs to the technical field of electric power automation. Background Art

[0002] The development of self-diagnosis technology for power system automatic devices has laid a foundation for intelligent operation and maintenance and inspection of substations. The accuracy and reliability of power system signal sampling directly affect the risk assessment and accident analysis of substation operation and maintenance.

[0003] With the development of communication technology, it provides a basis for the remote transmission and fusion analysis of the recording data of relay protection devices and intelligent recorders. Intelligent recorders record the changes of various electrical quantities in the whole process before, during and after the system fault at a high sampling rate.

[0004] Therefore, if the recording files of the protection device and the recording files of the intelligent recorder are correlated, compared and analyzed, and cross-checked, the occurrence of intelligent substation failures can be effectively reduced and prevented. However, these homologous waveforms are asynchronously sampled and cannot be directly analyzed on the same time axis. Moreover, due to the different filtering characteristics of their respective sampling loops and the non-periodic transmission characteristics of CT, large differences are caused when comparing and analyzing the fundamental content signals. Summary of the invention

[0005] Purpose: In order to overcome the deficiencies in the prior art, the present invention provides a transient homologous waveform analysis method, which solves the problem that multiple homologous waveforms cannot be analyzed on the same time axis and scale due to different sampling rates and asynchronous sampling, and the problem of low accuracy of fundamental wave signal comparison and analysis due to differences in sampling circuits and CT characteristics. By comparing and analyzing the fault recording files of the relay protection device based on the transient sampling values ​​of the intelligent recorder, a basis is provided for the intelligent recorder to perform real-time monitoring and early warning of the sampling values ​​of the relay protection device.

[0006] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A transient homologous waveform analysis method comprises the following steps:

[0008] Obtain the Comtrade waveform files generated by the same set of disturbances or faults of the protection device and the intelligent recorder, parse the .DAT and .CFG files, and obtain the channel configuration parameters and waveform sampling values ​​of each recording channel.

[0009] The channel configuration parameters of each recording channel and the current sampling values ​​at each moment in the waveform sampling value are used to identify the mutation point position T of the same source fault interval using the current mutation amount. fault .

[0010] In the voltage channel, from the mutation point position Tfault Look forward to find the nearest zero-crossing point T of one cycle from the mutation point ± , with the zero crossing position T ± Take the sampling waveforms of the first P cycles and the last Q cycles of all channels of the protection device and the intelligent recorder as the standard.

[0011] The sampling waveforms of the first P cycles and the last Q cycles of the protection device are subjected to sampling frequency normalization processing to obtain the normalized sampling waveform s(x).

[0012] Calculate the phase angle θ at the first sampling point of the normalized sampled waveform s(x) and the waveform of the first P cycles and the last Q cycles of the intelligent recorder 1 and θ 2 , find the phase angle difference Δθ = θ 1 -θ 2 ; Shift the normalized sampling waveform s(x) by Δθ, and calculate the phase angle θ at the first sampling point after the phase shift 1 and θ 2 , and then calculate the phase angle difference Δθ. If the angle difference Δθ ≥ the threshold, iteratively shift the phase of the normalized sampling waveform s(x) after phase shift until the calculated phase angle difference Δθ is within the threshold. The waveforms of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampling waveform s(x) after the final phase shift of the protection device are output.

[0013] As a preferred solution, it also includes: filtering the waveforms of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampling waveform s(x) after the final phase shift of the protection device using a cascaded FIR low-pass filter.

[0014] As a preferred solution, it also includes: obtaining the parameter x of the signal function expression of the non-periodic component according to the waveform of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampled waveform s(x) after the final phase shift of the protection device 0 、x d , τ, θ 0 and x m The signal function expression of the non-periodic component is as follows:

[0015]

[0016] Where: x(t) is the sampling value at time t, x 0 is a constant DC component, x d is the decaying DC component, τ is the decay period, θ 0 is the initial phase angle of the fundamental wave, l is the natural logarithm, x m is the maximum amplitude of the sinusoidal signal.

[0017] As a preferred solution, the channel configuration parameters of each recording channel and the current sampling values ​​at each moment in the waveform sampling values ​​are used to identify the mutation point position T of the same source fault interval by using the current mutation amount. fault The method comprises the following steps:

[0018] Calculate the current mutation component ΔI point by point from the starting position of each current channel waveform, and find out the mutation component ΔI greater than the fixed value I set The sampling point position is the mutation point position T fault .

[0019] The calculation formula of the mutation component ΔI is as follows:

[0020]

[0021] Wherein, i(t) is the current sampling value at the current sampling moment; is the current sampling value of half a cycle before the current sampling moment; i(tN) is the current sampling value of one cycle before the current sampling moment; N is the number of sampling points per cycle.

