CN100520425C

CN100520425C - Post-wavelet analysis treating method and device for electric power transient signal

Info

Publication number: CN100520425C
Application number: CNB2006100205581A
Authority: CN
Inventors: 何正友; 陈小勤; 麦瑞坤; 李小将
Original assignee: Southwest Jiaotong University
Current assignee: Southwest Jiaotong University
Priority date: 2006-03-24
Filing date: 2006-03-24
Publication date: 2009-07-29
Anticipated expiration: 2026-03-24
Also published as: CN1847867A

Abstract

The present invention discloses post-wavelet analysis treating method and device for electric power transient signal. The treating method for electric power transient signal after wavelet analysis and before feeding to the electric power monitoring center includes the following treatment on wavelet coefficient: the extraction of module maximum and the detection of irregularity; the statistics and cluster analysis of wavelet coefficient; neural network classification; energy analysis; and wavelet entropy calculation. The present invention can extract the characteristic of electric power transient signal for the application in traveling wave ranging, fault recognition, electric energy quality analysis and equipment fault diagnosis in the transmission line of power system.

Description

Wavelet analysis post-processing method and device for electric power transient signal

Technical Field

The invention relates to a wavelet analysis post-processing method and a wavelet analysis post-processing device for a power transient signal.

Background

In power systems, transient signals are generated due to faults, adverse power change events, or normal operation of power electronic switching devices, etc. In order to grasp the characteristics of transient signals which may occur and the influence and hazard of the transient signals on a power system, diagnose and predict the operation or fault conditions of a power grid, a power transmission line and large-scale power equipment at an early stage, carry out state maintenance of the power grid and the power equipment and the like, the transient signals of the power are monitored for a long time, and the detected and recorded signals are processed and distinguished so as to provide users or power companies with knowledge of the transient state of the generated power; meanwhile, during large-scale electrical tests, such as high-voltage power equipment, line switching and the like, power transient signals in the test process also need to be acquired; in addition, the power transient signals need to be further studied and analyzed for accurate fault location, line selection, distance measurement, etc. of the transmission/distribution lines. Therefore, the acquisition and analysis of the transient signals are significant to the normal operation of various devices in the power system.

The power transient signal is an unsteady state signal with short duration and strong mutation, has the obvious and outstanding characteristics of a steady state signal different from a normal state, and is applicable to FFT (fast Fourier transform) analysis of steady signal analysis and can not meet the requirement of the power transient signal. At present, most of analysis methods of transient signals are focused on a new tool of wavelet transform, and the analysis methods can simultaneously perform time domain and frequency domain localized fine analysis on the signals, and are particularly sensitive to the change of transient sudden change signals or weak signals, so that the wavelet transform meets the analysis requirements of transient signal sudden change characteristics.

The transient state can be detected by wavelet transformation, but the transformed coefficients are very many, if all the transformed coefficients are taken as characteristics, the performance of the classifier is inevitably reduced, and the method is not suitable for occasions of real-time application and needs dimension reduction; moreover, it is difficult to directly extract or classify the features of the signal components in different frequency bands, and a wavelet transform post-processing method for transient signals is necessary only from two aspects. However, a complete and systematic analysis and processing method and a special device thereof do not exist at present, and the existing wavelet analysis instrument is lack of post-processing analysis and single in function and cannot be directly used for transient analysis of a power system. Therefore, a special device with the function of wavelet analysis and post-processing of the power transient signal is urgently needed to be developed.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a small post-processing method for transient signals of a power system, which can perform wavelet analysis on the transient signals of the power system and perform systematic post-processing on wavelet analysis results, effectively extracts the characteristics of the transient signals of the power and is applied to traveling wave distance measurement and fault identification, power quality analysis and equipment fault diagnosis of a power transmission line of the power system.

