A control method and system for distinguishing transformer winding and core vibration signals

Abstract

The present invention discloses a control method and system for distinguishing transformer winding and core vibration signals, the method comprising the following steps: collecting multiple groups of transformer mixed vibration signals and performing signal preprocessing; performing VMD decomposition on each group of transformer mixed vibration signals to decompose multiple subspaces; calculating the mutual information value of the core and winding signals in each subspace; normalizing the mutual information value of the core and winding signals in each subspace to reconstruct the transformer mixed vibration signal; performing FASTICA algorithm decomposition on each group of reconstructed signals to obtain a separation matrix W to separate the vibration signals of the core and winding. The present invention can find the independent parts between the core and winding signals in each group of transformer mixed vibration signals and separate the signals by using the method of independent component separation, directly output the transformer winding vibration signal and the core vibration signal, can better retain the information component of the original signal, and more effectively decompose the vibration signal.

Description

Technical Field

[0001] The present invention relates to the technical field of vibration signal detection of electric power equipment, and in particular to a control method and system for distinguishing vibration signals of transformer windings and iron cores. Background Art

[0002] At present, the sensors used for transformer vibration detection are mainly wired vibration sensors. In practical applications, multiple vibration sensors need to be arranged to accurately detect transformers. Moreover, many transformers are mostly installed outdoors. The use of wired vibration sensors has problems such as complex wiring and difficult installation. With the rapid development of sensor technology, some wireless vibration sensors are also used in a small range for transformer vibration detection, but most of these sensors are battery-powered. The use of wireless vibration sensors requires regular battery charging or battery replacement, which limits the single operation time of the sensor system. At this stage, passive wireless vibration sensor technology is developing rapidly, and some passive wireless vibration sensors used in transformers have also appeared, but these sensors directly output mixed vibration signals of the transformer core and windings, resulting in cumbersome subsequent analysis steps of the transformer vibration signal.

[0003] First of all, the existing technology collects mixed vibration signals of transformers. The mixed vibration signals of transformers are composed of the core signal generated by the magnetostriction of the core and the winding vibration signal caused by the dynamic electromagnetic force generated between the windings, between the coils, and between the turns when the current passes through the windings. In addition, there are some power supply noise, environmental noise, and electromagnetic interference that make the transformer signal more complicated. The complex transformer signal can easily interfere with the subsequent analysis or make the steps cumbersome.

[0004] The signal of the transformer winding is mainly in the frequency band of 100Hz. Since the strain force of the winding insulation pad has certain nonlinear characteristics, the transformer winding vibration signal has high-order harmonic components, but they are much smaller than the fundamental frequency components. The vibration signal of the iron core is mainly generated by the magnetostriction of the ferromagnetic material. Because the magnetic field change has a strong nonlinearity, the iron core vibration signal, in addition to the 100Hz fundamental frequency, also contains 200Hz, 300Hz and 400Hz and other high-order harmonic components, and they are all very large. Therefore, the spectral characteristics of the transformer's mixed vibration signal are mainly concentrated in the medium and low frequencies, and due to the similarity of the iron core vibration signal and the winding vibration signal characteristics, the existing technology can judge the transformer fault from the abnormality of the mixed vibration signal, but the specific fault location accuracy needs to be improved. Although for mixed signals, the presence of high-frequency components can be used to determine faults such as loose core, it is difficult to determine whether the fault occurs in the winding or the core for winding faults and some faults that do not have high-frequency components. For example, if the fundamental frequency amplitude increases, it is difficult to determine whether it is caused by a winding fault or a core fault. The fundamental frequency amplitude of a loose core will increase, and the fundamental frequency amplitude of a loose winding will also increase. Summary of the invention

