A three-dimensional high-precision positioning method for multiple sound sources of power transformers based on a cross-shaped microphone array

Abstract

The present invention discloses a three-dimensional high-precision positioning method for multiple sound sources of a power transformer based on a cross-shaped microphone array, comprising the following steps: installing a laser ranging sensor at the center of an azimuth linear array and an elevation linear array, and measuring the distance R from the center of the array to the plane where the sound source is located in real time; forming an azimuth sum and difference beam; forming an elevation sum and difference beam; separating multiple sound sources based on a polyphase filter, and using a polyphase filter to divide the entire Nyquist spectrum into K non-overlapping sub-bands for the azimuth and elevation sum and difference beams respectively; measuring the angle based on a single pulse of azimuth and difference output in the frequency domain; measuring the angle based on a single pulse of elevation and difference output in the frequency domain; calculating the coordinates of the sound source to obtain the three-dimensional coordinates of each sound source. The present invention can realize three-dimensional high-precision positioning of multiple sound sources of a power transformer, support the operation and inspection personnel to quickly lock the fault investigation area in the early stage of abnormality, improve the maintenance efficiency, and avoid the serious harm caused by the gradual deterioration of the transformer.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to a three-dimensional high-precision positioning method for multiple sound sources of a power transformer based on a cross-shaped microphone array. Background Art

[0002] As the core equipment in the power system, the operating conditions of power transformers directly affect the operation of the power grid. Improving the reliability of transformer operation and timely discovering and locating transformer fault areas are crucial to ensuring the safe operation of the power grid. About 30% of transformer faults can be detected and located early through abnormal sound signals. However, due to the superposition of interference sounds in the transformer operating environment, the complex structure of the transformer, and the diverse operating conditions, it is currently impossible to accurately locate multiple sound sources of the transformer, and it is impossible to support maintenance personnel to quickly lock the fault investigation area in the early stage of the fault.

[0003] The sound source localization technology based on microphone array integrates knowledge from multiple disciplines such as array signal processing, audio signal processing, data fusion, and signal detection. At present, microphone array sound source localization technology is mainly divided into three categories: time delay estimation localization algorithm, high-resolution spectrum estimation localization algorithm, and controllable beamforming localization algorithm. The time delay estimation localization algorithm depends on the array structure and the data sampling rate of the microphone. It has limited accuracy and weak anti-interference ability. It is generally only applicable to single-source sound source localization. Among the high-resolution spectrum estimation localization algorithms, the MUSIC algorithm is the most widely used, but this algorithm involves feature decomposition and requires that the signal and noise of the sound source are irrelevant and stable. Due to the large amount of calculation and high requirements for the speech environment, the engineering application is relatively complex. The controllable beamforming algorithm is essentially a maximum likelihood estimation of the power of the speech signal. In actual use, the algorithm requires a global search, the amount of calculation is relatively large, and the real-time performance is poor. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a three-dimensional high-precision positioning method for multiple sound sources of a power transformer based on a cross-shaped microphone array.

[0005] Technical solution: The present invention comprises the following steps:

[0006] (1) A laser ranging sensor is installed at the center of the azimuth and elevation linear arrays to measure the distance R from the center of the array to the plane where the sound source is located in real time;

[0007] (2) Forming azimuth and difference beams;

[0008] (3) Forming elevation and elevation beams;

[0009] (4) Multi-source separation based on polyphase filters: polyphase filters are used to divide the entire Nyquist spectrum into K non-overlapping sub-bands for the azimuth and elevation sum and difference beams respectively;

[0010] (5) Azimuth and difference monopulse angle measurement based on frequency domain output;

[0011] (6) Pitch and difference monopulse angle measurement based on frequency domain output;

[0012] (7) Calculate the coordinates of the sound source. Apply steps (4) to (7) to each sound source of the transformer to obtain the three-dimensional coordinates of each sound source in turn.

[0013] Furthermore, the azimuth and elevation linear array receiving signals defined in step (1) are respectively:

[0014]

[0015]

[0016]

[0017]

[0018] Among them, X is the azimuth array receiving signal, Y is the elevation array receiving signal, x i is the complex amplitude of the ith sound signal, and are the receiving steering vectors of the azimuth array and the elevation array of the i-th acoustic signal, respectively, and f i is the signal spectrum, is the incident azimuth of the signal, is the signal incident pitch angle, c is the speed of sound, d is the array element spacing, M is the number of array elements of the uniform linear array, M can be an even number, Q is the number of sound sources, N 1 and N 2 are the receiving noise of the azimuth and elevation arrays respectively, Re(*) is the real part operation, and the superscript T is the transposition operator.