The invention solves the technical problem, and adopts the technical scheme that: a power transient signal wavelet analysis post-processing method is characterized in that after the power transient signal is subjected to wavelet analysis, one or more post-processing is carried out on wavelet coefficients, and the result is sent to a power monitoring center for processing; the post-treatment comprises the following steps: (1) extracting a modulus maximum value and detecting singularity to obtain a modulus maximum value point and singularity; (2) statistical processing or cluster analysis of wavelet coefficients; (3) classifying the neural network, namely taking the wavelet coefficient or data after wavelet coefficient post-processing as the input of the neural network and outputting the neural network classification result; (4) energy analysis is carried out to obtain an energy distribution coefficient and an energy fluctuation coefficient; (5) calculating wavelet entropy to obtain wavelet time-frequency entropy or wavelet average entropy or wavelet distance entropy, wherein the specific calculation method of the wavelet time-frequency entropy, the wavelet average entropy and the wavelet distance entropy is as follows:

in the following methods (a) to (c), the wavelet transform coefficients of the signal at the scale a and time t are denoted as WT_x(t，a)；

(a) Wavelet time-frequency entropy

Let the vector be z (r), r ═ 1, 2, …, m_LThen the entropy of the definable vector is

Wherein

P_{z (r)} = {| z (r) |}^{2} / {| | z (r) | |}_{2}^{2};

Signal x (t) discrete wavelet representation WT_x(kT，a＝2^j) Wherein T is a discrete time sampling interval, and a vector sequence can be obtained along variables k and j, then the wavelet time-frequency entropy:

{WTFE}_{x} (kT, j) = [E_{{WT}_{x}} (kT), E_{{WT}_{x}} (a = 2^{j})];

wherein:

(b) mean entropy of wavelets

Wavelet transform coefficient of signal x (t) at scale a and time t is WT_x(t, a) dividing the time-frequency plane (t, a) of the wavelet transform of the signal into N time-frequency window region blocks which are scaled by scale, and the energy in each region block is E_i(i ═ 1, 2,. N), the total energy across the time-frequency plane is

Mean entropy of wavelets

(c) Wavelet distance entropy

Discrete wavelet representation WT for signal x (t)_x(t＝kT，a＝2^j) Where T is the discrete time sampling interval, along the variable k, a vector sequence WT can be obtained_x(k) (ii) a Wavelet distance entropy:

wherein,

d_kl＝‖WT_x(kT)-WT_x(lT)‖，k，l＝1，2，…m。

compared with the prior art, the invention has the beneficial effects that: after wavelet analysis is carried out on the electric power transient signal to obtain a wavelet coefficient, effective characteristics of the electric power transient signal are extracted through various post-processing methods so as to be applied to traveling wave ranging and fault identification, electric energy quality analysis and equipment fault diagnosis of a power transmission line of an electric power system:

(1) the mode maximum value is obtained and is related to a singular point, and the mode maximum value points on all scales are converged into the singular point, so that the mode maximum value can be used for traveling wave distance measurement and signal denoising (removing non-extreme points) by the power monitoring center, and the singularity degree can more directly reflect the fluctuation and the flicker of the voltage;

(2) the coefficients after wavelet transformation are used for statistical classification analysis, so that the high frequency and the low frequency of the acquired voltage and current can be distinguished, and harmonic components are analyzed;

(3) the neural network classification result can be used for fault diagnosis, fault line selection/classification, stability analysis, dynamic safety evaluation of the power system and the like;

(4) the energy distribution coefficient can be used for reflecting the evolution law of the wavelet transform of the transient signal across the characteristic subspace, and the fluctuation coefficient can describe the local fluctuation form of the wavelet transform of the transient signal;

(5) various entropies are related to information, the information is related to control decision, and the state of the power system is related to the development direction of the power system, so that the wavelet entropy is closely related to the research of the power system, and more effective basis is provided for stability evaluation and fault analysis of the power system. The method can obtain three new entropies, and the three entropies reflect the specific characteristics and meanings of the power supply quality of the power system: wavelet time-frequency entropy

The wavelet time-frequency entropy measurement result is composed of two vectors (or sequences), wherein the first vector has time ergodicity, and the second vector has frequency ergodicity. The large entropy value at the moment kT represents that the wavelet coefficients with wide distribution fill a frequency domain space, and on the contrary, the small entropy value represents that the wavelet coefficients are concentrated in a few frequency points (or frequency bands), so that the wavelet time-frequency entropy value can measure the information characteristics of signals at any moment and at any frequency, thereby classifying different signals and having better application potential in the field of fault detection and diagnosis.

