Converter station equipment fault monitoring method and system based on robot acquisition data

Abstract

The invention relates to the technical field of converter station power equipment detection, in particular to a converter station equipment fault monitoring method and system based on robot acquisition data, and the method comprises the steps: obtaining a to-be-detected sound signal near a detected device through a movable robot; processing the to-be-detected sound signal to obtain a frequency component and a frequency center value of the to-be-detected sound signal; obtaining a to-be-detected sound signal energy value of the to-be-detected sound signal according to the frequency component and the frequency center value of the to-be-detected sound signal; and according to the to-be-detected sound signal energy value of the to-be-detected sound signal and the audio energy value of the normal sound signal, carrying out fault monitoring on the detected equipment. According to the method, the sound signals are divided and processed carefully, the fault monitoring accuracy of the detected equipment is effectively improved, in addition, the sound signals are obtained through the mobile robot, the intelligent level is improved, the labor cost is reduced, accurate collection can be conducted, and the fault monitoring accuracy is further improved.

Description

Technical Field

[0001] The present application relates to the technical field of converter station power equipment detection, and in particular to a converter station equipment fault monitoring method and system based on robot data collection. Background Art

[0002] The power equipment in the converter station is huge, and the equipment, represented by motors, works continuously for a long time and is prone to failure. If it is not maintained in time after a failure, it will lead to equipment failure and serious consequences. The traditional way to find faults in power equipment is to use manual inspections, in which staff make judgments based on their work experience and the operating conditions of the motors. This method is not very accurate, relies heavily on manual experience, and lacks intelligent means.

[0003] In the prior art, Chinese patent CN117309065B discloses a remote monitoring method and system for converter stations based on drones, which collects environmental parameter data for analysis through sensors carried by drones, reducing labor costs and improving work efficiency. However, the sound and other data collected by drones will be interfered by factors such as noise from the drone itself, the surrounding environment, and the distance, which will affect the accuracy of the collected data and the accuracy of fault detection.

[0004] Therefore, a new solution is urgently needed to improve the accuracy of fault diagnosis of power equipment in converter stations. Summary of the invention

[0005] In view of this, it is necessary to provide a converter station equipment fault monitoring method based on robot data collection to solve the problems in the prior art.

[0006] A converter station equipment fault monitoring method based on robot data collection, comprising:

[0007] Acquiring the sound signal to be detected near the device under test by using a mobile robot;

[0008] Processing the sound signal to be detected to obtain the frequency components and the frequency center value of the sound signal to be detected; obtaining the sound signal energy value to be detected of the sound signal to be detected according to the frequency components and the frequency center value of the sound signal to be detected;

[0009] Fault monitoring is performed on the device under test according to the sound signal energy value to be detected of the sound signal to be detected and the audio energy value of the normal sound signal.

[0010] Preferably, performing fault monitoring on the device under test according to the energy value of the sound signal to be detected of the sound signal to be detected and the energy value of the sound signal to be detected of the normal sound signal comprises:

[0011] Determine a difference between the energy value of the sound signal to be detected and the audio energy value of the normal sound signal, satisfying:

[0012] T=E normal +(E fault -E normal ) / 3

[0013] Wherein, T is the difference value, E normal is the audio energy value of the normal sound signal, E fault is the sound signal energy value to be detected of the sound signal to be detected;

[0014] If the energy value of the sound signal to be detected of the sound signal to be detected is greater than the difference value, it is determined that the device under test has a fault; if the energy value of the sound signal to be detected of the sound signal to be detected is not greater than the difference value, it is determined that the device under test is normal.

[0015] Preferably, processing the sound signal to be detected to obtain the frequency component and the frequency center value of the sound signal to be detected includes:

[0016] Processing the sound signal to be detected by Fourier transform to obtain the frequency component of the sound signal to be detected;

[0017] Calculate a frequency component difference sequence between the sound signal to be detected and the normal sound signal; and determine a frequency center value of the sound signal to be detected based on the frequency component difference sequence.