[0022] As a preferred solution, I set Set to 0.1A.

[0023] As a preferred solution, P is set to 2 and Q is set to 10.

[0024] As a preferred solution, the method for normalizing the sampling frequency of the sampling waveforms of the first P cycles and the last Q cycles of the protection device comprises the following steps:

[0025] The sampling waveforms of the first P cycles and the last Q cycles of the protection device are resampled by cubic spline interpolation using the sampling rate of the intelligent recorder.

[0026] The cubic spline interpolation formula is as follows:

[0027]

[0028] Where: y i and i+1 is the sampling value of two adjacent sampling points of the protection device, x i and x i+1 Is a protective device i and i+1 The sampling time corresponding to two adjacent sampling points, x is the position to be interpolated, s(x) is the cubic spline interpolation at x, h i is the inverse of the sampling rate of the intelligent recorder, M i and M i+1 is the interpolation coefficient.

[0029] As a preferred solution, the normalized sampled waveform s(x) is subjected to a second-order interpolation phase shift Δθ.

[0030] As a preferred solution, the threshold is set to 1°

[0031] As a preferred solution, the calculation formula of the cascaded FIR low-pass filter is as follows:

[0032]

[0033] Where y(k) is the kth sampling point, y(kj) is the kjth sampling point, N is the sampling rate of the intelligent recorder, k is the sequence number of the kth sampling point, and j is the sequence number of the jth sampling point.

[0034] As a preferred solution, l is taken as 2.7182.

[0035] Beneficial effect: The present invention provides a transient homologous waveform analysis method, which can resample and synchronously fuse multiple transient homologous recording files to generate a Comtrade waveform file, and use cubic spline interpolation, multi-level digital filtering and least squares curve fitting and other digital signal processing methods to achieve coaxial analysis of multiple homologous waveforms, providing a basis for analyzing hidden dangers in the power system and realizing real-time monitoring and early warning of the sampling values ​​of relay protection devices by intelligent recorders. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Schematic diagram of the cubic spline interpolation function.

[0037] Figure 2 It is the original Ua waveform of the same source as the protection device and the acquisition unit.

[0038] Figure 3 It is the waveform after filtering of the same source Ua of the protection device and the acquisition unit.

[0039] Figure 4 It is a waveform with superimposed non-periodic components. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with specific embodiments.

[0041] A transient homologous waveform analysis method comprises the following steps:

[0042] Step 1: Get the Comtrade waveform files generated by the same set of disturbances or faults of the protection device and the intelligent recorder, parse the .DAT and .CFG files, and get the channel configuration parameters and waveform sampling values ​​of each recording channel. According to the current sampling values ​​of each channel at each moment, use the current mutation amount to identify the mutation point position T of the same source fault interval fault .

[0043] According to the formula: Calculate the current mutation component ΔI point by point from the starting position of each current channel waveform, and find out the mutation component ΔI greater than the fixed value I set The sampling point position is the mutation point position T fault , where: i(t) is the current sampling value at the current sampling moment; is the current sampling value of the half cycle before the current sampling moment; i(tN) is the current sampling value of the cycle before the current sampling moment; N is the number of sampling points per cycle, I set To set the mutation value, the default value is 0.1A.

[0044] Step 2: Take the mutation point position T found in step 1 fault According to the formula and conditions: u(t)×u(t+1)<0&&u(t)<0, in the voltage channel, from the mutation point position T fault Look forward to find the nearest zero-crossing point T of one cycle from the mutation point ± , with the zero crossing position T ± Take the sampling waveforms of the first P cycles (P is recommended to be 2) and the last Q cycles (Q is recommended to be 10) of all channels.

[0045] Wherein u(t) and u(t+1) are two adjacent sampling points of the voltage channel, && represents logical AND; the formula and condition represent that the sampling voltage value at time t is a negative value, and the sampling voltage value at time t+1 is a positive value.

[0046] Step 3: Perform sampling frequency normalization processing on the sampling waveforms of the first P cycles and the last Q cycles of the protection device obtained in step 2 to obtain the normalized sampling waveform s(x).

