OLTC mechanical fault diagnosis method and system based on multi-source information fusion

Abstract

The application relates to the technical field of fault diagnosis, and discloses an OLTC mechanical fault diagnosis method and system based on multi-source information fusion, which comprises the following steps: dividing an action interval in an OLTC mechanical action process according to time axis information in the OLTC mechanical action process, and generating an initial vibration time-frequency graph and an initial acoustic emission time-frequency graph of the OLTC mechanical action process through short-time Fourier transform; at the same time point, calculating a difference amount between energy distribution in the initial vibration time-frequency graph and the initial acoustic emission time-frequency graph, and enhancing a frequency band area with an energy change trend opposite to that of the initial acoustic emission time-frequency graph in the initial vibration time-frequency graph according to the difference amount to obtain a difference time-frequency graph of the OLTC mechanical action process; mapping a fault sensitive feature vector of the difference time-frequency graph to a preset reference feature vector library, and outputting a mechanical fault diagnosis result of the OLTC mechanical action process; and the application can improve the efficiency of OLTC mechanical fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to an OLTC mechanical fault diagnosis method and system based on multi-source information fusion.

[0002] During the switching process, multiple pairs of contacts inside an on-load tap changer (OLTC) move and separate sequentially under the drive of energy storage springs, generating multiple transient impact events arranged in sequence. The vibration signal reflects the mechanical structural response generated by the contact movement, while the acoustic emission signal reflects the release of transient elastic stress waves on the surface of the contact material caused by friction, microcracks, etc. The two have different physical generation mechanisms and frequency band characteristics. Background Technology

[0003] Existing on-load tap changer mechanical fault diagnosis technology relies solely on a single vibration signal to complete fault analysis, failing to incorporate acoustic emission signals into the diagnostic analysis scope. This makes it impossible to establish a collaborative diagnostic system that integrates multi-source information from vibration and acoustic emission. When performing time-frequency characteristic analysis on the entire mechanical operation process of on-load tap changers, it cannot combine the energy distribution patterns of vibration and acoustic emission signals for joint processing, thus failing to comprehensively capture key characteristic information throughout the entire mechanical operation cycle and making it difficult to effectively extract core characteristic data highly correlated with mechanical faults.

[0004] The existing on-load tap changer mechanical fault diagnosis method does not perform feature enhancement operations on frequency bands where the energy change trends of vibration signals and acoustic emission signals are opposite. This fails to highlight the differences in fault-sensitive features, resulting in insufficient distinguishability and recognition of fault-sensitive features. When mapping the extracted feature data to the reference feature library for fault determination, the accuracy of feature matching is low, the operation efficiency of the diagnosis process is poor, and it is impossible to stably output accurate and complete mechanical fault diagnosis conclusions, making it difficult to guarantee the reliability of fault diagnosis results. Summary of the Invention

[0005] This invention provides a method and system for OLTC mechanical fault diagnosis based on multi-source information fusion, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an OLTC mechanical fault diagnosis method based on multi-source information fusion, comprising: Pt.1. Collect low-frequency vibration signals and high-frequency acoustic emission signals during the mechanical operation of the OLTC, divide the operation interval according to the time axis information of the mechanical operation of the OLTC, and perform short-time Fourier transform on the low-frequency vibration signals and the high-frequency acoustic emission signals respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the mechanical operation of the OLTC. Pt.2. At the same time point, calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, to obtain the difference time-frequency diagram of the OLTC mechanical action process; Pt.3. Map the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library, and output the mechanical fault diagnosis result of the OLTC mechanical action process.

[0007] In a preferred embodiment, the step of separately acquiring low-frequency vibration signals and high-frequency acoustic emission signals during the OLTC's mechanical movement, and dividing the movement interval based on the time axis information during the OLTC's mechanical movement, includes: The signal amplitude is detected from the multi-source information data of the OLTC mechanical action process, and the moment when it first exceeds the background activity threshold is determined as the mechanical action start time, thus obtaining the start time identifier; Continuing to detect the signal amplitude after the start time marker, when the signal amplitude is continuously lower than the action completion threshold and the duration reaches the stable determination time, the end time of the duration is determined as the mechanical action termination time, and the termination time marker is obtained. Record the occurrence times of transient peaks with abrupt amplitude changes within the complete vibration time span defined by the start time marker and the end time marker to obtain the action boundary point sequence; Based on the two adjacent occurrence times in the action boundary point sequence, the complete vibration time span is divided into multiple continuous time periods, with each time period serving as an action interval.

[0008] In a preferred embodiment, prior to the short-time Fourier transform, the process includes: Based on the total duration of the action interval, the time span of a single analysis window is determined to be able to fully cover the duration of a mechanical impact event, and a window duration parameter is generated. The overlap duration between two adjacent analysis windows is determined to be half of the window duration parameter, and the overlap duration parameter is generated. Select the Hanning window from the preset window function library as the window function for the current transformation and generate a window function identifier; The window duration parameter, the overlap duration parameter, and the window function identifier are combined into a parameter configuration group, and the parameter configuration group is output as the basis for the execution of the short-time Fourier transform.