Mean entropy of wavelets

The more uniform the energy distribution, the larger the wavelet mean entropy and vice versa. When a system (such as a power transmission line) is normal, current and voltage mainly contain fundamental waves and various harmonics, energy is distributed on a few frequency points and is relatively uneven, so that the average entropy of wavelets is small, and when a fault occurs, the current and voltage contain high-frequency transient state, the frequency domain distribution is wide, and the entropy can be increased.

Wavelet distance entropy

The wavelet distance entropy reflects the difference of information in different scales and different time, so the difference of the wavelet distance entropy of the transient signal and the wavelet distance entropy of the steady-state signal is larger, and the entropy is increased when the transient occurs.

In short, the method can make the power monitoring center obtain comprehensive and complete evaluation for the power supply quality, the system stability and the fault analysis of the power system by comprehensively performing the analysis, and accordingly take corresponding measures to improve the power supply quality and the system stability.

Another object of the present invention is to provide a special apparatus for the wavelet analysis post-processing method of the power transient signal.

The invention discloses a special device for a wavelet analysis post-processing method of a power transient signal, which comprises the following components: the background computer is connected with the preposed data acquisition and processing unit through a USB interface and is connected with the GPS synchronous clock through a universal serial port; meanwhile, the prepositive data acquisition and processing unit is connected with the GPS synchronous clock through a serial port.

The pre-data acquisition and processing unit acquires electric power transient signals, the electric power transient signals are sent to the background computer through the USB interface, the background computer performs wavelet analysis and post-processing on the received data according to the method of the invention, and the GPS synchronous clock synchronizes the time of the pre-data acquisition and processing unit and the background computer. Therefore, the device of the special and efficient wavelet analysis post-processing method of the power transient signal is provided for the power system.

The prepositive data acquisition and processing unit comprises: the input port of the analog quantity shaping module is connected with a power grid mutual inductor, and the output port of the analog quantity shaping module is connected with the input port of the voltage and current high-speed acquisition and processing module; the output port of the synchronous clock interface module is connected with the serial port of the voltage and current high-speed acquisition processing module, and the input port of the synchronous clock interface module is connected with the synchronous clock.

The analog quantity shaping module converts the voltage and current signals of the power grid into a voltage range suitable for A/D acquisition, and the voltage and current high-speed acquisition processing module can acquire the signals sent by the analog quantity shaping module at a high speed, so that the requirement of the invention on high-speed acquisition of power transient signals is met.

The core chip of the prepositive data acquisition processing unit is a TMS320F2812 digital signal processing chip. The mature high-main-frequency special data acquisition and processing chip ensures the reliable performance of the device and easy implementation.

The USB interface chip of the pre-data acquisition processing unit is CY7C 68001. The interface chip ensures that the interface of the prepositive data acquisition processing unit and the background computer meets the USB2.0 standard, the data transmission speed can reach 480Mbps, and the requirement of high transmission rate of the prepositive data acquisition processing unit and the background computer is met.

Drawings

Fig. 1 is a block diagram of hardware configuration according to the embodiment of the present invention.

Fig. 2 is a schematic configuration diagram of a pre-data acquisition processing unit according to an embodiment of the present invention.

Fig. 3 is a background PC wavelet analysis display interface according to an embodiment of the present invention.

Detailed Description

The invention is further described with reference to the following figures and detailed description.

Examples

One specific embodiment of the invention is as follows: a wavelet analysis post-processing method for electric power transient signals is characterized in that after the electric power transient signals are subjected to wavelet analysis, one or more of the following post-processing is carried out on wavelet coefficients, and the results are transmitted to a power monitoring center for processing:

(1) extracting a modulus maximum value and detecting singularity to obtain a modulus maximum value point and singularity;

(2) statistical processing or cluster analysis of wavelet coefficients;

(3) classifying the neural network, namely taking the wavelet coefficient or data after wavelet coefficient post-processing as the input of the neural network and outputting the neural network classification result;

(4) energy analysis is carried out to obtain an energy distribution coefficient and an energy fluctuation coefficient;

(5) and (4) calculating wavelet entropy to obtain wavelet energy entropy, wavelet time entropy, wavelet singular entropy, wavelet time-frequency entropy, wavelet average entropy or wavelet distance entropy.