[0018] Preferably, the energy value of the sound signal to be detected of the sound signal to be detected is obtained according to the frequency components and the frequency center value of the sound signal to be detected, which satisfies:

[0019]

[0020] Wherein, E is the energy value of the sound signal to be detected, F(k) is the frequency component of the sound signal to be detected, and k center is the frequency center value of the sound signal to be detected, k w is the frequency tone coverage width of the sound signal to be detected, and i is the sampling sequence number of the sound signal to be detected.

[0021] Preferably, the frequency tone coverage width of the sound signal to be detected satisfies:

[0022] k w =k center / 2

[0023] Wherein, kw is the frequency coverage width of the sound signal to be detected, k centeris the frequency center value of the sound signal to be detected.

[0024] Preferably, processing the sound signal to be detected by Fourier transform further includes:

[0025] Performing normalization processing on the sound signal to be detected;

[0026] The normalized sound signal to be detected is pre-emphasized and filtered to meet the following conditions:

[0027] y i =x i -c1x i-1 ,i=1,2,3,...,N

[0028] Among them, i is the sampling sequence number, x i is the output sequence after normalization, y i is the output sequence of the pre-emphasis filter; c1 is the pre-emphasis filter coefficient, 0<c1<1.

[0029] Preferably, the method further comprises performing spectral subtraction filtering on the output sequence obtained by pre-emphasis filtering:

[0030] The output sequence obtained by the pre-emphasis filtering is processed by Fourier transform to obtain the frequency components of the sound signal to be detected, which satisfies:

[0031]

[0032] Among them, y i is the output sequence of the pre-emphasis filtering, k is the subscript of the frequency domain component;

[0033] Spectral subtraction filtering is performed based on the average energy of the normal sound signal and the frequency component of the sound signal to be detected, satisfying:

[0034]

[0035] Among them, a and b are two constants;

[0036] D(k) is the average energy of the normal sound signal, satisfying:

[0037]

[0038] Among them, the duration of the normal sound signal is IS, and the corresponding number of frames is NIS.

[0039] Preferably, obtaining the sound signal energy value to be detected of the sound signal to be detected further includes:

[0040] Processing the sound signal to be detected to obtain multiple samples;

[0041] Calculating the average value of the energy values ​​of the sound signal to be detected of the plurality of samples;

[0042] The average value is used as the sound signal energy value to be detected of the sound signal to be detected.

[0043] Preferably, processing the sound signal to be detected includes windowing operation and frame shifting operation to obtain multiple samples.

[0044] Compared with the prior art, the beneficial effect of the present application is that the accuracy of fault monitoring of the equipment under test is effectively improved by the detailed division and processing of the sound signal. In addition, by acquiring the sound signal through the mobile robot, the intelligence level is improved and the labor cost is reduced, and the sound signal near the equipment under test can be accurately collected, thereby avoiding the interference of some noises and further improving the accuracy of fault monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of a converter station equipment fault monitoring method based on robot data collection provided in an embodiment of the present application.

[0046] Figure 2 This is an example diagram of a recording signal waveform provided in an embodiment of the present application.

[0047] Figure 3 This is an example diagram of high-frequency fault sound energy values ​​and discrimination of samples at different times provided in an embodiment of the present application.

[0048] Figure 4 It is a structural schematic diagram of a converter station equipment fault monitoring system based on robot data collection provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Please refer to Figure 1-Figure 3 , Figure 1 It is a flow chart of a converter station equipment fault monitoring method based on robot data collection provided in an embodiment of the present application; Figure 2 is an example diagram of a recording signal waveform provided in an embodiment of the present application; Figure 3 The energy value and discrimination example diagram of high-frequency fault sound of samples at different times provided in the embodiment of the present application. The converter station equipment fault monitoring method based on robot data collection includes the following steps:

[0051] S101: Acquire a sound signal to be detected near the device under test by using a mobile robot.

[0052] A mobile robot equipped with a sound signal acquisition device is used to record the device under test. The robot can be deployed near the motor under test and can effectively collect the sound signals emitted by the motor. To effectively collect the high-frequency sound signals to be detected from the motor, the acquisition device should have a high sampling frequency. The mobile robot can move along a preset trajectory or be controlled by wired or remote control. In addition, the sound signal acquisition device carried has a high sampling frequency to collect sound signals from a wider range as much as possible.