[0005] In order to overcome the defects and shortcomings of the prior art, the present invention provides a control method and system for distinguishing the vibration signals of transformer windings and cores. The present invention can find the independent parts between the core and winding signals in each group of transformer mixed vibration signals and then separate the signals by separating the independent components. The transformer winding vibration signals and core vibration signals are directly output, which can better retain the information components of the original signals and can more effectively decompose the vibration signals.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a control method for distinguishing transformer winding and iron core vibration signals, comprising the following steps:

[0008] Collect multiple groups of transformer mixed vibration signals, and perform signal preprocessing on each group of transformer mixed vibration signals collected, including denoising filtering and signal amplification processing;

[0009] Perform VMD decomposition on the mixed vibration signals of each group of transformers to decompose multiple subspaces;

[0010] The mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, and the mutual information value of the core and winding signals in each subspace is obtained;

[0011] Normalize the mutual information values ​​of the core and winding signals in each subspace, set a threshold for selecting the mutual information value, reconstruct the subspace corresponding to the selected threshold, and obtain a reconstructed signal of the transformer mixed vibration signal;

[0012] The FASTICA algorithm is used to decompose each group of reconstructed signals, and the separation matrix W is obtained to separate the vibration signals of the core and the winding. The vibration signals with higher harmonic content are output as core vibration signals, and the vibration signals with lower harmonic content are output as winding vibration signals.

[0013] As a preferred technical solution, when VMD decomposition is performed on each group of transformer mixed vibration signals, the number of modal components is set, and each group of transformer mixed vibration signals is decomposed into multiple subspaces according to the number of modal components, and the subspace bandwidth is set based on the normal signal spectrum range of the transformer.

[0014] As a preferred technical solution, the mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, and the mutual information value of the core and winding signals in each subspace is obtained. The specific calculation formula is expressed as follows:

[0015]

[0016] Among them, I(X;Y) represents the mutual information value of the core and winding signals in each subspace, X is the transformer core source signal, Y is the transformer winding source signal, p(x,y) is the joint probability density function of the two, p(x) and p(y) are the marginal probability density functions of the transformer core source signal and the transformer winding source signal, respectively.

[0017] The present invention also provides a control system for distinguishing transformer winding and core vibration signals, comprising: a vibration sensor, a signal conditioning module, a signal processing control module, a power management module, a wireless data transmission module and a GSM SMS alarm module;

[0018] The vibration sensor is connected to the signal conditioning module, the signal conditioning module is connected to the signal processing control module, the signal processing control module is respectively connected to the wireless data transmission module and the GSM short message alarm module, and the power management module is respectively connected to the vibration sensor, the signal conditioning module, the signal processing control module, the wireless data transmission module, and the GSM short message alarm module;

[0019] The vibration sensor is used to collect multiple groups of transformer mixed vibration signals;

[0020] The signal conditioning module is used to perform signal preprocessing on the collected mixed vibration signals of each group of transformers, including denoising filtering and signal amplification processing;

[0021] The power management module is used to supply power to the vibration sensor, signal conditioning module, signal processing control module, wireless data transmission module, and GSM SMS alarm module;

[0022] The wireless data transmission module is used to transmit vibration signals;

[0023] The GSM SMS alarm module is used to send an alarm SMS when the vibration signal exceeds a set threshold;

[0024] The signal processing control module is used to perform VMD decomposition on each group of transformer mixed vibration signals to decompose multiple subspaces; adopt a mutual information calculation method to obtain independent subspaces of vibration signals of the transformer core and winding, adopt a statistical method to obtain the correlation of the vibration signals of the core and winding, and obtain the mutual information value of the core and winding signals in each subspace; normalize the mutual information value of the core and winding signals in each subspace, set a selection threshold of the mutual information value, reconstruct according to the subspace corresponding to the selection threshold, and obtain a reconstructed signal of the transformer mixed vibration signal; decompose each group of reconstructed signals using the FASTICA algorithm to obtain a separation matrix W, separate the vibration signals of the core and winding, output the vibration signal with a higher high-order harmonic content as the core vibration signal, and output the vibration signal with a lower high-order harmonic content as the winding vibration signal.