[0019] Furthermore, it is characterized in that the step (2) comprises:

[0020] (2.1) The difference beam weight is Then the azimuth difference beam output is:

[0021] (2.2) Take the sum beam weight as The azimuth and beam outputs are:

[0022] Furthermore, it is characterized in that the step (3) comprises:

[0023] (3.1) The difference beam weight is Then the elevation difference beam output is:

[0024] (3.2) Take the sum beam weight as Then the pitch and beam outputs are:

[0025] Furthermore, the step (4) is based on the separation of various sound sources using a polyphase filter, including K sub-band processing channels, and there is an analysis filter in each sub-band processing channel.

[0026] Furthermore, it is characterized in that step (4) is based on the separation of each sound source of the multi-phase filter, the 0th sub-band and the K / 2th sub-band are real signals, the remaining sub-bands are complex signals, and the information in the 1st to K / 2-1th sub-bands is the same as the information in the K / 2+1th to K-1th sub-bands.

[0027] Furthermore, it is characterized in that the step (4) is based on the separation of each sound source of the polyphase filter, and each sub-band filter is composed of a prototype low-pass filter h 0 (n) is translated to obtain the filter response h of the qth subband q (n) is:

[0028]

[0029] where w q is the center frequency of the qth subband and

[0030]

[0031] Furthermore, the step (4) comprises:

[0032] (4.1) According to the number of sub-bands K and the sampling frequency f s Design a prototype low-pass filter h with bandwidth B and order N 0 (n), assuming that the system function of the prototype low-pass filter is

[0033]

[0034] Among them, N is a multiple of K, f s =BK;

[0035] (4.2) The orientation and beam and the prototype low-pass filter h 0 (n) multiphase decomposition, i.e.

[0036]

[0037] e q (l) = h 0 (Kl+q)

[0038] Where, q = 0, 1, ... K-1, K is the number of sub-bands, is the qth multiphase signal, e q (l) is the coefficient of the q-th polyphase filter, m and l represent the signal after polyphase decomposition and the length of each subsequence of the filter, respectively;

[0039] (4.3)Join With e q (l) The qth subband time domain signal can be quickly calculated by the following formula:

[0040]

[0041] Here, IDFT stands for Inverse Discrete Fourier Transform.

[0042] Furthermore, the step (5) comprises:

[0043] (5.1) Calculate the azimuth single pulse ratio slope Where λ is the wavelength of the center frequency of the qth sub-band;

[0044] (5.2) Using the formula Estimate the azimuth incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the azimuth and difference beams respectively;

[0045] (5.3) According to the formula Calculate the horizontal coordinate X of the plane where the qth sub-band sound source is located q .

[0046] Furthermore, the step (6) comprises:

[0047] (6.1) Calculate the slope of the pitch single pulse ratio Where λ is the wavelength of the center frequency of the qth sub-band;

[0048] (6.2) Using formula Estimate the pitch incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the elevation and difference beams respectively;

[0049] (6.3) According to the formula Calculate the vertical coordinate Y of the plane where the qth sub-band sound source is located q .

[0050] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: it can realize three-dimensional high-precision positioning of multiple sound sources of power transformers, and the positioning accuracy increases with the increase of the sound source frequency value. The three-dimensional positioning more accurately focuses on the sound source position of multiple types of transformer faults, thereby supporting operation and maintenance personnel to quickly lock the fault investigation area in the early stage of the abnormality, improve maintenance efficiency, and avoid serious harm caused by the gradual deterioration of the transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is the principle diagram of the present invention;

[0052] Figure 2 It is a schematic diagram of the process of the present invention;

[0053] Figure 3 It is a schematic diagram of the channelized structure of multi-phase filtering;

[0054] Figure 4 It is a schematic diagram of the spectrum of the received signal of each array element in the azimuth array;

[0055] Figure 5 It is a schematic diagram of the spectrum of the received signal of each array element in the pitch array;

[0056] Figure 6 It is a schematic diagram of the directivity of the sum and difference beam antenna corresponding to the frequencies of each sound source of the transformer;

[0057] Figure 7 It is the frequency azimuth angle detection curve of each sound source of the transformer;