The specific algorithm of the post-processing method is as follows:

(1) modulus maximum extraction, singularity detection

Wavelet transform of the first step: carrying out wavelet transformation on the collected current and voltage signals i (t), u (t) and t to obtain a transformed function W_f(s，t)。

Second step modulus maximum extraction: if the pair belongs to t₀An arbitrary point t of a certain neighborhood has | W_f(s₀，t)|≤|W_f(s₀，t₀) | title(s)₀，t₀) Is the modulus maximum point of the wavelet transform.

Third step window singular index: i.e. to find the best c and alpha problem. Note the book

Is at t₀Windowed signal discrete sampling of points

Wavelet decomposition of

j＝1，2，…M

The solution for α is performed in three steps:

a) discrete small decomposition of signals

d_{k}^{j} = \max | d_{k}^{j} | > 0,

The problem becomes to optimize c and alpha so that

b) To find

b_{j}^{*} = \log d_{j}^{*},

And b is logc, then there are

c) To solve the set of inequalities, let

Then

The problem becomes to solve a and b so that they satisfy

By using a least square method

(2) And (4) statistical processing and cluster analysis of the wavelet transform coefficients.

Respectively calculating wavelet coefficients D under each scale by referring to a method commonly used in engineering_j(k) Such as:

a) average value:

b) absolute average value:

c) variance:

d) willison magnitude number:

e) zero crossing number:

(3) neural network classification

x_jRepresenting a feature vector (typically wavelet coefficients or other post-processing result), s, for an input signal_iFor external input signals, u_iIs the internal state of the neuron, theta_iAs a threshold, the model can be described as

When a neuron has no internal state, let y_i＝u_iAnd h ═ f, where f is referred to as the activation function.

(4) Energy analysis

For the collected current and voltage, the energy of each scale after wavelet transformation can be directly obtained by the square of the wavelet coefficient. Let d_j，kFor the decomposition coefficient of the signal on the wavelet, based on the energy of each scale of the orthogonal wavelet transform, defining the energy distribution coefficient and the energy fluctuation coefficient respectively as follows:

(5) wavelet entropy calculation: in the following wavelet entropy calculation methods (a) to (c), the wavelet transform coefficient of a signal at the scale a and time t is denoted as WT_x(t, a); in methods (D) to (f), the discrete wavelet coefficients of the multi-resolution analysis of the signal x (t) at the scale j are represented by D ═ D_j(k)，k＝1，…n，j＝1，2…，m}，

Is the wavelet energy spectrum on the scale j.

(a) Wavelet time-frequency entropy

Let the vector be z (r), r 1, 2 … m_LThen the entropy of the definable vector is

Wherein

P_{z (r)} = {{| z (r) |}^{2} / | | z (r) | |}_{2}^{2} .

Discrete wavelet representation WT for the aforementioned signals_x(kT，a＝2^j) Where T is the discrete time sampling interval, a two-dimensional matrix is essential, and vector sequences can be obtained along variables k and j, thus defining a wavelet time-frequency entropy measure W_TFE(wavelet time-Frequencies entry) is:

{WTFE}_{x} (kT, j) = [E_{{WT}_{x}} (kT), E_{{WT}_{x}} (a = 2^{j})]

wherein

(b) Mean entropy of wavelets

Let signals u (t), i (t) be at scale a and time t₀The wavelet transform coefficient of (1) is WT (t, a) divides the time-frequency plane (t, a) of the wavelet transform of the signal into N time-frequency window area blocks kT-a delta which are scaled according to the scale_ψ，KT+aΔ_ψ]×[ω^*/a-Δ_ω/a，ω^*/a+Δ_ω/a]，(Δ_ψ、Δ_ωIs the base wavelet time domain radius and frequency domain radius, omega^*For the center of the frequency domain of the fundamental wavelet, T is the time discrete interval, k is 0, 1 … N discrete sequence), let the energy in each region block be E_i(i 1, 2.. N), the total energy of the whole time-frequency plane is