[0025] As a preferred technical solution, when VMD decomposition is performed on each group of transformer mixed vibration signals, the number of modal components is set, and each group of transformer mixed vibration signals is decomposed into multiple subspaces according to the number of modal components, and the subspace bandwidth is set based on the normal signal spectrum range of the transformer.

[0026] As a preferred technical solution, the mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, and the mutual information value of the core and winding signals in each subspace is obtained. The specific calculation formula is expressed as follows:

[0027]

[0028] Among them, I(X;Y) represents the mutual information value of the core and winding signals in each subspace, X is the transformer core source signal, Y is the transformer winding source signal, p(x,y) is the joint probability density function of the two, p(x) and p(y) are the marginal probability density functions of the transformer core source signal and the transformer winding source signal, respectively.

[0029] As a preferred technical solution, the power management module is provided with a solar panel, a power management chip CN3065, a capacitor C1, a capacitor C2, a light-emitting diode D1, a light-emitting diode D2, a diode D3, a resistor R1, a resistor R2, a resistor R3, a resistor R4 and a lithium battery;

[0030] One end of the solar panel output is grounded, and the other end is connected to one end of capacitor C1, and the other end of capacitor C1 is grounded;

[0031] One end of the capacitor C1 is connected to one end of the resistor R1, the other end of the resistor R1 is connected to the anodes of the light-emitting diode D1 and the light-emitting diode D2 respectively, the cathode of the light-emitting diode D1 is connected to the CHARGE pin of the power management chip CN3065, and the cathode of the light-emitting diode D2 is connected to the DONE pin of the power management chip CN3065;

[0032] The other end of the solar panel output is connected to the VIN pin of the power management chip CN3065;

[0033] The ISET pin of the power management chip CN3065 is connected to the resistor R2.

[0034] The BAT pin of the power management chip CN3065 is connected to the positive electrode of the lithium battery, the negative electrode of the lithium battery is grounded, and the capacitor C2 is connected in parallel at both ends of the lithium battery;

[0035] The TEMP pin of the power management chip CN3065 is connected to the thermistor, and the TEMP pin is connected to the VIN pin through the resistor R3;

[0036] The FB pin of the power management chip CN3065 is connected to the positive electrode of the lithium battery to monitor the positive electrode voltage of the lithium battery;

[0037] The FB pin of the power management chip CN3065 is also connected to one end of the resistor R4 to output the charging voltage. The other end of the resistor R4 is connected to the cathode of the diode D3. The anode of the diode D3 is grounded. The resistor R4 and the diode D3 form a voltage stabilizing circuit.

[0038] As a preferred technical solution, the signal conditioning module is provided with an operational amplifier OPA2277, which is used to perform voltage division processing on the output signal and then output it to the signal processing control module.

[0039] As a preferred technical solution, the wireless data transmission module adopts the LoRa network and uses the 470MHz frequency band for communication.

[0040] As a preferred technical solution, the signal processing control module is provided with a TMS320F2812 chip, and the GSM SMS alarm module is provided with a TC35I chip. The 24th to 29th pins of the TC35I chip are interconnected with the corresponding pins of the SIM card, the 9th pin TXD and the 10th pin RXD are respectively connected to the RXD pin and TXD pin of the TMS320F2812, and are also connected to the Vcc pin and GND pin of the TC35I chip.

[0041] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0042] (1) The present invention can find the independent parts between the core and winding signals in each group of transformer mixed vibration signals and then separate the signals by using the independent component separation method. The VMD method is used to separate the subspace, which has good adaptability, can better retain the information components of the original signal, has the ability to process non-stationary signals, and can more effectively decompose the vibration signal.

