Rapid broadband measuring device and method based on ESPRIT algorithm
Technical Field
The invention relates to a broadband measuring device, in particular to a rapid broadband measuring device and a rapid broadband measuring method based on an ESPRIT algorithm.
Background
With the development and utilization of large-scale renewable energy sources and the development of smart power grids, China currently builds an ultra-large-scale complex interconnected power system. The high permeability power electronic device makes the dynamic characteristics of the power system signal become more and more complex, and the frequency band range presents the characteristic of wide frequency
Power electronics can also cause frequency components up to 300Hz to the system, which also causes new grid stability problems. And a large amount of field wave recording data analysis shows that the frequency of the 0-300Hz component is dense, and the frequency and the amplitude can change rapidly along with time (the amplitude can change by 50 percent within 1 s). In addition, more and more power electronic equipment is put into operation, and the operation of the equipment improves the flexibility and the reliability of the operation of the power grid, but also generates a large amount of harmonic pollution to influence the power quality of the power grid. Power electronics also generate a large number of higher harmonics (up to 50 harmonics), which can cause severe harmonic pollution in power systems. Therefore, it is necessary to perform real-time synchronous measurement on the broadband signal of 0-2500Hz to provide data for research on broadband signal source, propagation path, and security control.
In the aspect of a measurement method of a broadband signal, Fast Fourier Transform (FFT) and an improved algorithm thereof are widely used for electrical quantity spectrum analysis of a power system due to the advantages of small calculated amount, easiness in hardware implementation and the like. However, the frequency resolution of the FFT is proportional to the time window length, i.e., a longer time window is required to achieve higher frequency resolution. This results in that it cannot take into account both the measurement accuracy and the tracking speed when measuring the fast changing frequency component of 0-300 Hz. And the method has a good measuring effect on integer subharmonics with large frequency intervals and relatively stable frequency intervals. In addition, wavelet transform and Prony algorithm are also common spectrum analysis methods, but these 2 methods have poor stability and cannot maintain good performance under the condition of low signal-to-noise ratio. The rotation invariant subspace (ESPRIT) algorithm is a spatial spectrum estimation method, which utilizes the orthogonal characteristics of the signal subspace and the noise subspace of a sampling signal, can ensure a shorter time window compared with the method while ensuring a higher frequency resolution, can better track the signal fast dynamic process and reduce the averaging effect of the time window, and has a good application prospect. However, this method is susceptible to noise when dividing the signal subspace and the noise subspace, which affects the performance of the algorithm.
Disclosure of Invention
The invention aims to provide a rapid broadband measuring device and a measuring method based on an ESPRIT algorithm. The method can accurately measure the rapidly-changed broadband signal and rapidly track the rapidly-changed broadband signal. The device has the advantages of short response time and high resolution, and meanwhile, the measurement accuracy of the device under the condition of low signal-to-noise ratio is improved.
In order to solve the existing technical problems. The technical scheme adopted by the invention is as follows:
the fast broadband measuring device based on the ESPRIT algorithm comprises a power supply module, a CPU module, a fast sampling and processing module, a broadband oscillation detection module, a high-speed bus backboard, a man-machine interface module and a data transmission module, wherein the power supply module is used for providing working voltage required by the device; the high-speed bus back board is an access board for all the plug-ins and is used for transmitting data among the plug-ins; the CPU module is inserted and connected on a high-speed bus backboard, converts a small signal sent by the alternating current input module into a digital signal through an analog-to-digital conversion chip AD to finish current and voltage signal sampling, calculates fundamental wave, harmonic wave, inter-harmonic wave, synchronous phasor and power of voltage and current according to the digital sampling signal of the current and voltage, and sends the voltage and the current to the liquid crystal module through the high-speed bus backboard for displaying; the man-machine interface module exchanges data with other modules through a high-speed bus backboard; the rapid sampling and processing module adopts an ESPRIT-based rapid broadband measurement algorithm; the broadband oscillation detection module collects data including a dominant component, an equivalent value and a transient energy flow. The data transmission module is divided into two links of an in-station link and a master station link. And transmitting measurement data, alarm events and file data to the in-station monitoring system, wherein various data are modeled according to a DL/T860 standard. Phasor, inter-harmonic waves and alarm events are transmitted to the master station in real time according to the GB/T26865.2 standard, and a recording wave data file is uploaded. And each dominant component of the inter-harmonic waves and the oscillation power is carried by real-time data frame time division multiplexing, so that compatible transmission with synchronous phasor data is realized.
Further, the man-machine interface module comprises a liquid crystal and a keyboard.