Normalizing the energy of each block to obtain P_i＝E_iE is then

The normalization condition of the information entropy is met, and then the wavelet mean entropy (WAE) is defined as:

(c) wavelet distance entropy

Discrete wavelet representation WT for power transient signals_x(kT，a＝2^j) Where T is the discrete time sampling interval, a two-dimensional matrix is essential, and the vector sequence WT is obtained along the variable k_x(k) Thus, with reference to the definition of the correlation distance, introducing a calculation method of the information, it is possible to define a wavelet distance entropy measure W_DE(Wavelet Distance Encopy) is:

wherein,

d_kl＝‖WT_x(kT)-WT_x(lT)‖，k，l＝1，2，…m。

(d) entropy of wavelet energy spectrum

u (t), i (t) obtaining wavelet energy spectrum E ═ E on m scale by orthogonal wavelet transform₁，E₂，…，E_m. The total power E of the signal being equal to the power E of the individual components_jAnd (4) summing. Let P_j＝E_jE, then

Wherein

Wavelet entropy W_EE(wavelet energy Engine) is

(e) Wavelet time entropy

Wavelet transform of the first step: u (t), i (t) are signals, and at the scale j, the discrete wavelet transform coefficients of the multi-resolution analysis are D ═ { D (k) }, k ═ 1, 2, …, N }

The second step defines the sliding window: defining a sliding window on the wavelet coefficient, setting the window width as omega ∈ N, and the sliding factor as delta ∈ N, so that the sliding window is W (m ∈ N.)_t；ω，δ)＝{d(k)，k＝1+m_tδ，…，ω+m_tδ}，m_t＝1，2，…，M.

Thirdly, calculating wavelet entropy: let p^mt(Z_l) Represents the wavelet coefficient d (k) epsilon W (m)_t(ii) a ω, δ) falls within the interval Z_lAccording to classical theory of probability, d (k) e W (m)_t(ii) a ω, δ) falls on Z_lNumber of (2) and W (m)_t(ii) a ω, δ) of the total number of wavelet coefficients ω. Then there is the following wavelet Time WTE (wavelet Time entry):

m_t＝1，2，…M

wherein M ═ N-w)/δ ∈ N, and for each scale, its WTE can be calculated accordingly^j(m_t)，m_t＝1，2，…M_j。

(f) Singular entropy of wavelets

Wavelet decomposition of signals u (t), i (t) into D at the scale of j (j ═ 1, 2, …, m)_j(n), the decomposition results at m scales may form an m × n matrix D_m×n＝U_m×1A_l+lV_l×n. Wherein the main diagonal element of the diagonal matrix Λ is defined by_i(i is 1, 2, …, l),

arranging λ in descending order₁≥λ₂≥…λ_lNot less than 0, is the wavelet transformation result matrix D_m×The singular value of (a). The j-th order delta wavelet singular entropy can then be calculated by:

then the wavelet singular entropy:

fig. 1 shows a special apparatus for a wavelet analysis post-processing method of a power transient signal according to the present embodiment, which includes: the background computer PC is connected with the preposed data acquisition processing unit DSP through a USB interface and is connected with the GPS synchronous clock GPS through a universal serial port, and meanwhile, the preposed data acquisition processing unit DSP is connected with the GPS synchronous clock GPS through a serial port. The back office computer PC of this embodiment is also connected to a printer P to facilitate printout of analysis results.

In this embodiment, a PC is used as a background machine, and the whole program can be divided into the following modules according to functions and applications: data acquisition procedures, display procedures, wavelet transform procedures, wavelet post-processing procedures, application procedures, spectral analysis, and other ancillary procedures. Fig. 3 shows the wavelet analyzer interface of the background PC, now separately introduced with several main functions as examples:

a) display function

The function can complete the display of the waveform according to the predetermined sampling rate, the total number of channels, the set number of data reading points each time, the set length of unit data and the like; specifically, it can accomplish the following several tasks: displaying the waveform according to the selected signal channel, regulating the step length of waveform rolling, pausing the waveform rolling, amplifying and reducing the waveform for careful observation when the display is paused, and indicating the amplitude and time of a certain point by moving a red cursor. This series of functions also facilitates the selection of wavelet transform signals, the selection of data points, and fault location in the future.