[0058] Figure 8 It is the pitch angle detection curve of each sound source frequency of the transformer. DETAILED DESCRIPTION

[0059] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0060] like Figure 2 As shown, the present invention discloses a three-dimensional high-precision positioning method for multiple sound sources of a power transformer based on a cross-shaped microphone array, which combines a polyphase filter (PF) with a sum-difference single pulse angle measurement, and includes the following steps:

[0061] (1) A laser ranging sensor is installed at the center of the azimuth array and the elevation array to measure the distance R from the center of the array to the plane where the sound source is located in real time. The distance R from the center of the array to the plane where the sound source is located is measured to be R = 3.4m. The schematic diagram is shown in the figure. Figure 1 shown.

[0062] Assume Q = 2 narrowband far-field signals x 1 , x 2 , respectively in azimuth and pitch angle The incident light is incident on two vertically placed uniform linear arrays consisting of M = 8 omnidirectional receiving elements, and f 1 =500Hz, f 2 =1500Hz, c=340m / s, the distance between adjacent array elements is d=0.0425m, then the received signals of the two microphone arrays can be written as:

[0063]

[0064]

[0065] (2) Forming azimuth sum and difference beams, which specifically includes the following steps:

[0066] (2.1) The difference beam weight is Then the azimuth difference beam output is:

[0067] (2.2) Take the sum beam weight as The azimuth and beam outputs are:

[0068] (3) Forming the elevation sum and difference beams specifically includes the following steps:

[0069] (3.1) The difference beam weight is Then the elevation difference beam output is:

[0070] (3.2) Take the sum beam weight as Then the pitch and beam outputs are:

[0071] (4) Multi-source separation based on polyphase filters. Polyphase filters are used to divide the entire Nyquist spectrum into K non-overlapping sub-bands for the azimuth and elevation sum and difference beams, respectively. Take the azimuth and elevation beams as an example.

[0072] Assume that the sampling frequency of the microphone array is f s =3200Hz, divide its Nyquist spectrum (-1600Hz, 1600Hz) into K=32 sub-bands, then the bandwidth of each sub-band is 100Hz, extract the sub-band where the signal to be tested is located in the last 16 sub-bands, then the first frequency to be tested is in the 21st sub-band, and the second frequency to be tested is in the 31st sub-band.

[0073] (4.1) According to the number of sub-bands K and the sampling frequency f s Design a prototype low-pass filter h with bandwidth B and order N 0 (n), assuming that the system function of the prototype low-pass filter is

[0074]

[0075] Among them, N is a multiple of K, f s =BK;

[0076] (4.2) The orientation and beam and the prototype low-pass filter h 0 (n) multiphase decomposition, i.e.

[0077]

[0078] e q (l) = h 0 (Kl+q)

[0079] Where, q = 0, 1, ... K-1, K is the number of sub-bands, is the qth multiphase signal, e q (l) is the coefficient of the q-th polyphase filter, m and l represent the signal after polyphase decomposition and the length of each subsequence of the filter, respectively;

[0080] (4.3)Join With e q (l) The qth subband time domain signal can be quickly calculated by the following formula:

[0081]

[0082] Here, IDFT stands for Inverse Discrete Fourier Transform.

[0083] The received signal is divided into multiple sub-bands using a polyphase filter. Only one filter is needed to replace the traditional filter bank. The sampling rate conversion property can be used to move the downsampling to before filtering, thereby effectively reducing the amount of calculation and improving the operation efficiency.

[0084] The received signal is divided into 32 sub-bands using a polyphase filter. All sub-band spectra have been moved to the vicinity of zero frequency. The 0th and 16th sub-bands contain real signals, while the remaining sub-bands contain complex signals, and half of the information is repeated. In subsequent processing, only the sub-bands where the signal to be tested is located can be extracted from the 16th to 31st sub-bands.

[0085] (5) Azimuth and difference single pulse angle measurement based on frequency domain output, taking the qth sound source output of the qth sub-band of the multi-phase filter as an example.

[0086] (5.1) Calculate the azimuth single pulse ratio slope Where λ is the wavelength of the center frequency of the qth sub-band;

[0087] (5.2) Using the formula Estimate the azimuth incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the azimuth and difference beams respectively;

[0088] (5.3) According to the formula Calculate the horizontal coordinate X of the plane where the qth sub-band sound source is located q .