[0053] S102: Process the sound signal to be detected to obtain the frequency components and the frequency center value of the sound signal to be detected.

[0054] The sound signal to be detected is collected and can be recorded as x i =x1,x2,x3,....,x N , i=1,2,3,...,N, where i is the sampling sequence number and N is the length of the sequence. The length of each sequence should remain consistent during implementation. For example, the duration of each sample is 0.4s. In the case of a 16Khz sampling frequency, N=16000*0.4=6400.

[0055] like Figure 2 The following is an example of the recording signal waveform, where the waveforms before 6:02 are the recordings of a normal motor, and the waveforms after that are the recordings of the faulty motor at different angles and distances.

[0056] Preferably, before the sound signal to be detected is processed to obtain the frequency components and the frequency center value of the sound signal to be detected, preprocessing may be performed, including normalization processing and pre-emphasis filtering processing. For example, normalization processing:

[0057] During the sound monitoring process, factors such as the distance and orientation of the acquisition equipment may cause differences in the decibel value of the collected sound signal. In order to remove this influence, the signal should be normalized and scaled to the range of -1 to 1 based on the maximum positive and negative values. The normalized output satisfies:

[0058] Among them, i is the sampling sequence number, x i is the collected sound signal sequence to be detected, y i Output sequence for normalization.

[0059] Pre-emphasis filtering:

[0060] Since there are many environmental noise interferences on site, such as cooling fan noise, and the specific sound signal to be detected is mainly in high-frequency form, the signal is filtered. The pre-emphasis filtering process meets the following requirements:

[0061] y i =x i -c1x i-1 ,i=1,2,3,...,N

[0062] Among them, i is the sampling sequence number, x i is the output sequence after normalization, y i is the output sequence of the pre-emphasis filter; c1 is the pre-emphasis filter coefficient, 0<c1<1.

[0063] Furthermore, the output sequence obtained by pre-emphasis filtering can also be subjected to spectral subtraction filtering. The goal of spectral subtraction filtering is to subtract the influence of environmental noise from the signal, thereby highlighting the sound signal to be detected. The signal spectrum is obtained by performing a fast Fourier transform (FFT) on the signal, and the environmental noise is also subjected to an FFT transform to obtain its spectrum, and then the noise component in the signal spectrum is removed by spectral subtraction. The purpose of this is to enhance the visibility of the sound signal to be detected relative to the background noise and improve the accuracy of fault detection.

[0064] For example, the output sequence obtained by the pre-emphasis filtering is processed by Fourier transform to obtain the frequency component of the sound signal to be detected, which satisfies:

[0065]

[0066] Among them, y i is the output sequence of the pre-emphasis filtering, k is the subscript of the frequency domain component;

[0067] Spectral subtraction filtering is performed based on the average energy of the normal sound signal and the frequency component of the sound signal to be detected, satisfying:

[0068]

[0069] Among them, a and b are two constants, a is called the over-reduction factor; b is called the gain compensation factor.

[0070] D(k) is the average energy of a normal sound signal, satisfying:

[0071]

[0072] The duration of the normal sound signal is IS, and the corresponding number of frames is NIS. The normal sound signal can be a long-term environmental noise recording on site that does not contain the sound signal to be detected, such as the noise of the cooling fan, the noise during normal operation of the equipment, etc.

[0073] Calculation of frequency center value: Based on the normalization and filtering transformation in the above steps, the frequency component sequence of the normal sound signal is obtained and recorded as F(k) normal , the frequency component sequence of the sound signal to be detected is recorded as F(k) fault , the difference sequence between the two is recorded as: F(k) diff =F(k) fault -F(k) normal , k=0,1,2,...,N / 2. Based on the properties of Fourier transform, k here can only be N / 2 and is an integer.