b) Wavelet transform function

The wavelet transformation function can complete continuous wavelet transformation, discrete wavelet transformation and wavelet packet transformation, and important parameters such as wavelet types, decomposition layer numbers and the like can be selected before transformation.

c) Post-processing functions

The post-processing functions of the software design include: module such as module maximum extraction, singularity detection, statistical treatment or cluster analysis of wavelet transform coefficient, wavelet neural network classification, energy distribution characteristic analysis based on wavelet analysis, wavelet entropy calculation, etc.

d) Application analysis function

Based on the transformation and post-processing, the background machine software is further designed with the following application analysis: traveling wave ranging, fault classification, fault line selection, power quality analysis, data compression and the like. For example, the travelling wave distance measurement is to observe the data after wavelet transformation and modulus maximum extraction, display the waveform thereof to analyze the time of the initial travelling wave and the opposite end reflected wave head reaching the measurement point, determine the wave velocity to measure the distance, and display in the human-computer interface.

In addition, the software functions include fourier transform, spectrum analysis, storage, printing, and help.

Fig. 2 shows that the front-end data acquisition processing unit DSP of the present embodiment is composed of: the input port of the analog quantity shaping module A is connected with a mutual inductor, and the output port of the analog quantity shaping module A is connected with the DAQ input port of the voltage and current high-speed acquisition processing module; the output port of the GPS synchronous clock interface module GI is connected with the serial port of the voltage and current high-speed acquisition processing module DAQ, and the input port is connected with the GPS synchronous clock.

The modules are described in detail as follows:

(1) DAQ (digital addressable Q) of high-speed voltage and current acquisition and processing module

The collection processing module DAQ comprises a core processing chip TMS320F2812, a current and voltage collection unit and a communication unit of a background computer. The module is used for collecting and recording current and voltage of power frequency steady state and fault transient state. The analog-to-digital conversion part of the module adopts A/D embedded in a core processing chip TMS320F 2812. The selection of the sampling frequency in the steady state of the power frequency can be set according to the actual requirement, the sampling conversion time is 80ns, the A/D conversion precision of the analog-to-digital converter is 12 bits, two paths of signals can be collected simultaneously, and 16 paths of analog quantity can be collected in a time-sharing manner.

The initial program is solidified in the BOOT ROM, and the program running in the core processing chip TMS320F2812 describes the work flow of the preposed data acquisition processing unit:

when in normal operation, the system work flow is as follows: the analog quantity shaping module A conditions the signal to be measured to be between 0V and 3V through linear gain amplification and inputs the signal into the AD converter; in the AD converter, an AD clock and a TMS320F2812 internal timer control the analog signal to be converted into a digital signal, data is written into a result buffer register in the AD converter, an interrupt is generated, and the TMS320F2812 is informed to read and process the data; TMS320F2812 calculates the effective value of the acquired data or performs FFT, and determines whether a fault has occurred and whether a fault recording procedure is started according to the result of the calculation.

When no fault exists, A/D is converted continuously, however, TMS320F2812 reads the sampling data and stores the sampling data in a cyclic storage area, and meanwhile, the effective value of the cyclic storage area is calculated or FFT conversion is carried out, so that whether fault recording is started or not is judged.

When a fault signal is detected, TMS320F2812 records the current address, the data recorded before the address is steady-state data, and each path has 12500 words and total 100k words; the data recorded after the address is transient data, each path records 32500 words and 250k words, the wave recording data are temporarily stored in an expanded off-chip storage space, the data are read after a background computer PC sends an instruction, the transmission mode between the front data acquisition processing unit DSP and the background computer PC can select a universal serial port or a USB2.0 mode, wherein the transmission mode of a large amount of data uses a USB2.0 interface, and the speed can reach 480 Mbps.