[0009] In a preferred embodiment, generating the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process via short-time Fourier transform includes: The original vibration signals of the vibration subsequence and acoustic subsequence in the action interval are divided into multiple short time segments that are connected end to end and have partial overlap, to obtain the window sequence of the OLTC mechanical action process. A cosine window function is applied to each analysis window in the window sequence to gradually reduce the amplitude at the window boundary to zero, thereby obtaining the vibration window data and acoustic emission window data of the OLTC mechanical action process. The vibration window data and the sound emission window data are multiplied point by point with the preset sine fundamental frequency signal and cosine fundamental frequency signal respectively and accumulated to obtain the local vibration time-frequency block and the local sound emission time-frequency block of the OLTC mechanical action process. The local vibration time-frequency block and the local acoustic emission time-frequency block are arranged in the order of the window time to obtain the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process.

[0010] In a preferred embodiment, calculating the difference in energy distribution between the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point includes: Perform energy value differential alignment operation between the first energy distribution curve in the initial vibration time-frequency diagram and the second energy distribution curve in the initial acoustic emission time-frequency diagram at each frequency node on the frequency axis; The differential alignment operation involves capturing the first energy value and the second energy value on the same frequency node one by one, and constructing a set of feature triplet queues of the first energy value and the second energy value associated with the frequency node. The difference amplitude is obtained by absolute processing of the difference between the first energy value and the second energy value, and the sum amplitude is obtained by summing and converging the first energy value and the second energy value. The difference characteristic value is obtained by normalizing the ratio of the difference amplitude to the sum amplitude, and the difference characteristic value is arranged in ascending order of frequency to form a difference characteristic spectral line. The difference characteristic spectral line is the difference between the energy distributions.

[0011] In a preferred embodiment, the formula for calculating the difference feature value is as follows: ; In the formula, Frequency position The difference feature value at that location, The frequency position in the first energy distribution curve The amplitude of vibration energy, The frequency position in the second energy distribution curve The amplitude of acoustic emission energy, This is a preset stability constant.

[0012] In a preferred embodiment, the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram includes: At the same frequency position, the vibration trend sequence of the initial vibration time-frequency diagram and the acoustic emission trend sequence of the initial acoustic emission time-frequency diagram are compared in trend direction; The trend direction comparison involves determining at each time point whether the direction of increase or decrease of vibration energy amplitude is opposite to the direction of increase or decrease of acoustic emission energy amplitude. When the direction of increase or decrease of the vibration energy amplitude and the direction of increase or decrease of the acoustic emission energy amplitude are opposite to each other at a certain frequency position for more than a preset number of time points, the frequency position is marked as a reverse trend frequency node. All frequency nodes marked as reverse trend frequency nodes are grouped together according to their frequency values ​​to form a continuous frequency range, thus obtaining the opposite frequency band region of the OLTC mechanical action process.

[0013] In a preferred embodiment, the step of enhancing the frequency band region in the initial vibration time-frequency diagram that has an energy change trend opposite to that in the initial acoustic emission time-frequency diagram based on the difference amount to obtain the difference time-frequency diagram of the OLTC mechanical action process includes: The subset of difference values ​​corresponding to the opposite frequency band region in the difference quantity is used as the basis for the enhanced weighting coefficient of the OLTC mechanical action process, based on the positive correlation between size. Based on the enhancement weighting coefficient, the original vibration energy amplitude of the opposite frequency band region is amplitude modulated to obtain the enhanced time-frequency block of the OLTC mechanical action process; The enhanced time-frequency block is replaced back to the position of the original time-frequency block in the initial vibration time-frequency diagram, while keeping the data of other frequency band regions outside the opposite frequency band region in the initial vibration time-frequency diagram unchanged, to obtain the difference time-frequency diagram of the OLTC mechanical action process.

[0014] In a preferred embodiment, mapping the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library and outputting the mechanical fault diagnosis result of the OLTC mechanical action process includes: The energy proportion parameters of multiple frequency band regions extracted from the difference time-frequency map, as well as the orientation parameters and curvature parameters of the time-frequency ridge, are aggregated into a fault-sensitive feature vector. The similarity of the fault-sensitive feature vectors with the reference feature vectors in the preset reference feature vector library is compared one by one to obtain the matching degree score queue of the OLTC mechanical action process. The reference feature vector corresponding to the highest matching score in the matching score queue is determined as the best matching reference vector; The fault type name and fault level information of the best matching reference vector are organized into a structured diagnostic report, and the structured diagnostic report is output as the mechanical fault diagnosis result of the OLTC mechanical action process.