[0043] (2) The present invention adopts solar energy for power supply and realizes intelligent charging and discharging through a power management chip, which has low power consumption and prolongs the battery life; the LoRa network is used for wireless data transmission, which has extremely low power consumption and long transmission distance, and the sensor can be installed flexibly without wiring; the DSP processor directly outputs the vibration signal of the transformer core and winding after separating the signal, making subsequent analysis more convenient, and the fault position can be quickly located by directly extracting the characteristics of the signal of the transformer core and winding. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A schematic flow chart of a control method for distinguishing transformer winding and core vibration signals according to the present invention;

[0045] Figure 2 A schematic diagram of a transformer mixed vibration signal according to the present invention;

[0046] Figure 3 It is a schematic diagram of the separated iron core vibration signal of the present invention;

[0047] Figure 4 A schematic diagram of a winding vibration signal separated by the present invention;

[0048] Figure 5 A schematic diagram of the framework of a control system for distinguishing transformer winding and core vibration signals according to the present invention;

[0049] Figure 6 A schematic diagram of the working process of the power management module of the present invention;

[0050] Figure 7 Schematic diagram of the circuit structure of the power management module of the present invention;

[0051] Figure 8 Schematic diagram of the circuit structure of the signal conditioning module of the present invention;

[0052] Fig. 9 Schematic diagram of the circuit structure of the signal processing control module of the present invention;

[0053] Fig.10 The figure is a schematic diagram of the circuit structure of the wireless data transmission module of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] Example 1

[0056] like Figure 1 As shown, this embodiment provides a control method for distinguishing transformer winding and core vibration signals, comprising the following steps:

[0057] S1: Collect more than two groups of transformer mixed vibration signals, and regard each group of mixed vibration signals Y(t) collected as multiple subspaces Y i The set of (t): The purpose is to find the independent parts between the core and winding signals in each group of mixed signals and then separate the signals by using the method of independent component separation.

[0058] S2: In order to find the independent parts between the core and winding signals in each group of mixed signals, the VMD method is used to separate the subspace Y i (t). Since VMD is sensitive to noise, it is necessary to perform denoising and filtering on the mixed vibration signals collected from each group to obtain a relatively pure transformer mixed vibration signal. When VMD is performed on each group of transformer mixed vibration signals, the number of modal components K is set to 7 (i.e., the original transformer mixed vibration signal is decomposed into 7 subspaces Y i (t), where i = 1, 2, 3, 4, 5, 6, 7). Since the normal signal spectrum of the transformer is within 2000, the subspace bandwidth is taken as 2000.

[0059] In this embodiment, the VMD method is used to separate the subspace, which has good adaptability, can better retain the information components of the original signal, has the ability to process non-stationary signals, and can more effectively decompose the vibration signal.

[0060] S3: The independent subspaces of the vibration signals of the core and winding are found by calculating the mutual information method. The correlation between the two random variables, the transformer winding vibration signal and the transformer core vibration signal, is found by statistical methods. The calculation formula is:

[0061]

[0062] Among them, X is the transformer core source signal, Y is the transformer winding source signal, p(x, y) is the joint probability density function of the two, p(x) and p(y) are the marginal probability density functions of the transformer core source signal and the transformer winding source signal respectively; by judging the value of the mutual information I(X; Y) of the core winding signal of each subspace after separation, it can be judged whether the subspace is an independent subspace. The smaller the mutual information value, the more independent the subspace.

[0063] S4: Normalize the calculated I(X;Y) and find 4 subspaces with large mutual interest values ​​for reconstruction. The reconstructed signal is a transformer mixed vibration signal mixed with completely independent transformer core winding vibration signals. The purpose of normalization is to prevent being affected by larger or smaller calculation results. The independent subspace reconstruction adopts the VMD reconstruction method.

[0064] S5: Decompose each group of reconstructed signals using the FASTICA algorithm to obtain a separation matrix W.