The ESPRIT algorithm of the rapid broadband measuring device based on the ESPRIT algorithm comprises the following steps: firstly, a dynamic model of a broadband multi-frequency signal in the power system is composed of k frequency components and noise components at time t, namely
In the formula: f. ofi,xmi(t) and θi(t) are each si(t) has a frequency, amplitude and phase angle of xm1(t)≥xm2(t)≥…≥xmk(t) of (d). Arranging the sampled data of the voltage or current signals into a sampled data matrix X, and arranging the sampled data of the voltage or current signals into a matrix X consisting of the first L-1 elements of the n +1 th column vector of X because each sampled value can be regarded as the superposition of the instantaneous value of each frequency component and the noise componentn1=[x(n),x(n+1),…,x(n+L-2)]TCan be expressed as:
in the formula: ts is the sampling interval.
The formula (2) is simplified into a matrix form
Xn1=A1S+N1 (3)
In the formula: a. the1Representing the phase angle difference of each frequency component of the reference element when the sampling elements meet the sampling characteristic matrix; s is a column vector consisting of each frequency component; n is a radical of1Is a noise vector.
For X, the same principle appliesnThe matrix X composed of the last L-1 elementsn2Comprises the following steps:
Xn2=A2S+N2 (4)
in the formula: a. the2Is a sampling characteristic matrix; n is a radical of2Is a noise vector.
Thirdly, the corresponding amplitude x can be obtained by the least square method through the frequency of each component of the signal
miAngle of sum
Matrix parameters are defined as shown in equations (5) and (6):
Xs=[x(0),x(1),…,x(L+M-2)]T (6)
fourthly, according to the least square method, a matrix S containing phasor information of each frequency component of the signal is
S=(WTW)-1WTXs=[s1,s2,…,sk]T (7)
The amplitude and phase angle corresponding to each component of the signal are:
and fifthly, in order to improve the measurement performance of the ESPRIT algorithm under the condition of containing noise, a signal frequency component estimation method based on kurtosis is provided.
Performing SVD processing on the L multiplied by M sampling data matrix X:
X=UPV (9)
in the formula: p is singular value matrix, P ═ diag (P)1,p2,…,pL)。
Sixthly, regarding the singular value of X from small to large as a discrete signal, the kurtosis K of the mth singular valuemThe expression is as follows:
in the formula: p ═ p
m-n+1,p
m-n,…,p
mThe m-n +1 th to m-th singular values of the sampled data matrix X are arranged from small to large;
is the mean value of all elements in p; σ is the standard deviation of p, and the expression is shown below.
The following equations (10) and (11) can be obtained:
when the elements in the group p are noise singular values, the kurtosis is due to the fact that no significant impact part of the noise singular values appears overall
Will not generate too large fluctuation and is always at the limit value K
aThe following; when m is
1+1 singular values into the array P, kurtosis
It will increase significantly. Thus by judging kurtosis value K
mCan estimate the frequency of the signal multiplied by a fraction k, i.e.
The invention has the advantages and beneficial effects that:
the invention relates to a rapid broadband measuring device based on an ESPRIT algorithm, which consists of a power supply module, a CPU module, a rapid sampling and processing module, a broadband oscillation detection module, a high-speed bus back plate, a data transmission module and a man-machine interface module. The fast sampling and processing module adopts an ESPRIT-based fast broadband measurement method, can accurately measure and fast track fast-changing broadband signals, estimates the frequency component of the signals through the kurtosis of singular values of the signals, has short response time and high resolution ratio for the fast-changing broadband signals, and improves the performance under the condition of low signal-to-noise ratio. The practical broadband measurement technology is promoted, and a new monitoring means is provided for the power electronic power grid.
Drawings
Fig. 1 is a schematic structural diagram of a fast broadband measurement device based on the ESPRIT algorithm according to the present invention.