[0074] Take the frequency component subscript k corresponding to the maximum value of the sequence, which is the frequency center of the sound signal to be detected, and the frequency sound coverage width is usually taken. Among them, the frequency domain component subscript is used to identify each frequency component in the frequency domain signal obtained after FFT transformation. These frequency components represent the energy distribution of the original signal at different frequencies. These frequency components are usually represented by a sequence or vector, and each element in the sequence will have an index value (i.e., subscript) to identify a specific frequency component. In applications, the frequency center of the sound signal to be detected and the frequency sound coverage width can also be determined by discrete Fourier transform of the sample signal and then by manual analysis methods, making it more accurate and targeted.

[0075] S103: Obtaining a sound signal energy value of the sound signal to be detected according to the frequency components and the frequency center value of the sound signal to be detected.

[0076] According to the frequency component F(k), the frequency center k of the sound signal to be detected is determined. center , and its frequency tone coverage width k w , calculate the high-frequency fault sound energy indicator index of the motor fault in the sound signal to be detected, the formula is:

[0077]

[0078] This formula can effectively highlight the energy of the specified frequency component and calculate its out-of-band radiated sound signal to comprehensively obtain the fault index.

[0079] S104: performing fault monitoring on the device under test according to the energy value of the sound signal to be detected and the audio energy value of the normal sound signal.

[0080] like Figure 3 As shown, Figure 3 This is an example diagram of high-frequency fault sound energy values ​​and discrimination of samples at different times provided in an embodiment of the present application.

[0081] The frequency component F(k) of the normal sound signal obtained based on steps S102-S103 normal 、The frequency component F(k) of the sound signal to be detected fault And the frequency center k of the sound signal to be detected center , the frequency coverage width k of the sound signal to be detected w , the energy value of the high-frequency sound signal to be detected of the normal audio sample sound signal and the energy value of the high-frequency sound signal to be detected of the sound signal to be detected can be obtained.

[0082] Determine a difference between the energy value of the sound signal to be detected and the audio energy value of the normal sound signal, satisfying:

[0083] T=E normal +(E fault -E normal ) / 3, wherein T is the difference value, is the audio sound energy value of the normal sound signal, and is the energy value of the sound signal to be detected of the sound signal to be detected. According to experience and a large number of experiments, when the threshold value coefficient in this formula is 3, the difference value can achieve a relatively ideal result. However, in some implementations, the threshold value coefficient can be adjusted according to time conditions, experience, etc.

[0084] If the energy value of the sound signal to be detected of the sound signal to be detected is greater than the difference value, it is determined that the device under test fails; if the energy value of the sound signal to be detected of the sound signal to be detected is not greater than the difference value, it is determined that the device under test is normal.

[0085] That is, the monitoring result Sn satisfies:

[0086] This can be used to determine whether motors and other equipment are faulty during actual monitoring.

[0087] Based on an inventive concept, the present application also provides a converter station equipment fault monitoring system based on robot data collection, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the structure of the converter station equipment fault monitoring system based on robot data collection provided in the embodiment of the present application. The system includes a mobile robot and a computing device; the mobile robot is equipped with a sound signal collection device for collecting sound signals to be detected near the device under test; the computing device is used to perform the above steps S101-S104 and any optional implementation method thereof according to the sound signals to be detected collected by the mobile robot, and perform fault monitoring on the device under test, which will not be repeated.

[0088] In summary, the solution provided by this application uses a flexible robot to automatically collect data. The sound monitoring method makes it easier to install and deploy the equipment without damaging the original structure of the motor. As a non-contact measurement, the monitoring system is not easily disturbed by high voltage. The calculation method of the energy value of high-frequency fault sound not only considers the energy fluctuation at its main frequency, but also takes into account the out-of-band radiation and the offset of the audio center, which improves the anti-interference ability and effectively improves the accuracy of fault monitoring.

Claims

1. A method for monitoring converter station equipment faults based on robot data collection, characterized in that: include: Acquiring the sound signal to be detected near the device under test by using a mobile robot; Processing the sound signal to be detected to obtain the frequency component and the frequency center value of the sound signal to be detected; Obtaining a sound signal energy value of the sound signal to be detected according to the frequency components and the frequency center value of the sound signal to be detected; Fault monitoring is performed on the device under test according to the sound signal energy value to be detected of the sound signal to be detected and the audio energy value of the normal sound signal.