[0065] S6: The expression of the collected pure transformer mixed signal is as follows:

[0066] Y(t)=AS(t)

[0067] Where Y(t) is the collected transformer mixed signal, S(t) is the source signal (core and winding vibration signal), and A is the mixing matrix. The above formula regards the mixing of the transformer core and winding vibration signals as a linear mixing process. This embodiment is also based on linear mixing, by comparing multiple groups of output mixed signals Y(t) (two groups are taken in this embodiment), to find a separation matrix W, the function of this separation matrix is ​​to estimate the original source signal S(t); the result obtained by this method through the separation matrix is ​​S'(t), and the specific solution formula is as follows:

[0068] S'(t)=WY(t)=WAS(t)

[0069] Where S'(t) is the separated source signal, and W is the separation matrix. In order to minimize the error between the separation (estimation) result and the source signal S(t), when WA=1, S'(t)=S(t), that is, the separated signal is the vibration source signal of the transformer core and winding.

[0070] The separation matrix W is known from step S5, and the collected mixed signal is multiplied by the separation matrix W on the left to obtain the vibration signals of the core and the winding.

[0071] The above steps can only separate two groups of signals, namely the core and winding signals. However, the order of the signals after the two groups of signals are separated is random, and it is not possible to intelligently determine which group is the core vibration signal and which group is the winding vibration signal. The following provides a method for determining which signal the separated signal is.

[0072] The vibration signals of the core and the winding are very similar, both are periodic signals, and both contain a 100HZ fundamental frequency, but the core vibration signal contains high-order harmonic components, while the winding signal has no high-order harmonics under ideal conditions. However, due to the nonlinear characteristics of the strain force of the winding insulation pad, the winding signal may also contain high-order harmonics, but the high-order harmonic content of the core vibration signal is greater than that of the winding, so the vibration signal with a higher high-order harmonic content is output as the core vibration signal, and the vibration signal with a lower high-order harmonic content is output as the winding vibration signal. In this way, it can be determined whether the separated signal corresponds to the winding vibration signal or the core vibration signal.

[0073] The existing ICA algorithm is mainly used to separate two independent signals, but the transformer winding and the iron core are connected together, and the vibration signals of the two will affect and alias each other, so the two signals are not completely independent, and the existing ICA algorithm cannot be used to separate them well. This application improves the ICA algorithm, finds the independent parts between the two signals, and then separates the signals by using the independent component separation method, thereby achieving the separation of non-completely independent signals.

[0074] like Figure 2-Figure 4 As shown, the separation results are obtained, where Figure 2 Mixed signal for transformer, Figure 3 is the separated core signal, Figure 4 is the separated winding signal.

[0075] Example 2

[0076] like Figure 5 As shown, this embodiment provides a control system for distinguishing transformer winding and core vibration signals, including: a vibration sensor, a signal conditioning module, a signal processing control module, a power management module, a wireless data transmission module and a GSM SMS alarm module;

[0077] In this embodiment, the vibration sensor adopts a piezoelectric acceleration sensor CA-YD-168, which can detect a frequency of up to 10kHz and is suitable for collecting transformer vibration signals;

[0078] In this embodiment, the signal conditioning module is used to perform signal preprocessing (signal amplification), and the output vibration signal can be recognized by the DSP;

[0079] In this embodiment, the signal processing control module adopts the TMS320F2812 model DSP processor, which is used to intelligently control and manage all functions of the sensor system, and intelligently control the sensor working mode: divided into low power consumption mode and normal working mode, digital filtering, control of sending alarm signals and other intelligent functions, and adopts the control method of distinguishing transformer winding and core vibration signals in the above-mentioned embodiment 1 to separate the original collected signals, and directly output the transformer winding vibration signal and the core vibration signal.

[0080] In this embodiment, the power management module supplies power to the vibration sensor, the signal conditioning module, the signal processing control module, the wireless data transmission module, and the GSM SMS alarm module;

[0081] like Figure 6 As shown, the solar panel converts solar energy into electrical energy, which is then used to charge the lithium battery through the power management chip CN3065, and then converted into different working voltages through the DC / DC converter to power each module (note that constant current power supply is required to power the sensor, and this embodiment uses the LT3092 chip to make it constant current).