[0015] To address the aforementioned problems, this invention also provides an OLTC mechanical fault diagnosis system based on multi-source information fusion, the system comprising: The partitioned time-frequency modeling module is used to collect low-frequency vibration signals and high-frequency acoustic emission signals during the mechanical operation of the OLTC, divide the operation interval according to the time axis information of the mechanical operation of the OLTC, and perform short-time Fourier transform on the low-frequency vibration signals and the high-frequency acoustic emission signals respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the mechanical operation of the OLTC. The differential spectral enhancement module is used to calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, so as to obtain the differential time-frequency diagram of the OLTC mechanical action process. The feature mapping diagnosis module is used to map the fault-sensitive feature vectors of the difference time-frequency diagram to a preset reference feature vector library and output the mechanical fault diagnosis results of the OLTC mechanical action process.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention relies on the OLTC mechanical motion time axis to accurately divide the motion interval, and generates the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram simultaneously through short-time Fourier transform, realizing the collaborative time-frequency modeling of multi-source signals of vibration and acoustic emission, and completely capturing the time-frequency characteristic information of the entire mechanical motion cycle. This makes the fault feature acquisition process more comprehensive and the data processing more standardized, significantly improving the completeness of fault feature extraction and the execution efficiency of the diagnostic process.

[0017] 2. This invention accurately calculates the energy distribution difference between the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram, specifically enhancing frequency bands with opposite energy change trends, significantly improving the identification and prominence of fault-sensitive features. Then, the fault-sensitive feature vectors of the difference time-frequency diagrams are mapped to a preset reference feature vector library to complete accurate matching, which can quickly output structured diagnostic results containing fault type and level, effectively improving the accuracy of fault judgment, the reliability of diagnostic conclusions, and the response speed of fault identification. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an embodiment of the OLTC mechanical fault diagnosis method based on multi-source information fusion provided by the present invention. Figure 2 This is a functional block diagram of an OLTC mechanical fault diagnosis system based on multi-source information fusion provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] This application provides an OLTC mechanical fault diagnosis method based on multi-source information fusion. The execution entity of the OLTC mechanical fault diagnosis method based on multi-source information fusion includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the OLTC mechanical fault diagnosis method based on multi-source information fusion can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0021] Reference Figure 1 The diagram shown is a flowchart illustrating an OLTC mechanical fault diagnosis method based on multi-source information fusion according to an embodiment of the present invention. In this embodiment, the OLTC mechanical fault diagnosis method based on multi-source information fusion includes: Pt.1. Collect low-frequency vibration signals and high-frequency acoustic emission signals during the mechanical operation of the OLTC, divide the operation interval according to the time axis information of the mechanical operation of the OLTC, and perform short-time Fourier transform on the low-frequency vibration signals and the high-frequency acoustic emission signals respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the mechanical operation of the OLTC. In this embodiment of the invention, the step of separately acquiring low-frequency vibration signals and high-frequency acoustic emission signals during the mechanical movement of the OLTC, and dividing the movement interval based on the time axis information during the mechanical movement of the OLTC, includes: The signal amplitude is detected from the multi-source information data of the OLTC mechanical action process, and the moment when it first exceeds the background activity threshold is determined as the mechanical action start time, thus obtaining the start time identifier; Continuing to detect the signal amplitude after the start time marker, when the signal amplitude is continuously lower than the action completion threshold and the duration reaches the stable determination time, the end time of the duration is determined as the mechanical action termination time, and the termination time marker is obtained. Record the occurrence times of transient peaks with abrupt amplitude changes within the complete vibration time span defined by the start time marker and the end time marker to obtain the action boundary point sequence; Based on the two adjacent occurrence times in the action boundary point sequence, the complete vibration time span is divided into multiple continuous time periods, with each time period serving as an action interval.

[0022] Before the short-time Fourier transform, the following are included: Based on the total duration of the action interval, the time span of a single analysis window is determined to be able to fully cover the duration of a mechanical impact event, and a window duration parameter is generated. The overlap duration between two adjacent analysis windows is determined to be half of the window duration parameter, and the overlap duration parameter is generated. Select the Hanning window from the preset window function library as the window function for the current transformation and generate a window function identifier; The window duration parameter, the overlap duration parameter, and the window function identifier are combined into a parameter configuration group, and the parameter configuration group is output as the basis for the execution of the short-time Fourier transform.

[0023] The generation of the initial vibration time-frequency diagram and initial acoustic emission time-frequency diagram of the OLTC mechanical action process through short-time Fourier transform includes: The original vibration signals of the vibration subsequence and acoustic subsequence in the action interval are divided into multiple short time segments that are connected end to end and have partial overlap, to obtain the window sequence of the OLTC mechanical action process. A cosine window function is applied to each analysis window in the window sequence to gradually reduce the amplitude at the window boundary to zero, thereby obtaining the vibration window data and acoustic emission window data of the OLTC mechanical action process. The vibration window data and the sound emission window data are multiplied point by point with the preset sine fundamental frequency signal and cosine fundamental frequency signal respectively and accumulated to obtain the local vibration time-frequency block and the local sound emission time-frequency block of the OLTC mechanical action process. The local vibration time-frequency block and the local acoustic emission time-frequency block are arranged in the order of the window time to obtain the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process.

