CN117848423A - On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell - Google Patents
On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell Download PDFInfo
- Publication number
- CN117848423A CN117848423A CN202410257398.0A CN202410257398A CN117848423A CN 117848423 A CN117848423 A CN 117848423A CN 202410257398 A CN202410257398 A CN 202410257398A CN 117848423 A CN117848423 A CN 117848423A
- Authority
- CN
- China
- Prior art keywords
- grid
- parameters
- monitoring
- sleeve
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 87
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000008859 change Effects 0.000 claims abstract description 37
- 230000006870 function Effects 0.000 claims abstract description 14
- 238000004458 analytical method Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 13
- 238000013136 deep learning model Methods 0.000 claims abstract description 12
- 230000002159 abnormal effect Effects 0.000 claims description 20
- 238000013527 convolutional neural network Methods 0.000 claims description 12
- 239000013598 vector Substances 0.000 claims description 11
- 230000004927 fusion Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000003247 decreasing effect Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 18
- 230000007547 defect Effects 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000006247 magnetic powder Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005297 material degradation process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to the field of intelligent detection control of power grid equipment, and discloses an online monitoring method, system, equipment and medium for the integrity of a casing of a converter transformer valve side casing, wherein the method comprises the steps of obtaining the size and an image of the casing, and establishing a first grid on the surface of the casing according to the size; expanding a first grid, and dividing a sleeve middle section and a connecting section according to image recognition; obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model; acquiring and monitoring current flow and pressure parameters of the connection section through an SVM algorithm; performing change analysis according to the acquired real-time data, and adjusting the division of the first grid to obtain a second grid; and monitoring the parameters of the second grid area in real time, comparing the differences between the data, and carrying out early warning on the shell problem. According to the invention, the detection range can be accurately divided, the daily monitoring amount is reduced, the prediction model is built by combining the data characteristics, the monitoring and prediction result is directly output, the real-time integrity is strong, and a certain early warning function can be realized.
Description
Technical Field
The invention relates to the field of intelligent detection control of power grid equipment, in particular to an online monitoring method, an online monitoring system, online monitoring equipment and online monitoring media for the integrity of a converter transformer valve side sleeve shell.
Background
Detecting the integrity of a casing on the converter transformer valve side is a key link for ensuring normal and safe operation of equipment, and the loss of the integrity of the casing can lead to electric leakage, explosion or mechanical failure, thereby causing serious safety problems and endangering the safety of equipment, personnel and environment; and damage or defects of the casing may affect the stable operation of the converter valve, leading to unstable and even power failure of the power system.
Most detection modes at present are visual detection, and identification is directly carried out through images; detecting internal defects or cracks of the material by utilizing an ultrasonic technology, and detecting the integrity of the material by sending ultrasonic waves and observing a reflection mode of the ultrasonic waves; detecting surface temperature change by using a thermal infrared imager, and identifying abnormal heat distribution so as to find defects on the surface and in the interior; the magnetic powder is smeared on the surface of the sleeve, and cracks or defects are detected by observing the aggregation of the magnetic powder at the defects. These different detection methods may be only suitable for specific types of defects, not cover all possible problems, require highly skilled professionals to operate and interpret the results, and some methods may require down detection or longer detection times, not enable real-time monitoring and feedback.
Disclosure of Invention
The present invention has been made in view of the above-described problems occurring in the prior art.
Therefore, the invention provides an online monitoring method and system for the integrity of a converter transformer valve side sleeve shell, which solve the problems that important detection parameters cannot be covered on the whole surface at the same time at present, and the problems that manual interpretation results are needed, the time consumption is long, the machine is likely to stop for detection, and real-time feedback and early warning cannot be performed.