[0082] like Figure 7 As shown, capacitor C1 is a filter capacitor to ensure the stability of the circuit; two light-emitting diodes D1 and D2 are connected to the drain indicator end of the triode inside the chip (i.e., the chip DONE and CHARGE pins) to indicate the working state of the circuit; resistor R2 prevents excessive current; the BAT pin of the chip CN3065 is connected to the lithium battery to charge the battery; the TEMP pin of the chip CN3065 is connected to the thermistor to monitor the temperature. If the temperature is too high, the chip CN3065 will stop charging. The principle uses the function of the TEMP pin. When the voltage of the TEMP pin is less than 46% of the input voltage, the charging will stop; the FB pin of the chip CN3065 is used to monitor the positive voltage of the lithium battery. The charging circuit realizes intelligent charging through the voltage difference between the chip VIN pin and the BAT pin. If the voltage difference is less than 20mV, the chip will stop charging and enter the dormant state; resistor R4 and diode D3 form a voltage stabilizing circuit to keep the output voltage of the lithium battery stable.

[0083] In this embodiment, a 14500 lithium battery is provided in the power management module, and the output voltage is 4.2V, which can supply power to each module after passing through a DC / DC converter, and has low power consumption.

[0084] like Figure 8 As shown, in the circuit of the signal conditioning module, the output signal is divided by the operational amplifier OPA2277 to ensure that the output signal is less than or equal to 3V and can be directly connected to the DSP processor;

[0085] like Fig. 9As shown, the signal processing control module adopts a C-type DSP processor, and its function is mainly to realize vibration detection processing through software programming. The working power supply voltage range is 1.8V to 3.6V, and the preferred embodiment is 3.3V. After power-on, the analog signal is input into the chip, and the AD conversion module of TMS320F2812 starts to work to convert it into a data signal for processing. The data processing function specifically includes data conversion, algorithm filtering, and vibration signal extraction. The DSP processor can also intelligently control the sensor working mode: it is divided into low power consumption mode and normal working mode. At the same time, it has intelligent functions such as information processing, mixed signal separation, digital filtering, and control of sending alarm signals (in low power consumption mode, the device can only perform data collection and preliminary threshold alarm functions, and the normal working mode includes all the above functions). The control circuit uses an SP708 reset chip for reset; the XOUT pin and the XIN pin are connected to a crystal oscillator circuit, which provides a basic clock signal for the system. TMS320F2812 communicates serially with the wireless data transmission module through four SPI ports. TMS320F2812 communicates serially with the GSM module through two SCI ports. DSP processor has powerful processing capability and extremely low power consumption. DSP processor must use JTAG interface circuit to realize program burning and simulation debugging. DSP processor can output relatively pure vibration signal.

[0086] like Fig.10 As shown, the wireless data transmission module adopts the LoRa network for the transmission of vibration data; the LoRa network has extremely low power consumption and long transmission distance compared with other wireless networks. This embodiment uses the 470MHz frequency band for communication, and the selected chip is SX1278. The sleep current of the SX1278 chip is less than 0.2uA, the current is 1.6mA when idle, and 9.9mA and 120mA when receiving and sending respectively, with lower power consumption. The power management module provides 3.3V voltage for the three power pins of the circuit, and the RF chip starts working. The received data is amplified, converted into radio signals, and then transmitted to the host computer through the RF switch through the antenna. In the circuit diagram, the power supply parallel filter capacitor ensures the stability of the power supply, and the capacitors C20, C22 and the inductor form a parallel resonant circuit for signal selection; the transmitting circuit and the receiving circuit both form an antenna matching circuit through capacitors and inductors, and reduce interference through filter capacitors; the RF chip adopts the PE4259 chip, which distinguishes the transmission and reception of the wireless circuit, improves the signal quality and reduces interference. The wireless data transmission module has the advantages of long transmission distance, low power consumption, strong anti-interference ability and high sensitivity, and can transmit transformer vibration signals well.