[0047] Get the respective recording sampling rates from the Comtrade files of the protection device and the intelligent recorder, and resample the recording data of the protection device according to the sampling frequency of the intelligent recorder. For example, the sampling rate of the protection device is generally 1200Hz, and the sampling rate of the intelligent recorder is 10000Hz. Use cubic spline interpolation to resample the sampling data of the protection device. Compared with linear interpolation and quadratic interpolation algorithms, cubic spline interpolation can ensure a higher interpolation accuracy during the transient process of the power system. The cubic spline interpolation function is as follows:

[0048]

[0049] Where: y i and i+1 is the sampling value of two adjacent sampling points of the protection device, x i and x i+1 Is a protective devicei and i+1 The sampling time corresponding to two adjacent sampling points, x is the position to be interpolated, s(x) is the cubic spline interpolation at x, h i =x i+1 -x i is the distance between two adjacent sampling values, that is, the node spacing, M i and M i+1 is the interpolation coefficient.

[0050] like Figure 1 As shown in the resampling diagram, f(x) indicates the sampling waveform of the protection device. For example, when the sampling rate of the protection device is 1200Hz, Because x i+1 =x i +h i , substituting into (1) we get:

[0051]

[0052] M i and M i+1 is the second-order derivative at the corresponding sampling point, is the interpolation coefficient, and is solved according to the following matrix equation.

[0053]

[0054] in:

[0055]

[0056] (x i-1 ,y i-1 )、(x i ,y i ) and (x i+1 ,y i+1 ) are three consecutive sampling points

[0057] Step 4: Calculate the phase angle θ at the first sampling point of the normalized sampled waveform s(x) obtained in step 3 and the waveform of the first P cycles and the last Q cycles of the intelligent recorder with the same frequency sampling rate 1 and θ 2 , find the phase angle difference Δθ = θ 1 -θ 2 ; Use second-order interpolation to shift the waveform of the protection device for the first P cycles and the last Q cycles by Δθ. After one phase shift, calculate the phase angle θ at the first sampling point of the intelligent recorder and the protection device. 1 and θ 2, calculate the phase angle difference Δθ. If the angle difference Δθ ≥ 1°, perform the second iterative phase shift until the calculated phase angle difference Δθ is within 1°. Finally, output the first P cycles and the last Q cycles of the synchronously sampled intelligent recorder and protection device waveforms.

[0058] Through steps 1-4, two asynchronously sampled waveform files with different sampling rates are fused to generate a Comtrade file with the same sampling frequency. Steps 1-4 solve the problem of waveform fusion into one sampling frequency. Steps 5 and 6 solve the problems of filtering and superposition components respectively.

[0059] Step 5: Since the cut-off frequency of the sampling circuit of the protection device is much smaller than the cut-off frequency of the sampling circuit of the intelligent recorder, when a fault occurs, the harmonic content of the two homologous waveforms is obviously different, such as Figure 2 As shown in the waveforms of the protection device and the intelligent recorder, in order to improve the fit of the two homologous waveforms and more accurately compare the fundamental waveform, the first P cycles and the last Q cycles of the intelligent recorder and the protection device generated in step 4 can be filtered out of the white noise of the homologous channel using a 200-order finite impulse response cascade FIR low-pass filter. The calculation formula is as follows

[0060]

[0061] Where y(k) is the kth sampling point, y(kj) is the kjth sampling point, N is the number of sampling points per week, which is the sampling rate of the intelligent recorder in this paper, that is, 200. k is the sequence number of the kth sampling point, j is the sequence number of the jth sampling point, The waveform of the same source channel after passing through the cascaded FIR digital filter. Figure 3 As shown, it is the homologous waveform after FIR filtering with the cutoff frequency set to 300Hz. After filtering, the fundamental component waveform of the intelligent recorder and the fundamental component waveform of the protection device have a high degree of fit.

[0062] Step 6: When a fault occurs in the power system, the fault signal often contains non-periodic components that decay exponentially. Due to the different CT characteristics of the protection device and the intelligent recorder and the difference in the filter circuit parameters, the same source waveforms superimposed with non-periodic components will be different, such as Figure 4 To quantitatively analyze the difference between the two, a non-periodic function can be constructed (such as formula 7), and the first P cycles and the last Q cycles of the intelligent recorder and protection device generated in step 4 can be curve fitted to obtain the non-periodic characteristic value x of the protection device and the intelligent recorder. 0 、x d , τ, θ 0 and x m The signal function expression of the attenuated non-periodic component is as follows:

[0063]

[0064] Where x(t) is the sampling value at time t, x 0 is a constant DC component, x d is the decaying DC component, τ is the decay period, θ 0 is the initial phase angle of the fundamental wave, l is the natural logarithm, take 2.7182, x m is the maximum amplitude of the sinusoidal signal