2. The method according to claim 1, characterized in that Performing fault monitoring on the device under test according to the sound signal energy value to be detected of the sound signal to be detected and the sound signal energy value to be detected of the normal sound signal, including: Determine a difference between the energy value of the sound signal to be detected and the audio energy value of the normal sound signal, satisfying: T=E normal +(And fault -AND normal ) / 3 Wherein, T is the difference value, E normal is the audio energy value of the normal sound signal, E fault is the sound signal energy value to be detected of the sound signal to be detected; If the energy value of the sound signal to be detected of the sound signal to be detected is greater than the difference value, it is determined that the device under test has a fault; if the energy value of the sound signal to be detected of the sound signal to be detected is not greater than the difference value, it is determined that the device under test is normal.

3. The method according to claim 1, characterized in that Processing the sound signal to be detected to obtain the frequency component and the frequency center value of the sound signal to be detected includes: Processing the sound signal to be detected by Fourier transform to obtain the frequency component of the sound signal to be detected; Calculating a difference sequence between the frequency components of the sound signal to be detected and the frequency components of a normal sound signal; Based on the frequency component difference sequence, a frequency center value of the sound signal to be detected is determined.

4. The method according to claim 3, characterized in that According to the frequency components and the frequency center value of the sound signal to be detected, the energy value of the sound signal to be detected of the sound signal to be detected is obtained, which satisfies: Wherein, E is the energy value of the sound signal to be detected, F(k) is the frequency component of the sound signal to be detected, and k center is the frequency center value of the sound signal to be detected, k w is the frequency tone coverage width of the sound signal to be detected, and i is the sampling sequence number of the sound signal to be detected.

5. The method according to claim 4, characterized in that The frequency coverage width of the sound signal to be detected satisfies: k w =k center / 2 Wherein, kw is the frequency coverage width of the sound signal to be detected, k center is the frequency center value of the sound signal to be detected.

6. The method according to claim 3, characterized in that Processing the sound signal to be detected by Fourier transform, further comprising: Performing normalization processing on the sound signal to be detected; The normalized sound signal to be detected is pre-emphasized and filtered to meet the following conditions: y i =x i -c1x i-1 ,i=1,2,3,...,N Among them, i is the sampling sequence number, x i is the output sequence after normalization, y i is the output sequence of the pre-emphasis filter; c1 is the pre-emphasis filter coefficient, 0<c1<1.

7. The method according to claim 6, characterized in that It also includes spectral subtraction filtering of the output sequence obtained by pre-emphasis filtering: The output sequence obtained by the pre-emphasis filtering is processed by Fourier transform to obtain the frequency components of the sound signal to be detected, which satisfies: Among them, y i is the output sequence of the pre-emphasis filtering, k is the subscript of the frequency domain component; Spectral subtraction filtering is performed based on the average energy of the normal sound signal and the frequency component of the sound signal to be detected, satisfying: Among them, a and b are two constants; D(k) is the average energy of the normal sound signal, satisfying: Among them, the duration of the normal sound signal is IS, and the corresponding number of frames is NIS.

8. The method according to any one of claims 1 to 7, characterized in that: Obtaining the sound signal energy value to be detected of the sound signal to be detected further includes: Processing the sound signal to be detected to obtain multiple samples; Calculating the average value of the energy values ​​of the sound signal to be detected of the plurality of samples; The average value is used as the sound signal energy value to be detected of the sound signal to be detected.

9. The method according to claim 8, characterized in that The processing of the sound signal to be detected includes windowing operation and frame shifting operation to obtain multiple samples.

10. A converter station equipment fault monitoring system based on robot data collection, characterized in that: including mobile robots and computing devices; The mobile robot is equipped with a sound signal collecting device for collecting the sound signals to be detected near the device under test; The computing device is used to execute the method described in any one of claims 1 to 9 to perform fault monitoring on the device under test based on the sound signal to be detected collected by the mobile robot.

Images