[0087] In this embodiment, the GSM SMS alarm module is used to send an alarm SMS to a mobile phone when the collected vibration signal exceeds a set threshold.

[0088] The GSM SMS alarm module uses TC35I. The 24th to 29th pins of TC35I are connected to the corresponding pins of the SIM card. The 9th pin TXD and 10th pin RXD of the GSM SMS alarm module are connected to the RXD pin and TXD pin of TMS320F2812 respectively, and the Vcc pin and GND pin of the TC35I chip are connected at the same time, so that the communication between the GSM SMS alarm module and the microcontroller can be completed. When the collected vibration signal exceeds the set threshold, the microcontroller drives the GSM SMS alarm module, and the GSM SMS alarm module sends an alarm SMS to the mobile phone.

[0089] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A control method for distinguishing transformer winding and core vibration signals, characterized in that: The steps include: Collect multiple groups of transformer mixed vibration signals, and perform signal preprocessing on each group of transformer mixed vibration signals collected, including denoising filtering and signal amplification processing; Perform VMD decomposition on the mixed vibration signals of each group of transformers to decompose multiple subspaces; When VMD decomposition is performed on each group of transformer mixed vibration signals, the number of modal components is set, and each group of transformer mixed vibration signals is decomposed into multiple subspaces according to the number of modal components, and the subspace bandwidth is set based on the normal signal spectrum range of the transformer; The mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, and the mutual information value of the core and winding signals in each subspace is obtained; The mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, so as to obtain the mutual information value of the core and winding signals in each subspace. The specific calculation formula is expressed as follows: Among them, I(X;Y) represents the mutual information value of the core and winding signals in each subspace, X is the transformer core source signal, Y is the transformer winding source signal, p(x,y) is the joint probability density function of the two, p(x) and p(y) are the marginal probability density functions of the transformer core source signal and the transformer winding source signal respectively; Normalize the mutual information values ​​of the core and winding signals in each subspace, set a threshold for selecting the mutual information value, reconstruct the subspace corresponding to the selected threshold, and obtain a reconstructed signal of the transformer mixed vibration signal; The FASTICA algorithm is used to decompose each group of reconstructed signals, and the separation matrix W is obtained to separate the vibration signals of the core and the winding. The vibration signals with higher harmonic content are output as core vibration signals, and the vibration signals with lower harmonic content are output as winding vibration signals.

2. A control system for distinguishing transformer winding and core vibration signals, characterized in that: include: Vibration sensor, signal conditioning module, signal processing control module, power management module, wireless data transmission module and GSM SMS alarm module; The vibration sensor is connected to the signal conditioning module, the signal conditioning module is connected to the signal processing control module, the signal processing control module is respectively connected to the wireless data transmission module and the GSM short message alarm module, and the power management module is respectively connected to the vibration sensor, the signal conditioning module, the signal processing control module, the wireless data transmission module, and the GSM short message alarm module; The vibration sensor is used to collect multiple groups of transformer mixed vibration signals; The signal conditioning module is used to perform signal preprocessing on the collected mixed vibration signals of each group of transformers, including denoising filtering and signal amplification processing; The power management module is used to supply power to the vibration sensor, signal conditioning module, signal processing control module, wireless data transmission module, and GSM SMS alarm module; The wireless data transmission module is used to transmit vibration signals; The GSM SMS alarm module is used to send an alarm SMS when the vibration signal exceeds a set threshold; The signal processing control module is used to perform VMD decomposition on each group of transformer mixed vibration signals to decompose multiple subspaces; adopt a mutual information calculation method to obtain independent subspaces of vibration signals of the transformer core and winding, adopt a statistical method to obtain the correlation of the vibration signals of the core and winding, and obtain the mutual information value of the core and winding signals in each subspace; normalize the mutual information value of the core and winding signals in each subspace, set a selection threshold of the mutual information value, reconstruct according to the subspace corresponding to the selection threshold, and obtain a reconstructed signal of the transformer mixed vibration signal; perform FASTICA algorithm decomposition on each group of reconstructed signals, obtain a separation matrix W, separate the vibration signals of the core and winding, output the vibration signal with a higher harmonic content as a core vibration signal, and output the vibration signal with a lower harmonic content as a winding vibration signal; When VMD decomposition is performed on each group of transformer mixed vibration signals, the number of modal components is set, and each group of transformer mixed vibration signals is decomposed into multiple subspaces according to the number of modal components, and the subspace bandwidth is set based on the normal signal spectrum range of the transformer; The mutual information calculation method is used to obtain the independent subspaces of the vibration signals of the transformer core and winding, and the correlation of the vibration signals of the core and winding is obtained by a statistical method, so as to obtain the mutual information value of the core and winding signals in each subspace. The specific calculation formula is expressed as follows: Among them, I(X;Y) represents the mutual information value of the core and winding signals in each subspace, X is the transformer core source signal, Y is the transformer winding source signal, p(x,y) is the joint probability density function of the two, p(x) and p(y) are the marginal probability density functions of the transformer core source signal and the transformer winding source signal, respectively.