[0065] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A transient homologous waveform analysis method, characterized in that: The steps include: Obtain the Comtrade waveform files generated by the same set of disturbances or faults of the protection device and the intelligent recorder, parse the .DAT and .CFG files, and obtain the channel configuration parameters and waveform sampling values ​​of each recording channel; The channel configuration parameters of each recording channel and the current sampling values ​​at each moment in the waveform sampling value are used to identify the mutation point position T of the same source fault interval using the current mutation amount. fault ; In the voltage channel, from the mutation point position T fault Look forward to find the nearest zero-crossing point T of one cycle from the mutation point ± , with the zero crossing position T ± Take the sampling waveforms of the first P cycles and the last Q cycles of all channels of the protection device and the intelligent recorder as the standard; The sampling waveforms of the first P cycles and the last Q cycles of the protection device are subjected to sampling frequency normalization processing to obtain the normalized sampling waveform s(x); Calculate the phase angles θ1 and θ2 at the first sampling point of the normalized sampled waveform s(x) and the waveform of the first P cycles and the last Q cycles of the intelligent recorder, and find the phase angle difference Δθ=θ1-θ2; shift the phase of the normalized sampled waveform s(x) by Δθ, calculate the phase angles θ1 and θ2 at the first sampling point after the phase shift, and then find the phase angle difference Δθ. If the angle difference Δθ≥threshold, iteratively shift the phase of the normalized sampled waveform s(x) after the phase shift until the obtained phase angle difference Δθ is within the threshold, and output the waveforms of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampled waveform s(x) after the final phase shift of the protection device; The method for normalizing the sampling frequency of the sampling waveforms of the first P cycles and the last Q cycles of the protection device comprises the following steps: The sampling waveforms of the first P cycles and the last Q cycles of the protection device are resampled by cubic spline interpolation using the sampling rate of the intelligent recorder; The cubic spline interpolation formula is as follows: Where: y i and i+1 is the sampling value of two adjacent sampling points of the protection device, x i and x i+1 Is a protective device i and i+1 The sampling time corresponding to two adjacent sampling points, x is the position to be interpolated, s(x) is the cubic spline interpolation at x, h i is the inverse of the sampling rate of the intelligent recorder, M i and M i+1 is the interpolation coefficient.

2. A transient homologous waveform analysis method according to claim 1, characterized in that: Also includes: The waveforms of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampling waveform s(x) after the final phase shift of the protection device are filtered using a cascade FIR low-pass filter.

3. A transient homologous waveform analysis method according to claim 1, characterized in that: Also includes: According to the waveforms of the first P cycles and the last Q cycles of the intelligent recorder and the normalized sampling waveform s(x) after the final phase shift of the protection device, the parameters x0 and x1 of the signal function expression of the non-periodic component are obtained. d , τ, θ0 and x m The signal function expression of the non-periodic component is as follows: Where: x(t) is the sampling value at time t, x0 is the constant DC component, x d is the decaying DC component, τ is the decay period, θ0 is the initial phase angle of the fundamental wave, l is the natural logarithm, x m is the maximum amplitude of the sinusoidal signal.

4. A transient homologous waveform analysis method according to any one of claims 1 to 3, characterized in that: The channel configuration parameters of each recording channel and the current sampling values ​​at each moment in the waveform sampling value are used to identify the mutation point position T of the same-source fault interval by using the current mutation amount. fault The method comprises the following steps: Calculate the current mutation component ΔI point by point from the starting position of each current channel waveform, and find out the mutation component ΔI greater than the fixed value I set The sampling point position is the mutation point position T fault ; The calculation formula of the mutation component ΔI is as follows: Wherein, i(t) is the current sampling value at the current sampling moment; is the current sampling value of half a cycle before the current sampling moment; i(tN) is the current sampling value of one cycle before the current sampling moment; N is the number of sampling points per cycle.

5. A transient homologous waveform analysis method according to claim 4, characterized in that: I set Set to 0.1A.

6. A transient homologous waveform analysis method according to any one of claims 1 to 3, characterized in that: The P is set to 2 and the Q is set to 10.

7. A transient homologous waveform analysis method according to any one of claims 1 to 3, characterized in that: After normalization, the sampled waveform s(x) is phase-shifted by Δθ using second-order interpolation.

8. A transient homologous waveform analysis method according to claim 2, characterized in that: The calculation formula of the cascaded FIR low-pass filter is as follows: Where y(k) is the kth sampling point, y(kj) is the kjth sampling point, N is the sampling rate of the intelligent recorder, k is the sequence number of the kth sampling point, and j is the sequence number of the jth sampling point.

9. A transient homologous waveform analysis method according to claim 3, characterized in that: The threshold is set to 1°, and l is taken as 2.7182.

Images