In order to solve the technical problems, the invention provides the following technical scheme:
in a first aspect, the invention provides an on-line monitoring method for the integrity of a converter transformer valve side sleeve shell, comprising the following steps:
acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
expanding a first grid, and dividing a sleeve middle section and a connecting section according to image recognition;
obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model;
acquiring and monitoring current flow and pressure parameters of the connection section through an SVM algorithm;
performing change analysis according to the acquired real-time data, and adjusting the division of the first grid, wherein the adjustment is to increase the grid density of the area with abnormal change to obtain a second grid;
and monitoring the parameters of the second grid area in real time, comparing the differences between the data, and carrying out early warning on the shell problem.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: the establishing a first grid on the surface of the sleeve according to the size comprises the following steps:
taking the height direction of the sleeve as an X axis and taking the direction perpendicular to the height direction as a Y axis;
taking the midpoint of the height direction as the origin of a coordinate system;
the sleeve surface is uniformly divided into rectangular block areas of the same size.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: the expanding the first grid, dividing the middle section and the connecting section of the sleeve according to image recognition, comprises: expanding the first grid along the Y axis, and if the sleeve structure has a convex structure, continuing to expand along the X axis until the sleeve structure is expanded into a two-dimensional plane;
on the unfolded two-dimensional plane, identifying and determining the area where the convex structure is located according to the original structure of the sleeve, and expanding according to an expanding rule;
the first expansion rule is that the edge of the area where the convex structure is located is continuously expanded by 2 rectangular block areas along the X axis and expanded to two-dimensional plane edge rectangular blocks along the Y axis so as to obtain a first plane, the first plane is marked as a connecting section, and the rest plane parts are middle sections;
the second expansion rule is that if a part of the convex structure has two-dimensional plane edges and cannot be expanded outwards, the expandable edges are expanded according to the first expansion rule to obtain a first plane, the first plane is marked as a joint section, and the rest plane parts are middle sections;
the third expansion rule is that if only one end of the sleeve has a convex structure, after expansion according to the first or second expansion rule, a second plane symmetrical to the first plane is obtained by taking the Y axis as a symmetry axis, and the second plane is marked as a connecting section, and the rest plane parts are middle sections;
if the convex structure does not exist, the Y axis is taken as a symmetry axis, two thirds of the length of the two-dimensional plane are respectively extended to the two ends of the X axis, a third plane is obtained, the mark is a middle section, and the rest plane parts are joint sections.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model, wherein the method comprises the following steps of: acquiring image data of an intermediate section and time sequence data of vibration and temperature parameters, and preprocessing;
extracting image features through a CNN model, wherein the image features comprise color changes, cracks, fine lines and edge contours;
inputting the image characteristics and vibration and temperature parameters into an LSTM network model for characteristic fusion to obtain characteristic vectors combining three data sources;
and inputting the feature vector into the CNN model again for convolutional neural network learning to form a first model, wherein the first model is a neural network model with vibration and temperature prediction.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: current flow and pressure parameters of the junction section are obtained and monitored through an SVM algorithm, and the method comprises the following steps: acquiring current flow and pressure parameters of the connection section as a sample set, extracting characteristic values and preprocessing;
extracting current flow characteristics comprises frequency distribution, current mean and variance and current difference value of preset time interval;
extracting pressure characteristics comprises the average value and peak value of pressure and the current difference value of a preset time interval;
the selection range of the preset time interval is 10-20 minutes;
and inputting the extracted characteristic values into an SVM model, taking RBF as a kernel function, performing parameter tuning, and forming a second model through cross verification, wherein the second model is provided with a model of a current value and a shell pressure value.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: performing change analysis according to the acquired real-time data, and adjusting the division of the first grid, wherein the adjustment is to increase the grid density of the area with abnormal change to obtain a second grid, and the method comprises the following steps:
acquiring real-time data through the first model and the second model, and locating a changed area through the feature vector when the acquired vibration value exceeds thirty percent of the historical mean value or a certain temperature value changes beyond the variance within the past half hour;
when the acquired continuous current values or pressure values are continuously increased or decreased, marking as abnormal, and directly positioning the changed area;
when the current value or the pressure value in a certain time period exceeds 50% of the historical average value, marking as abnormal, and directly positioning a changed area;
and increasing the grid density of the change area to 1.5 times of the original density to obtain a second grid.
As a preferred scheme of the on-line monitoring method for the integrity of the converter transformer valve side sleeve shell, the invention comprises the following steps: monitoring the second grid region parameters in real time, comparing the differences between the data, and performing fault prediction, wherein the method comprises the following steps:
acquiring parameters of a second grid region, extracting parameters of a change region, inputting the parameters into respective corresponding models for parameter value prediction, and obtaining first prediction data;
taking the average value of the residual errors of the actually monitored data and the first prediction data as a reference, and if the residual errors exceed twice the standard deviation of the average value, carrying out shell problem early warning.