3. The control system for distinguishing transformer winding and core vibration signals according to claim 2, characterized in that: The power management module is provided with a solar panel, a power management chip CN3065, a capacitor C1, a capacitor C2, a light emitting diode D1, a light emitting diode D2, a diode D3, a resistor R1, a resistor R2, a resistor R3, a resistor R4 and a lithium battery; One end of the solar panel output is grounded, and the other end is connected to one end of capacitor C1, and the other end of capacitor C1 is grounded; One end of the capacitor C1 is connected to one end of the resistor R1, the other end of the resistor R1 is connected to the anodes of the light-emitting diode D1 and the light-emitting diode D2 respectively, the cathode of the light-emitting diode D1 is connected to the CHARGE pin of the power management chip CN3065, and the cathode of the light-emitting diode D2 is connected to the DONE pin of the power management chip CN3065; The other end of the solar panel output is connected to the VIN pin of the power management chip CN3065; The ISET pin of the power management chip CN3065 is connected to the resistor R2. The BAT pin of the power management chip CN3065 is connected to the positive electrode of the lithium battery, the negative electrode of the lithium battery is grounded, and the capacitor C2 is connected in parallel at both ends of the lithium battery; The TEMP pin of the power management chip CN3065 is connected to the thermistor, and the TEMP pin is connected to the VIN pin through the resistor R3; The FB pin of the power management chip CN3065 is connected to the positive electrode of the lithium battery to monitor the positive electrode voltage of the lithium battery; The FB pin of the power management chip CN3065 is also connected to one end of the resistor R4 to output the charging voltage. The other end of the resistor R4 is connected to the cathode of the diode D3. The anode of the diode D3 is grounded. The resistor R4 and the diode D3 form a voltage stabilizing circuit.

4. The control system for distinguishing transformer winding and core vibration signals according to claim 2, characterized in that: The signal conditioning module is provided with an operational amplifier OPA2277, which is used to perform voltage division processing on the output signal and then output it to the signal processing control module.

5. The control system for distinguishing transformer winding and core vibration signals according to claim 2, characterized in that: The wireless data transmission module adopts the LoRa network and uses the 470MHz frequency band for communication.

6. The control system for distinguishing transformer winding and core vibration signals according to claim 2, characterized in that: The signal processing control module is provided with a TMS320F2812 chip, and the GSM SMS alarm module is provided with a TC35I chip. The 24th to 29th pins of the TC35I chip are interconnected with the corresponding pins of the SIM card, the 9th pin TXD and the 10th pin RXD are respectively connected to the RXD pin and TXD pin of the TMS320F2812, and are also connected to the Vcc pin and the GND pin of the TC35I chip.

Images