[0089] (6) Pitch and difference single pulse angle measurement based on frequency domain output, taking the qth sound source of the qth sub-band output of the multi-phase filter as an example.

[0090] (6.1) Calculate the slope of the pitch single pulse ratio Where λ is the wavelength of the center frequency of the qth sub-band;

[0091] (6.2) Using formula Estimate the pitch incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the elevation and difference beams respectively;

[0092] (6.3) According to the formula Calculate the vertical coordinate Y of the plane where the qth sub-band sound source is located q .

[0093] (7) Calculate the coordinates of the sound source. The three-dimensional coordinates of the qth sound source are (R, X q , Y q ).

[0094] In this specific embodiment, the effectiveness of this method is further verified by computer simulation. The parameter settings of this simulation experiment are shown in Table 1:

[0095] Table 1 System simulation parameters

[0096]

[0097] like Figure 3 The following is a schematic diagram of the multiphase filter channelization structure. This simulation is based on Figure 2 and Figure 3 The combined positioning method first forms azimuth and elevation sum and difference beams of the azimuth and elevation respectively, and then the time domain signals of each sum and difference beam are divided into multiple sub-bands through PF; then, the azimuth frequency domain sum and difference single pulse angle measurement and the elevation frequency domain sum and difference single pulse angle measurement are performed in the sub-band where each sound source of the transformer is located to obtain the azimuth and elevation incident angles of the sub-band sound source; finally, the distance information between the array and the plane where the transformer is located is used to realize the three-dimensional high-precision positioning of each sound source, such as Figure 4 , Figure 5 shown.

[0098] like Figure 6The figure shows the sum and difference beam antenna directivity diagram corresponding to the sound source frequency, as shown in Figure 7 The corresponding azimuth angle detection curve is shown as Figure 8 The corresponding pitch angle detection curve is shown in Figure 1. It can be seen that the main lobe of the sum beam is aligned with 0 degrees, while the difference beam forms a notch at 0 degrees. It should be noted that the single pulse angle measurement uses the linear interval of the angle detection curve near the main lobe. Therefore, the higher the sound source frequency, the narrower the main lobe, the larger the k value, and the higher the angle measurement accuracy. Generally, the linear interval of angle measurement is taken within the range of 3dB attenuation of the main lobe; then Figure 6 Medium 1 The corresponding angle measurement range is approximately between (-20°, 20°), f 2 The corresponding angle measurement range is approximately between (-15°, 15°).

[0099] According to the method proposed in the present invention, f 1 Measured The spatial coordinates are converted to (3.4m, 0.2669m, 0.0593m), which have only 0m, -0.0306m and -0.0143m errors on the R, X and Y axes respectively compared with the real spatial coordinates (3.4m, 0.2975m, 0.0736m). 2 Measured, It is converted into spatial coordinates (3.4m, 0.1187m, 0.6055m), which have only errors of 0m, 0.0097m and 0.006m on the R, X and Y axes respectively compared with the real spatial coordinates (3.4m, 0.1090m, 0.5995m). The comparison between the real value and the estimated value is shown in Table 2.

[0100] Table 2 Comparison of actual value and estimated value

[0101] <![CDATA[Sound source f to be measured 1 > <![CDATA[Sound source f to be measured 2 > R(m) 3.4 3.4 True azimuth (°) 5 2 Measured azimuth (°) 4.4892 1.8364 True pitch angle (°) 1 10 Measured pitch angle (°) 1.2408 10.0975 Real space coordinates (m) (3.4,0.2975,0.0736) (3.4，0.1090,0.5995) Measured spatial coordinates (m) (3.4,0.2669,0.0593) (3.4，0.1187,0.6055) Error(m) (0，-0.0306，-0.0143) (0,0.0097,0.006)