[0024] The multi-source information data consisting of vibration signals and acoustic emission signals synchronously collected throughout the entire mechanical action of the OLTC is used to perform real-time detection of the signal amplitude at each moment according to a fixed time step. The moment when the detected signal amplitude first exceeds the preset background activity threshold is directly identified as the start moment of the OLTC mechanical action, thereby generating a start moment identifier to mark the start of the action.

[0025] After the start time marker, the signal amplitude is continuously detected in real time according to the same fixed time step. When the detected signal amplitude is continuously below the preset action completion threshold and the duration of the low amplitude state reaches the preset stability judgment duration, the final end time of the continuous state is identified as the end time of the OLTC mechanical action, thereby generating an end time marker to mark the end of the action.

[0026] Within the complete vibration time span defined by the start and end time markers, the dynamic change trajectory of the signal amplitude is continuously tracked in real time. The corresponding time points when all signal amplitudes suddenly and drastically change and form transient peaks are accurately recorded. All recorded time points are arranged one by one in natural chronological order to form a sequence of action boundary points used to divide the action phases.

[0027] Based on the independent time range defined by each pair of adjacent occurrences in the action boundary point sequence, the complete vibration time span is precisely divided into multiple consecutive independent time periods without time overlap. Each independent time period serves as the vibration action interval for feature analysis during the OLTC mechanical action process.

[0028] Based on the total duration of each vibration action interval that has been accurately divided in the early stage, the time span of a single analysis window is set to be the duration that can completely encompass the entire process of a mechanical impact event in the OLTC mechanical action from occurrence to end, thereby generating a window duration parameter to limit the length of the analysis window.

[0029] The overlap duration between two adjacent analysis windows is directly set to half the duration corresponding to the generated window duration parameter, thereby generating an overlap duration parameter used to control the degree of window overlap.

[0030] From a pre-built window function library that stores various window functions, the Hanning window is selected as the dedicated window function used in this short-time Fourier transform processing operation, thereby generating a window function identifier to identify the window function type.

[0031] The generated window duration parameters, overlap duration parameters, and window function identifiers are systematically integrated and uniformly collected to form a complete parameter configuration group containing all transformation configuration information. This parameter configuration group is stably output to the outside world as the direct and sole execution basis for subsequent short-time Fourier transform operations.

[0032] Based on the previously determined window duration and overlap duration parameters, the original vibration signals corresponding to the vibration subsequences and the original acoustic emission signals corresponding to the acoustic emission subsequences within each vibration action interval are precisely segmented and cut. This ensures that all the short-time segments obtained are arranged end-to-end with a fixed proportion of overlap, thereby obtaining the window sequence for the OLTC mechanical action process used for short-time Fourier transform.

[0033] Cosine window functions are applied to each independent analysis window in the window sequence to smooth the amplitude. The signal amplitude is gradually and smoothly reduced from the beginning boundary position to the end boundary position of the window, so that the signal amplitude at both boundary positions of the window smoothly drops to the zero amplitude state. Vibration window data and acoustic emission window data of OLTC mechanical action process are obtained for time-frequency analysis.

[0034] The vibration window data is multiplied point by point with the pre-set sine and cosine fundamental frequency signals. The results of multiplying all the points are then accumulated and integrated point by point to obtain the local vibration time-frequency block of the OLTC mechanical action process used for stitching the time-frequency diagram. The sound emission window data is multiplied point by point with the pre-set sine and cosine fundamental frequency signals. The results of multiplying all the points are then accumulated and integrated point by point to obtain the local sound emission time-frequency block of the OLTC mechanical action process used for stitching the time-frequency diagram.

[0035] All generated local vibration time-frequency blocks are arranged and seamlessly stitched together in the natural order of time corresponding to the analysis window to obtain the initial vibration time-frequency diagram of the OLTC mechanical action process for subsequent difference analysis. All generated local acoustic emission time-frequency blocks are arranged and seamlessly stitched together in the natural order of time corresponding to the analysis window to obtain the initial acoustic emission time-frequency diagram of the OLTC mechanical action process for subsequent difference analysis.

[0036] It should be noted that the low-frequency vibration signal and the high-frequency acoustic emission signal in this application have different physical generation mechanisms and frequency bands. They are acquired synchronously by independently set vibration sensors and acoustic emission sensors, respectively, and are independent signal sources. Therefore, the initial vibration time-frequency map and the initial acoustic emission time-frequency map generated in this application are two independent time-frequency maps, rather than multiple sub-maps obtained by dividing a single signal into frequency bands. The subsequent description in this application of 'calculating the difference in energy distribution between the initial vibration time-frequency map and the initial acoustic emission time-frequency map' refers to a joint analysis of the two independent time-frequency maps, rather than a comparison between different frequency bands within the same time-frequency map.