In addition, the switching value input and output information from the external circuit is connected to a general purpose input and output interface (GPIO) of the TMS320F2812 through the bus driving chip 74LS245 and is refreshed every 1 millisecond for detecting the switching value displacement condition and giving the switching value output information.

(2) GPS clock synchronous interface module (GI)

The GPS clock synchronization interface module (GI) is used for receiving serial time information and accurate second pulse from a GPS power system synchronous clock GPS and generating an accurate clock source. The system mainly comprises a central processing unit CPU, a data memory RAM, a program memory EPROM, a high-precision crystal oscillator and a counter, wherein the data memory RAM, the program memory EPROM, the high-precision crystal oscillator and the counter are connected with the central processing unit CPU, and the central processing unit CPU is an 8051XA series single chip microcomputer produced by Philips company and has the model of PXAC37 KFA; the data memory RAM is 628512 ALP-7; the program memory EPROM is M27C 64; the high-precision crystal oscillator is a 33Z megahertz crystal oscillator; the counter is a 74LS161 circuit.

The program solidified in the EPROM and running in the 8051XA singlechip has the following working process:

when the system runs normally, the CPU receives accurate GPS clock information through the serial port once in 1 second. The error generated by the GPS power system synchronous clock is less than 1 microsecond accurate second pulse, and the second pulse is divided into microsecond pulses through a counter circuit.

When the detected equipment has a fault, the starting signal triggers the counter circuit to stop pulse counting and informs the CPU. At this point, the CPU (8051XA) will detect the activation signal and record the exact trigger time. This is the exact microsecond of the time that the fault occurred.

The CPU further combines the time of year, month, day, minute and second information received every second with the just microsecond time to obtain complete fault occurrence time information.

The CPU will also transmit this time information to the back-end computer using the serial port for printout.

The core chip of the prepositive data acquisition processing unit (DSP) of this embodiment is a TMS320F2812 digital signal processing chip.

TMS320F2812 is a new recently introduced product, which is a high-performance 32-bit processor with static CMOS architecture and can provide 1.5 Million Instructions Per Second (MIPS), 32 × 32 bits per cycle MAC function, 128KW flash memory on chip, 1K word OTP ROM, 4K word BOOT ROM, 18K word SARAM, and 12 bit analog-to-digital converter (ADC) on chip. The off-chip memory is extended with 512k words of RAM.

The USB interface chip of the pre-data acquisition processing unit (DSP) of this embodiment is CY7C 68001.

CY7C68001 is a USB2.0 interface chip of Cypress company, and a USB2.0 transceiver and a USB2.0 serial interface engine SIE are integrated on the chip; its transmission rate can be up to 480Mbps, and its interior has 4K byte FIFO resource, and possesses internal phase-locked loop and synchronous and asynchronous FIFO interface.

Technical indexes of the device of the embodiment are as follows:

1. a main processor: TMS320F2812, dominant frequency 150 MHz;

2. SRAM: in the sheet: 18 Kx 16-bit, 0 wait

Outside the chip: 512 Kx 16-bit, 12ns wait;

3. FLASH: in the sheet: 128K × 16-bits, 36ns wait;

4. ROM: boot ROM in the chip: 4 Kx 16-bit

On-chip OTPROM: 1 Kx 16-bit 36ns wait;

5. A/D: 2 multiplied by 8 channel, 12-bit, 80ns conversion time and 0-3V range in chip

Recording the number of electrical quantities:

firstly, collecting 4-circuit voltage (three-phase voltage and zero-sequence voltage) data;

collecting data of 4 paths of current (three-phase current and zero sequence current);

the sampling frequency can be adjusted at any time according to actual needs, and the maximum sampling frequency can reach 12.5 Msps;

6. asynchronous serial port: 2-channel, RS232/RS422/RS485 configurable

Transmission rate: RS 232: 1 Mbaud; RS422/RS 485: 9.375 Mbaud;

7. USB bus: 1 channel, conforming to the USB2.0 specification, highest transmission rate: 480 Mbps;

8. the linearity of the current transformer and the voltage transformer is better than 0.1 percent, and the current transformer and the voltage transformer are sealed by full resin, have high isolation and strong impact resistance, are small and portable and can be directly welded on a printed circuit board;