Claims

1. A three-dimensional high-precision positioning method for multiple sound sources of power transformers based on a cross-shaped microphone array, characterized in that: The steps include: (1) A laser ranging sensor is installed at the center of the azimuth and elevation linear arrays to measure the distance R from the center of the array to the plane where the sound source is located in real time; (2) Forming azimuth sum and difference beams; (3) Forming elevation and elevation beams; (4) Multi-source separation based on polyphase filters: polyphase filters are used to divide the entire Nyquist spectrum into K non-overlapping sub-bands for the azimuth and elevation sum and difference beams respectively; (5) Azimuth and difference monopulse angle measurement based on frequency domain output; (6) Pitch and difference monopulse angle measurement based on frequency domain output; (7) Calculate the coordinates of the sound source. Apply steps (4) to (7) to each sound source of the transformer to obtain the three-dimensional coordinates of each sound source in turn. The step (4) comprises: (4.1) According to the number of sub-bands K and the sampling frequency f S Design a prototype low-pass filter h0(n) with bandwidth B and order N. Assume that the system function of the prototype low-pass filter is Among them, N is a multiple of K, f s =BK; (4.2) Perform multiphase decomposition on the orientation and beam and the prototype low-pass filter h0(n), that is, And q (l)=h0(Kl+q) Where, q = 0, 1, ... K-1, K is the number of sub-bands, is the qth multiphase signal, e q (l) is the coefficient of the q-th polyphase filter, m and l represent the signal after polyphase decomposition and the length of each subsequence of the filter, respectively; (4.3)Join With e q (l) The qth subband time domain signal can be quickly calculated by the following formula: Wherein, IDFT stands for Inverse Discrete Fourier Transform; The step (5) comprises: (5.1) Calculate the azimuth single pulse ratio slope Where λ is the wavelength of the qth sub-band center frequency, d is the distance between adjacent array elements, Q sound sources are incident on two vertically placed uniform linear arrays composed of M omnidirectional receiving array elements, and M is the number of array elements in the uniform linear array; (5.2) Using the formula Estimate the azimuth incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the azimuth and difference beams, respectively, and Re(*) is the real part operation; (5.3) According to the formula Calculate the horizontal coordinate X of the plane where the qth sub-band sound source is located q ; The step (6) comprises: (6.1) Calculate the slope of the pitch single pulse ratio Where λ is the wavelength of the qth sub-band center frequency, d is the distance between adjacent array elements, and M is the number of array elements of the uniform linear array; (6.2) Using formula Estimate the pitch incident angle of the qth sound source in are the frequency domain outputs of the qth subband of the elevation and difference beams, respectively, and Re(*) is the real part operation; (6.3) According to the formula Calculate the vertical coordinate Y of the plane where the qth sub-band sound source is located q .

2. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1 is characterized in that: The azimuth and elevation linear array receiving signals defined in step (1) are: Among them, X is the azimuth array receiving signal, Y is the elevation array receiving signal, x i is the complex amplitude of the fth sound signal, and are the receiving steering vectors of the f-th acoustic signal azimuth array and elevation array, respectively, i is the signal spectrum, is the incident azimuth of the signal, is the signal incident elevation angle, c is the speed of sound, d is the spacing between adjacent array elements, M is the number of elements of the uniform linear array, M is an even number, Q is the number of sound sources, N1 and N2 are the receiving noises of the azimuth and elevation arrays respectively, Re(*) is the real part operation, and the superscript T is the transpose operator.

3. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1 is characterized in that: The step (2) comprises: (2.1) The difference beam weight is Then the azimuth difference beam output is: X is the azimuth array receiving signal, M is the number of array elements of the uniform linear array; (2.2) Take the sum beam weight as The azimuth and beam outputs are: X is the azimuth array receiving signal, and M is the number of array elements of the uniform linear array.

4. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1 is characterized in that: The step (3) comprises: (3.1) The difference beam weight is Then the elevation difference beam output is: Y is the received signal of the elevation array, and M is the number of elements of the uniform linear array; (3.2) Take the sum beam weight as Then the pitch and beam outputs are: Y is the received signal of the elevation array, and M is the number of elements of the uniform linear array.

5. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1, characterized in that: The step (4) is based on the separation of various sound sources using a polyphase filter, and includes K sub-band processing channels. There is an analysis filter in each sub-band processing channel, where K is the number of divided sub-bands.

6. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1, characterized in that: The step (4) is based on the separation of each sound source by a multi-phase filter, the 0th sub-band and the K / 2th sub-band are real signals, the remaining sub-bands are complex signals, and the information in the 1st to K / 2-1th sub-bands is the same as the information in the K / 2+1th to K-1th sub-bands.

7. The method for three-dimensional high-precision positioning of multiple sound sources of power transformers based on a cross-shaped microphone array according to claim 1, characterized in that: The step (4) is based on the separation of each sound source by a polyphase filter. Each sub-band filter is obtained by shifting a prototype low-pass filter h0(n). The filter response h of the qth sub-band is q (n) is: where w q is the center frequency of the qth sub-band, K is the number of sub-bands, and

Images