[0037] Pt.2. At the same time point, calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, to obtain the difference time-frequency diagram of the OLTC mechanical action process; In this embodiment of the invention, calculating the difference in energy distribution between the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point includes: Perform energy value differential alignment operation between the first energy distribution curve in the initial vibration time-frequency diagram and the second energy distribution curve in the initial acoustic emission time-frequency diagram at each frequency node on the frequency axis; The differential alignment operation involves capturing the first energy value and the second energy value on the same frequency node one by one, and constructing a set of feature triplet queues of the first energy value and the second energy value associated with the frequency node. The difference amplitude is obtained by absolute processing of the difference between the first energy value and the second energy value, and the sum amplitude is obtained by summing and converging the first energy value and the second energy value. The difference characteristic value is obtained by normalizing the ratio of the difference amplitude to the sum amplitude, and the difference characteristic value is arranged in ascending order of frequency to form a difference characteristic spectral line. The difference characteristic spectral line is the difference between the energy distributions.

[0038] The formula for calculating the difference feature value is as follows: ; In the formula, Frequency position The difference feature value at that location, The frequency position in the first energy distribution curve The amplitude of vibration energy, The frequency position in the second energy distribution curve The amplitude of acoustic emission energy, This is a preset stability constant.

[0039] The frequency band regions in the initial vibration time-frequency diagram that exhibit the opposite energy change trend to the initial acoustic emission time-frequency diagram include: At the same frequency position, the vibration trend sequence of the initial vibration time-frequency diagram and the acoustic emission trend sequence of the initial acoustic emission time-frequency diagram are compared in trend direction; The trend direction comparison involves determining at each time point whether the direction of increase or decrease of vibration energy amplitude is opposite to the direction of increase or decrease of acoustic emission energy amplitude. When the direction of increase or decrease of the vibration energy amplitude and the direction of increase or decrease of the acoustic emission energy amplitude are opposite to each other at a certain frequency position for more than a preset number of time points, the frequency position is marked as a reverse trend frequency node. All frequency nodes marked as reverse trend frequency nodes are grouped together according to their frequency values ​​to form a continuous frequency range, thus obtaining the opposite frequency band region of the OLTC mechanical action process.

[0040] The step of enhancing the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference amount to obtain the difference time-frequency diagram of the OLTC mechanical action process includes: The subset of difference values ​​corresponding to the opposite frequency band region in the difference quantity is used as the basis for the enhanced weighting coefficient of the OLTC mechanical action process, based on the positive correlation between size. Based on the enhancement weighting coefficient, the original vibration energy amplitude of the opposite frequency band region is amplitude modulated to obtain the enhanced time-frequency block of the OLTC mechanical action process; The enhanced time-frequency block is replaced back to the position of the original time-frequency block in the initial vibration time-frequency diagram, while keeping the data of other frequency band regions outside the opposite frequency band region in the initial vibration time-frequency diagram unchanged, to obtain the difference time-frequency diagram of the OLTC mechanical action process.

[0041] For the first energy distribution curve contained in the initial vibration time-frequency diagram and the second energy distribution curve contained in the initial acoustic emission time-frequency diagram, differential alignment of energy values ​​is performed one by one along the frequency axis from the lowest frequency node to the highest frequency node to ensure that the two types of energy data at the same frequency node are accurately matched.

[0042] When performing differential alignment, the first energy value and the second energy value corresponding to the same frequency node are extracted one by one according to the frequency axis. The frequency node is used as the core association item, and the corresponding frequency node, the first energy value and the second energy value are combined into independent data units. All independent data units are arranged in ascending order of frequency to form a feature triplet queue in which the first energy value and the second energy value are associated with the frequency node.

[0043] For each independent data unit in the feature triplet queue, the difference between the first energy value and the second energy value in the group is calculated and then the absolute value is taken to obtain the difference amplitude corresponding to the data unit. The first energy value and the second energy value in the same data unit are directly added to obtain the sum amplitude corresponding to the data unit.

[0044] The difference amplitude and the corresponding sum amplitude of each data unit in the feature triplet queue are normalized by ratio to obtain the difference feature value corresponding to the frequency node. The difference feature values ​​corresponding to all frequency nodes are arranged in ascending order of frequency value to form a continuous and complete difference feature spectrum. This difference feature spectrum is the difference between the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram in terms of energy distribution.

[0045] The difference characteristic value corresponding to the frequency position is generated by the vibration energy amplitude and acoustic emission energy amplitude at the same frequency node, which are then processed by difference absoluteization, summation and convergence, and ratio normalization.

[0046] The vibration energy amplitude at a frequency position is directly extracted from the fixed position of the corresponding frequency node on the first energy distribution curve inside the initial vibration time-frequency diagram. The extraction process does not change the original value of the energy amplitude.

[0047] The acoustic emission energy amplitude at a frequency position is directly extracted from the fixed position of the corresponding frequency node on the second energy distribution curve inside the initial acoustic emission time-frequency diagram. The extraction process does not change the original value of the energy amplitude.

[0048] The stability constant is a fixed value set in advance. This value is added to the calculation result of the sum range during the calculation process to ensure that the denominator does not become zero when the ratio is normalized, thus maintaining the stability and continuity of the entire calculation process.

[0049] This calculation method is used to accurately quantify the degree of energy distribution difference between the initial vibration time-frequency map and the initial acoustic emission time-frequency map at the same frequency node location, providing a quantitative basis for subsequent feature enhancement.