Detailed Description
As shown in fig. 1, the fast broadband measurement device based on the ESPRIT algorithm of the present invention includes a power module 1, a CPU module 2, a fast sampling and processing module 3, a broadband oscillation detection module 4, a high-speed bus backplane 5, a human-computer interface module 6, and a data transmission module 7, wherein the power module is used for providing a working voltage required by the device; the high-speed bus back board is an access board for all the plug-ins and is used for transmitting data among the plug-ins; the system is provided with a man-machine interface, so that a rapid sampling and processing module can be combined with a rapid broadband measurement algorithm of ESPRIT, and accurate measurement and rapid tracking of broadband data signals are realized; the CPU module is inserted and connected on a high-speed bus backboard, converts a small signal sent by the alternating current input module into a digital signal through an analog-to-digital conversion chip AD, completes current and voltage signal sampling, calculates parameters such as voltage, fundamental wave, harmonic wave, inter-harmonic wave, synchronous phasor and power of current according to the digital sampling signal of the current and voltage, and sends the parameters to the liquid crystal module for display through the high-speed bus backboard; the man-machine interface module exchanges data with other modules through a high-speed bus backboard; the rapid sampling and processing module adopts an ESPRIT-based rapid broadband measurement algorithm, and has shorter response time and higher resolution ratio for rapidly-changed broadband signals; the broadband oscillation detection module collects data including a dominant component, an equivalent value and a transient energy flow. Detecting the dominant component according to the frequency resolution of 1Hz to realize the accurate estimation of frequency, amplitude and phase; the equivalent value is the comprehensive evaluation quantity of subsynchronous/supersynchronous current, voltage total energy and distortion power thereof, and is used as out-of-limit alarm quantity to be monitored in real time; the data transmission module is divided into two links of an in-station link and a master station link. And transmitting measurement data, alarm events and file data to the in-station monitoring system, wherein various data are modeled according to a DL/T860 standard. Phasor, inter-harmonic and alarm events are transmitted in real time with the main station according to the GB/T26865.2 standard, and a recording wave data file is uploaded. And each dominant component of (inter) harmonic waves and oscillation power is carried by real-time data frame time division multiplexing, so that compatible transmission with synchronous phasor data is realized. The man-machine interface module comprises a liquid crystal and a keyboard.
The ESPRIT algorithm of the rapid broadband measuring device based on the ESPRIT algorithm comprises the following steps:
the dynamic model of the broadband multi-frequency signal in the power system is composed of k frequency components and noise components at time t, that is:
in the formula: f. ofi,xmi(t) and θi(t) are each si(t) has a frequency, amplitude and phase angle of xm1(t)≥xm2(t)≥…≥xmk(t)。
Arranging the sampled data of the voltage or current signals into a sampled data matrix X, and arranging the sampled data of the voltage or current signals into a matrix X consisting of the first L-1 elements of the n +1 th column vector of X because each sampled value can be regarded as the superposition of the instantaneous value of each frequency component and the noise componentn1=[x(n),x(n+1),…,x(n+L-2)]TCan be expressed as:
in the formula: ts is the sampling interval.
The equation (2) is simplified to a matrix form:
Xn1=A1S+N1 (3)
in the formula: a. the1For sampling the characteristic matrix, characterizing the meeting of the sampling elements to the reference element at each frequencyRate component phase angle difference; s is a column vector consisting of each frequency component; n is a radical of1Is a noise vector.
For X, the same principle appliesnThe matrix X composed of the last L-1 elementsn2Comprises the following steps:
Xn2=A2S+N2 (4)
in the formula: a. the2Is a sampling characteristic matrix; n is a radical of2Is a noise vector.
Thirdly, the corresponding amplitude x can be obtained by the least square method through the frequency of each component of the signal
miAngle of sum
Matrix parameters are defined as shown in equations (5) and (6):
Xs=[x(0),x(1),…,x(L+M-2)]T (6)
fourthly, according to a least square method, a matrix S containing phasor information of each frequency component of the signal is as follows:
S=(WTW)-1WTXs=[s1,s2,…,sk]T (7)
the amplitude and phase angle corresponding to each component of the signal are:
and fifthly, in order to improve the measurement performance of the ESPRIT algorithm under the condition of containing noise, a signal frequency component estimation method based on kurtosis is provided.
Performing SVD processing on the L multiplied by M sampling data matrix X:
X=UPV (9)
in the formula: p is singular value matrix, P ═ diag (P)1,p2,…,pL)。
Sixthly, the singularity that X is arranged from small to largeThe value is treated as a discrete signal whose m-th singular value has a kurtosis KmThe expression is as follows:
in the formula: p ═ p
m-n+1,p
m-n,…,p
mThe m-n +1 th to m-th singular values of the sampled data matrix X are arranged from small to large;
is the mean value of all elements in p; σ is the standard deviation of p, and the expression is shown below.
The following equations (10) and (11) can be obtained:
when the elements in the group p are noise singular values, the kurtosis is due to the fact that no significant impact part of the noise singular values appears overall
Will not generate too large fluctuation and is always at the limit value K
aThe following; when m is
1+1 singular values into the array P, kurtosis
It will increase significantly. Thus by judging kurtosis value K
mThe frequency multiplied by the fraction k of the signal can be estimated as: