Power line spike signal test method based on LMS adaptive filter

Abstract

The invention discloses a power line peak signal measurement method based on an LMS adaptive filter, and the method can automatically process the time domain waveform of an oscilloscope obtained in a CE107 test through the LMS adaptive filter in a main control computer, can efficiently separate a power line signal from a peak signal, avoids the manual recognition of a tester, can save hardware equipment, and improves the measurement efficiency. The construction cost of the test system is reduced, and the test requirements in the electromagnetic compatibility field are met.

Description

Technical Field

[0001] The invention relates to the field of electromagnetic compatibility testing, and in particular to a power line peak signal testing method based on an LMS adaptive filter. Background Art

[0002] GIB151B-2013 "Requirements and Measurements for Electromagnetic Emissions and Sensitivity of Military Equipment and Subsystems" proposes test requirements for spike signals that appear in the AC power grid when the equipment is switched on and off, namely CE107: Power Line Spike Signal Conducted Emission Test, and provides a test configuration block diagram, such as Figure 1 During the test, two linear impedance stabilization networks (LISN) were connected to the neutral line N and the live line L of the power supply line of the equipment respectively. An oscilloscope was used to observe the voltage spike signals appearing on the live line L and the neutral line N when the equipment switch was turned on and off.

[0003] The standard clearly states that "the amplitude of the spike signal is based on the voltage that appears at the power supply voltage waveform at the moment of the switching operation, not on the 0V of the oscilloscope's vertical axis", that is, the test should be Figure 5 The peak value is the amplitude a, not the amplitude b, but the standard does not provide a clear test method. For this peak value, the early method was for the tester to manually measure it using the oscilloscope cursor (Marker), which was a complicated test.

[0004] The GJB151B implementation guide states that a hardware notch filter can be used for testing. The test configuration is as follows: Figure 2 As shown. For AC power, when measuring with an oscilloscope, the existence of the power waveform has a great influence on the accurate measurement of the transient amplitude. The smaller the amplitude of the transient being measured, the greater the influence. Because it is difficult to see both large and small signals clearly on an oscilloscope, a notch filter can be used to filter out the waveform of the power frequency before measuring. The commonly used notch filter is Solar's 9533-1, as shown in Figure 3 shown.

[0005] The notch filter is inserted between the oscilloscope probe and the oscilloscope. At this time, the oscilloscope displays only the transient waveform to be measured, while the power waveform is filtered out. Figure 1 The addition of one device to the test configuration shown increases the cost of the test system. Summary of the invention

[0006] The present invention provides a method for testing a power line spike signal of a digital adaptive filter. The test configuration is as follows: Figure 4 As shown, the following steps are included: (1) It includes a first linear impedance stabilization network, a second linear impedance stabilization network, a first load, a second load, a high impedance oscilloscope, a main control computer, and an LMS adaptive filter, wherein the first linear impedance stabilization network is connected to the first load; the second linear impedance stabilization network is connected to the second load; the first linear impedance stabilization network is connected to the live wire L of the power line of the device under test, and the second linear impedance stabilization network is connected to the neutral wire N of the power line of the device under test; one end of the high impedance oscilloscope is connected to the live wire L of the power line connecting the first linear impedance stabilization network and the device under test, and the other end is connected to the ground terminal of the first linear impedance stabilization network; the main control computer includes an LMS adaptive filter and is connected to the high impedance oscilloscope; the EUT switch is turned on and off to obtain a time domain waveform containing a spike signal; (2) Design an LMS adaptive filter; (3) Performing LMS adaptive filtering on the time domain waveform in the data processing software, and obtaining the amplitude of each peak from the filter output; wherein step (1) includes: performing Figure 4 The test configuration shown is connected, and the switch inside the EUT is turned on and off to obtain a time domain waveform containing a spike signal; preferably, step (2) includes: designing an LMS adaptive filter, the basic framework of which is as follows: It mainly consists of two parts: a FIR filter and an adaptive weight estimation mechanism, such as Figure 6 As shown, the calculation equation of the FIR filter is as follows: (1) Where: U(n)—input vector; y(n)—filter output; —filter tap weight vector; The calculation equation of the adaptive weight adjustment mechanism is as follows: (2) (3) Where: — expected response; y(n)—filter output; e(n) — estimation error; μ—Step size parameter.

[0007] Step (3) includes: using the time domain waveform collected in step as the input vector U(n) of the adaptive filter, and using the time domain waveform value U(n-1) at the previous moment as the expected response d(n) of the adaptive filter, then the output y(n) of the filter is the power line signal, and the estimated error e(n) is the peak signal on the power line.

[0008] This technical solution proposes a power line spike signal measurement method based on LMS adaptive filter, which can automatically process the oscilloscope time domain waveform obtained in the CE107 test, and efficiently separate the power line signal and the spike signal, thereby avoiding manual identification by testers, saving hardware equipment, reducing the construction cost of the test system, and meeting the test requirements in the field of electromagnetic compatibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 CE107 power line spike signal (time domain) conducted emission test configuration diagram.

[0010] Figure 2 Diagram of the conducted emission test configuration for power line spike signals (time domain) based on a notch filter.

[0011] Figure 3 Notch filter 9533-1.

[0012] Figure 4 Diagram of the conducted emission test configuration for power line spike signals (time domain) based on LMS filters.

[0013] Figure 5 Test signal waveform example.

[0014] Figure 6 Framework of the LMS adaptive filter.

[0015] Figure 7 CE107 measured oscilloscope waveform.

[0016] Figure 8 Equipment under test (Bull GN-605 socket strip).

[0017] Fig. 9 Spike markers.

[0018] Fig.10 Comparison of filter input and output signals (0.840s~0.8415s).

[0019] Fig.11 Comparison of LMS filter input and output signals (0.840s~0.891s). DETAILED DESCRIPTION

[0020] Example 1.

[0021] The first linear impedance stabilization network is connected to the first load (50Ω); the second linear impedance stabilization network is connected to the second load (50Ω); the first linear impedance stabilization network is connected to the live wire L of the power line of the device under test, and the second linear impedance stabilization network is connected to the neutral wire N of the power line of the device under test; one end of the high-impedance oscilloscope is connected to the live wire L of the power line connecting the first linear impedance stabilization network and the device under test, and the other end is connected to the ground terminal of the first linear impedance stabilization network; the main control computer includes an LMS adaptive filter and is connected to the high-impedance oscilloscope; the LMS adaptive filter consists of two major parts: an FIR filter and an adaptive weight estimation mechanism; the EUT switch is turned on and off to obtain a time domain waveform containing a spike signal; the signal waveform measured by the Bull GN-605 socket strip in the CE107 test is taken, as shown Figure 7 As shown, the device under test is Figure 8 The test uses a 100:1 oscilloscope probe, and it can be seen that the peak voltage of the sine wave is (220×1.414) / 100=3.11V. The waveform conforms to the characteristics of the power grid. During the on-off process of the power strip, some spike signals are generated, and the time range is t=0.863s~0.866s.

[0022] Amplify the signal in the red box, such as Fig. 9 As shown, the amplitudes of 20 larger peaks were manually marked and measured.

[0023] Pick Figure 7 The original voltage data of the waveform shown in the figure has a total of 100,000 points, which is the input vector U(n) of the adaptive filter. The expected signal d(n)=U(n-1). The order of the adaptive filter is 32, and the tap weight vectors of each order are The initial value is 0. Follow the steps below to perform adaptive filtering: 1) Perform FIR filtering according to formula (1): Since the filter calculation requires the use of the first 32 data u(1)~u(32), take n=33 as the first calculation point, then the output of the FIR filter is: Where: —The weight vector No. 33 of the filter, that is, the initial tap weight vector of the FIR filter, which is an all-0 vector; —Voltage vector No. 33, i.e., original voltage data u(1)~u(32); 2) Calculate the estimation error according to formula (2): Take the expected response d(33)=U(32)=1.05, then the estimation error 3) Update the weight vector according to formula (3): take the step size μ as 0.001 4) Bring the updated weights into step 1) and continue to calculate the second calculation point, that is, n=34, that is 5) Repeat the above steps to calculate the output of the FIR filter, and feed back the filter weight vector according to the estimated error to obtain the filter output waveform y(n) and the estimated error output waveform e(n). At the initial time (0.840s~0.8415s), the LMS filter input and output signal comparison is as follows: Fig.10 As shown, it can be seen that with the increase of calculation points (i.e. time), the output signal y(n) of the filter quickly approaches the input signal, while the error estimate quickly drops to near 0V; 6) The entire time domain waveform, LMS filter input and output signal comparison is as follows Fig.11 As shown, it can be seen that in the entire time domain waveform, the output signal y(n) of the filter is a good estimate of the input signal u(n). When a spike signal appears, the output signal y(n) is a good estimate of the 220VAC grid signal, and the estimation error is just the spike signal that appears on the grid. 7) Yes Fig. 9 The spike signal marked in the figure is manually measured and compared with the estimated error waveform output by the LMS filter. The comparison results are shown in Table 1. It can be seen that among the 20 compared spikes, only two spikes have test errors exceeding 5%, and the remaining spikes are all within 5%.

[0024] surface Comparison of manual peak test and LMS automatic peak test .

Claims

1. A power line spike signal testing method based on an LMS adaptive filter, comprising a first linear impedance stabilization network, a second linear impedance stabilization network, a first load, a second load, a high impedance oscilloscope, a main control computer, and an LMS adaptive filter, characterized in that: The first linear impedance stabilization network is connected to the first load; the second linear impedance stabilization network is connected to the second load; the first linear impedance stabilization network is connected to the live wire L of the power line of the device under test, and the second linear impedance stabilization network is connected to the neutral wire N of the power line of the device under test; one end of the high-impedance oscilloscope is connected to the live wire L of the power line connecting the first linear impedance stabilization network and the device under test, and the other end is connected to the ground terminal of the first linear impedance stabilization network; the main control computer includes an LMS adaptive filter and is connected to the high-impedance oscilloscope; the switch of the device under test is turned on and off to obtain a time domain waveform containing a peak signal.

2. The power line spike signal testing method based on LMS adaptive filter according to claim 1, characterized in that: The LMS adaptive filter consists of two parts: FIR filter and adaptive weight estimation mechanism.

3. The power line spike signal testing method based on LMS adaptive filter according to claim 2 is characterized in that: In the data processing software, the time domain waveform is subjected to LMS adaptive filtering, and the filter output obtains the amplitude of each peak; The calculation equation of the FIR filter is as follows: (1) Where: U(n)—input vector; y(n)—filter output; —filter tap weight vector; The calculation equation of the adaptive weight adjustment mechanism is as follows: (2) (3) Where: — expected response; y(n)—filter output; e(n) — estimation error; μ—Step size parameter.

4. The power line spike signal testing method based on LMS adaptive filter according to claim 3 is characterized in that: The collected time domain waveform is used as the input vector U(n) of the adaptive filter, and the time domain waveform value U(n-1) at the previous moment is used as the expected response d(n) of the adaptive filter. The output y(n) of the filter is the power line signal, and the estimated error e(n) is the peak signal on the power line.

Images