[0050] This calculation method generates standardized difference feature values ​​through ratio normalization, eliminating the influence of differences in energy amplitude at different frequency nodes, and providing accurate and unified numerical support for subsequent feature enhancement in the reverse frequency band region.

[0051] All the difference characteristic values ​​obtained by this calculation method can be arranged in ascending order of frequency values ​​to form difference characteristic spectral lines. These spectral lines can be directly used as the difference between the energy distribution of the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram for subsequent use.

[0052] At the same fixed frequency position, a trend direction comparison operation covering the entire time axis is carried out between the vibration trend sequence corresponding to the initial vibration time-frequency diagram and the acoustic emission trend sequence corresponding to the initial acoustic emission time-frequency diagram.

[0053] When conducting trend direction comparison, the direction of increase or decrease of vibration energy amplitude and direction of increase or decrease of acoustic emission energy amplitude are judged one by one along the time axis from the start time to the end time, and it is determined whether the two change directions are completely opposite to each other.

[0054] When the total number of time points at a certain fixed frequency position where the direction of increase or decrease in vibration energy amplitude is different from the direction of increase or decrease in acoustic emission energy amplitude exceeds the preset judgment number, the fixed frequency position is marked as a reverse trend frequency node.

[0055] All marked reverse trend frequency nodes are continuously collected and organized in ascending order of frequency value. Adjacent and consecutive reverse trend frequency nodes are combined into a continuous frequency range, which is the opposite frequency band region in the OLTC mechanical operation process.

[0056] From the difference quantities formed by the difference characteristic spectral lines, extract all difference values ​​corresponding to the frequency range covered by the opposite frequency band region, and integrate and combine these difference values ​​to form a subset of difference values.

[0057] According to the rule that the magnitude of the difference value is positively correlated with the enhancement weight coefficient, the larger the difference value, the larger the corresponding enhancement weight coefficient is generated. The enhancement weight coefficient is generated for each difference value in the subset of difference values ​​to obtain the enhancement weight coefficient of the OLTC mechanical motion process.

[0058] Based on the generated enhancement weighting coefficients, the amplitude of the original vibration energy corresponding to each frequency node in the opposite frequency band region is adjusted point by point. The adjusted time and frequency data are combined as a whole to form the enhanced time and frequency block of the OLTC mechanical action process.

[0059] The generated enhanced time-frequency block is precisely placed at the fixed position corresponding to the original time-frequency block in the initial vibration time-frequency map. The data of all frequency band regions outside the coverage area of ​​the opposite frequency band region in the initial vibration time-frequency map are kept unchanged in their original values ​​and shapes. Finally, the difference time-frequency map of the OLTC mechanical action process is obtained for fault feature extraction.

[0060] Pt.3. Map the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library, and output the mechanical fault diagnosis result of the OLTC mechanical action process.

[0061] In this embodiment of the invention, mapping the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library and outputting the mechanical fault diagnosis result of the OLTC mechanical action process includes: The energy proportion parameters of multiple frequency band regions extracted from the difference time-frequency map, as well as the orientation parameters and curvature parameters of the time-frequency ridge, are aggregated into a fault-sensitive feature vector. The similarity of the fault-sensitive feature vectors with the reference feature vectors in the preset reference feature vector library is compared one by one to obtain the matching degree score queue of the OLTC mechanical action process. The reference feature vector corresponding to the highest matching score in the matching score queue is determined as the best matching reference vector; The fault type name and fault level information of the best matching reference vector are organized into a structured diagnostic report, and the structured diagnostic report is output as the mechanical fault diagnosis result of the OLTC mechanical action process.

[0062] The energy proportion parameters corresponding to each of the divided frequency band regions are extracted from the difference time-frequency map. Simultaneously, the direction parameters and curvature parameters corresponding to the time-frequency ridges inside the difference time-frequency map are extracted accurately. All the extracted energy proportion parameters, time-frequency ridge direction parameters, and time-frequency ridge curvature parameters are integrated and collected in a fixed feature dimension order to form a fault-sensitive feature vector with complete dimensions and comprehensive features.

[0063] The generated fault-sensitive feature vectors are compared with each standard reference feature vector stored in the pre-built reference feature vector library in sequence, and full-dimensional feature matching is carried out. For each set of feature vectors to be compared, the corresponding matching degree value is obtained by comparing the numerical consistency and change trend consistency of each feature dimension. All the matching degree values ​​generated by the comparison are sorted in order according to the order of comparison with the reference feature vectors to form a complete matching degree scoring queue of OLTC mechanical action process.

[0064] The matching score queue is fully traversed, and all scores are compared and filtered one by one. The matching score with the highest value is accurately located and selected. The reference feature vector associated with the highest matching score is determined as the best matching reference vector for this fault diagnosis.

[0065] From the determined best matching reference vector, the pre-bound corresponding fault type name information and fault level information are fully extracted. The two core information types, fault type name and fault level, are standardized and integrated according to the preset fixed report format to generate a structured diagnostic report with complete structure and clear information. This structured diagnostic report is then officially output as the final mechanical fault diagnosis result of the OLTC mechanical action process.