9. voltage of

Rated voltage effective input: 220V, 50 HZ;

measurement range: 5% -150% of the nominal value;

loading: less than or equal to 0.1 VA;

precision: plus or minus 0.1 percent in the measurement range;

continuous overpressure: 200% of nominal;

response frequency: the frequency response of the 50HZ system is 45HZ to 1200 HZ;

10. electric current

Rated current active input: 5A 50 HZ;

measurement range: 2% -150% of the nominal value;

loading: less than or equal to 0.01 ohm;

precision: plus or minus 0.1 percent in the measurement range;

continuous overload: 200% of nominal;

thermal overload: 2 times of the nominal value is continuous; 100 times nominal value for 1 second;

11. active and reactive

Rated current, voltage valid input: 0-5A, 0-220V and 50 Hz;

measurement range: active power: 500VA 0-225%; reactive power is 500VA 0-225%;

precision: plus or minus 0.2 percent in the measurement range;

12. the working temperature of the equipment is as follows: 0 to 70 DEG C

13. And (3) carrying out a series of wavelet transformation and post-processing on the acquired data by analysis software running on a background machine:

the functions are as follows: the system not only has the main functions of wavelet transformation, wavelet post-processing, application analysis and the like, but also has various auxiliary functions of Fourier transformation, spectrum analysis, waveform display, data storage, printing and the like;

the applicable conditions are as follows: a prepositive unit with data acquisition or a stored existing data text file;

the applicable environment is as follows: windows2000 or more systems on ordinary computers;

analysis of the object: the method is mainly used for transient signals of the power system and can also be used for processing and analyzing other engineering signals;

installation and maintenance: and the optical disc is installed, and the software can be updated and upgraded.

The invention can be used for transient signal acquisition, wavelet analysis and post-processing in power plants and electrical equipment of transformer substations of various voltage grades, such as high-voltage transmission lines, generators, large-scale power transformers, capacitors, phase modulators and the like, and can realize traveling wave ranging and fault identification, power quality analysis and equipment fault diagnosis of power transmission lines of power systems.

Claims

1. A power transient signal wavelet analysis post-processing method is characterized in that after the power transient signal is subjected to wavelet analysis, one or more post-processing is carried out on wavelet coefficients, and the result is sent to a power monitoring center for processing; the post-treatment comprises the following steps: (1) extracting a modulus maximum value and detecting singularity to obtain a modulus maximum value point and singularity; (2) statistical processing or cluster analysis of wavelet coefficients; (3) classifying the neural network, namely taking the wavelet coefficient or data after wavelet coefficient post-processing as the input of the neural network and outputting the neural network classification result; (4) energy analysis is carried out to obtain an energy distribution coefficient and an energy fluctuation coefficient; (5) calculating wavelet entropy to obtain wavelet time-frequency entropy or wavelet average entropy or wavelet distance entropy, wherein the specific calculation method of the wavelet time-frequency entropy, the wavelet average entropy and the wavelet distance entropy is as follows:

in the following methods (a) to (c), the wavelet transform coefficients of the signal x (t) at the scale a and time t are denoted as WT_x(t，a)；

(a) Wavelet time-frequency entropy

Wherein

P_{z (r)} = {| z (r) |}^{2} / {| | z (r) | |}_{2}^{2};

For signal x (t) discrete waveletThe transform coefficients represent WT_x(kT，a＝2^j) Wherein T is a discrete time sampling interval, and a vector sequence can be obtained along variables k and j, then the wavelet time-frequency entropy:

{WTFE}_{x} (kT, j) = [E_{{WT}_{x}} (kT), E_{{WT}_{x}} (a = 2^{j})];

wherein:

(b) mean entropy of wavelets

Mean entropy of wavelets

(c) Wavelet distance entropy

Discrete wavelet transform coefficient representation WT for signal x (t)_x(t＝kT，a＝2^j) Where T is the discrete time sampling interval, along the variable k, a vector sequence WT can be obtained_x(k) (ii) a Wavelet distance entropy:

wherein,

d_k＝‖WT_x(kT)-WT_x(lT)‖，k，l＝1，2，…m。