[0066] like Figure 2 The diagram shown is a functional block diagram of an OLTC mechanical fault diagnosis system based on multi-source information fusion provided in an embodiment of the present invention.

[0067] The OLTC mechanical fault diagnosis system 100 based on multi-source information fusion described in this invention can be installed in an electronic device. Depending on the functions implemented, the OLTC mechanical fault diagnosis system 100 based on multi-source information fusion may include a partitioned time-frequency modeling module, a differential spectral domain enhancement module, and a feature mapping diagnosis module. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and is stored in the memory of the electronic device.

[0068] In this embodiment, the functions of each module / unit are as follows: The partitioned time-frequency modeling module is used to collect low-frequency vibration signals and high-frequency acoustic emission signals during the mechanical operation of the OLTC, divide the operation interval according to the time axis information of the mechanical operation of the OLTC, and perform short-time Fourier transform on the low-frequency vibration signals and the high-frequency acoustic emission signals respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the mechanical operation of the OLTC. The differential spectral enhancement module is used to calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, so as to obtain the differential time-frequency diagram of the OLTC mechanical action process. The feature mapping diagnosis module is used to map the fault-sensitive feature vectors of the difference time-frequency diagram to a preset reference feature vector library, and output the mechanical fault diagnosis results of the OLTC mechanical action process. In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for example, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0069] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0070] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0071] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0072] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for OLTC mechanical fault diagnosis based on multi-source information fusion, characterized in that, The method includes: Pt.

1. Low-frequency vibration signals and high-frequency acoustic emission signals were collected during the mechanical movement of the OLTC, and the movement interval was divided according to the time axis information of the OLTC's mechanical movement, including: The signal amplitude is detected from the multi-source information data of the OLTC mechanical action process, and the moment when it first exceeds the background activity threshold is determined as the mechanical action start time, thus obtaining the start time identifier; Continuing to detect the signal amplitude after the start time marker, when the signal amplitude is continuously lower than the action completion threshold and the duration reaches the stable determination time, the end time of the duration is determined as the mechanical action termination time, and the termination time marker is obtained. Record the occurrence times of transient peaks with abrupt amplitude changes within the complete vibration time span defined by the start time marker and the end time marker to obtain the action boundary point sequence; Based on the two adjacent occurrence times in the action boundary point sequence, the complete vibration time span is divided into multiple consecutive time periods, with each time period serving as an action interval; Short-time Fourier transforms are performed on the low-frequency vibration signal and the high-frequency acoustic emission signal respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process; Pt.

2. At the same time point, calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, to obtain the difference time-frequency diagram of the OLTC mechanical action process; Pt.

3. Map the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library, and output the mechanical fault diagnosis result of the OLTC mechanical action process.

2. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 1, characterized in that, Before the short-time Fourier transform, the following are included: Based on the total duration of the action interval, the time span of a single analysis window is determined to be able to fully cover the duration of a mechanical impact event, and a window duration parameter is generated. The overlap duration between two adjacent analysis windows is determined to be half of the window duration parameter, and the overlap duration parameter is generated. Select the Hanning window from the preset window function library as the window function for the current transformation and generate a window function identifier; The window duration parameter, the overlap duration parameter, and the window function identifier are combined into a parameter configuration group, and the parameter configuration group is output as the basis for the execution of the short-time Fourier transform.

3. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 1, characterized in that, The generation of the initial vibration time-frequency diagram and initial acoustic emission time-frequency diagram of the OLTC mechanical action process through short-time Fourier transform includes: The original vibration signals of the vibration subsequence and acoustic subsequence in the action interval are divided into multiple short time segments that are connected end to end and have partial overlap, to obtain the window sequence of the OLTC mechanical action process. A cosine window function is applied to each analysis window in the window sequence to gradually reduce the amplitude at the window boundary to zero, thereby obtaining the vibration window data and acoustic emission window data of the OLTC mechanical action process. The vibration window data and the sound emission window data are multiplied point by point with the preset sine fundamental frequency signal and cosine fundamental frequency signal respectively and accumulated to obtain the local vibration time-frequency block and the local sound emission time-frequency block of the OLTC mechanical action process. The local vibration time-frequency block and the local acoustic emission time-frequency block are arranged in the order of the window time to obtain the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process.

4. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 1, characterized in that, The calculation of the difference in energy distribution between the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point includes: Perform energy value differential alignment operation between the first energy distribution curve in the initial vibration time-frequency diagram and the second energy distribution curve in the initial acoustic emission time-frequency diagram at each frequency node on the frequency axis; The differential alignment operation involves capturing the first energy value and the second energy value on the same frequency node one by one, and constructing a set of feature triplet queues of the first energy value and the second energy value associated with the frequency node. The difference amplitude is obtained by absolute processing of the difference between the first energy value and the second energy value, and the sum amplitude is obtained by summing and converging the first energy value and the second energy value. The difference characteristic value is obtained by normalizing the ratio of the difference amplitude to the sum amplitude, and the difference characteristic value is arranged in ascending order of frequency to form a difference characteristic spectral line. The difference characteristic spectral line is the difference between the energy distributions.

5. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 4, characterized in that, The formula for calculating the difference feature value is as follows: ； In the formula, Frequency position The difference feature value at that location, The frequency position in the first energy distribution curve The amplitude of vibration energy, The frequency position in the second energy distribution curve The amplitude of acoustic emission energy, This is a preset stability constant.

6. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 5, characterized in that, The frequency band regions in the initial vibration time-frequency diagram that exhibit the opposite energy change trend to the initial acoustic emission time-frequency diagram include: At the same frequency position, the vibration trend sequence of the initial vibration time-frequency diagram and the acoustic emission trend sequence of the initial acoustic emission time-frequency diagram are compared in trend direction; The trend direction comparison involves determining at each time point whether the direction of increase or decrease of vibration energy amplitude is opposite to the direction of increase or decrease of acoustic emission energy amplitude. When the direction of increase or decrease of the vibration energy amplitude and the direction of increase or decrease of the acoustic emission energy amplitude are opposite to each other at a certain frequency position for more than a preset number of time points, the frequency position is marked as a reverse trend frequency node. All frequency nodes marked as reverse trend frequency nodes are grouped together according to their frequency values ​​to form a continuous frequency range, thus obtaining the opposite frequency band region of the OLTC mechanical action process.

7. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 6, characterized in that, The step of enhancing the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference amount to obtain the difference time-frequency diagram of the OLTC mechanical action process includes: The subset of difference values ​​corresponding to the opposite frequency band region in the difference quantity is used as the basis for the enhanced weighting coefficient of the OLTC mechanical action process, based on the positive correlation between size. Based on the enhancement weighting coefficient, the original vibration energy amplitude of the opposite frequency band region is amplitude modulated to obtain the enhanced time-frequency block of the OLTC mechanical action process; The enhanced time-frequency block is replaced back to the position of the original time-frequency block in the initial vibration time-frequency diagram, while keeping the data of other frequency band regions outside the opposite frequency band region in the initial vibration time-frequency diagram unchanged, to obtain the difference time-frequency diagram of the OLTC mechanical action process.

8. The OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 1, characterized in that, The step of mapping the fault-sensitive feature vector of the difference time-frequency diagram to a preset reference feature vector library and outputting the mechanical fault diagnosis result of the OLTC mechanical action process includes: The energy proportion parameters of multiple frequency band regions extracted from the difference time-frequency map, as well as the orientation parameters and curvature parameters of the time-frequency ridge, are aggregated into a fault-sensitive feature vector. The similarity of the fault-sensitive feature vectors with the reference feature vectors in the preset reference feature vector library is compared one by one to obtain the matching degree score queue of the OLTC mechanical action process. The reference feature vector corresponding to the highest matching score in the matching score queue is determined as the best matching reference vector; The fault type name and fault level information of the best matching reference vector are organized into a structured diagnostic report, and the structured diagnostic report is output as the mechanical fault diagnosis result of the OLTC mechanical action process.

9. An OLTC mechanical fault diagnosis system based on multi-source information fusion, characterized in that, The system for implementing the OLTC mechanical fault diagnosis method based on multi-source information fusion as described in claim 1 includes: The partitioned time-frequency modeling module is used to separately collect low-frequency vibration signals and high-frequency acoustic emission signals during the OLTC's mechanical operation. It divides the operation intervals based on the time axis information of the OLTC's mechanical operation, specifically for: The signal amplitude is detected from the multi-source information data of the OLTC mechanical action process, and the moment when it first exceeds the background activity threshold is determined as the mechanical action start time, thus obtaining the start time identifier; Continuing to detect the signal amplitude after the start time marker, when the signal amplitude is continuously lower than the action completion threshold and the duration reaches the stable determination time, the end time of the duration is determined as the mechanical action termination time, and the termination time marker is obtained. Record the occurrence times of transient peaks with abrupt amplitude changes within the complete vibration time span defined by the start time marker and the end time marker to obtain the action boundary point sequence; Based on the two adjacent occurrence times in the action boundary point sequence, the complete vibration time span is divided into multiple consecutive time periods, with each time period serving as an action interval; Short-time Fourier transforms are performed on the low-frequency vibration signal and the high-frequency acoustic emission signal respectively to independently generate the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram of the OLTC mechanical action process; The differential spectral enhancement module is used to calculate the difference between the energy distribution in the initial vibration time-frequency diagram and the initial acoustic emission time-frequency diagram at the same time point, and enhance the frequency band region in the initial vibration time-frequency diagram that has the opposite energy change trend to the initial acoustic emission time-frequency diagram based on the difference, so as to obtain the differential time-frequency diagram of the OLTC mechanical action process. The feature mapping diagnosis module is used to map the fault-sensitive feature vectors of the difference time-frequency diagram to a preset reference feature vector library and output the mechanical fault diagnosis results of the OLTC mechanical action process.

Images