In a second aspect, the present invention provides an on-line monitoring system for converter transformer valve side casing integrity, comprising: the first grid construction module is used for acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
the first dividing module is used for expanding the first grid and dividing the middle section and the connecting section of the sleeve according to image recognition;
the first model monitoring module is used for acquiring and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model;
the second model monitoring module is used for acquiring and monitoring current flow and pressure parameters of the connecting section through an SVM algorithm;
the second dividing module is used for carrying out change analysis according to the acquired real-time data, adjusting the division of the first grid, wherein the adjustment is to increase the grid density of the area with abnormal change to obtain a second grid;
and the early warning module is used for monitoring the parameters of the second grid area in real time, comparing the differences between the data and carrying out early warning on the problems of the shell.
In a third aspect, the present invention provides a computing device comprising:
a memory and a processor;
the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions, and the computer executable instructions realize the steps of the online monitoring method for the integrity of the converter transformer valve side sleeve shell when being executed by the processor.
In a fourth aspect, the present invention provides a computer readable storage medium storing computer executable instructions which, when executed by a processor, implement the steps of the method for on-line monitoring of converter valve side casing integrity.
Compared with the prior art, the invention has the beneficial effects that: according to the invention, the detection range is accurately divided by establishing the grid and the two-dimensional plane, so that the condition of redundant data of daily monitoring is reduced, the daily monitoring data is subjected to condition detection, and the daily monitoring needs are met; the key data are obtained by setting the dividing rules corresponding to the parameters, the prediction model is built by combining the data characteristics, and the data are subjected to real-time training prediction, so that the monitoring and prediction results can be directly output, the visualization degree is high, the real-time performance is high, and a certain early warning function can be realized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic overall flow chart of an on-line monitoring method for the integrity of a casing on a converter transformer valve side according to a first embodiment of the invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
Example 1
Referring to fig. 1, for one embodiment of the present invention, there is provided an on-line monitoring method of converter valve side casing integrity, comprising,
s1: acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
still further, establishing a first grid on the surface of the cannula, depending on the size, includes,
taking the height direction of the sleeve as an X axis and taking the direction perpendicular to the height direction as a Y axis;
taking the midpoint of the height direction as the origin of a coordinate system;
the sleeve surface is uniformly divided into rectangular block areas of the same size.
It should be noted that, selecting the midpoint of the height value as the origin of the coordinate system facilitates subsequent symmetric or asymmetric region selection according to different shapes of the casing, which can reduce the computational effort required for modeling.
S2: expanding a first grid, and dividing a sleeve middle section and a connecting section according to image recognition;
further, the first grid is unfolded, and the middle section and the connecting section of the sleeve are divided according to image recognition, wherein the first grid is unfolded along the Y axis, and if a convex structure exists in the sleeve structure, the first grid is unfolded along the X axis until the first grid is unfolded into a two-dimensional plane;
on the unfolded two-dimensional plane, identifying and determining the area where the convex structure is located according to the original structure of the sleeve, and expanding according to an expanding rule;
the first expansion rule is that the edge of the area where the convex structure is located is continuously expanded by 2 rectangular block areas along the X axis and expanded to two-dimensional plane edge rectangular blocks along the Y axis so as to obtain a first plane, the first plane is marked as a connecting section, and the rest plane parts are middle sections;
the second expansion rule is that if a part of the convex structure has two-dimensional plane edges and cannot be expanded outwards, the expandable edges are expanded according to the first expansion rule to obtain a first plane, the first plane is marked as a joint section, and the rest plane parts are middle sections;
the third expansion rule is that if only one end of the sleeve has a convex structure, after expansion according to the first or second expansion rule, a second plane symmetrical to the first plane is obtained by taking the Y axis as a symmetry axis, and the second plane is marked as a connecting section, and the rest plane parts are middle sections;
if the convex structure does not exist, the Y axis is taken as a symmetry axis, two thirds of the length of the two-dimensional plane are respectively extended to the two ends of the X axis, a third plane is obtained, the mark is a middle section, and the rest plane parts are joint sections.
It should be noted that, because the daily monitoring is large in data volume, the real-time online integrity detection is only needed for the most core parameters of each part of the sleeve, the sleeve and the data are classified and detected in a dividing mode, and the intermediate section and the connecting section are respectively subjected to grid establishment and expansion. Corresponding unfolding rules are set according to different sleeve structures, so that the sleeve structures can be fully displayed to the greatest extent and the greatest efficiency.
S3: obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model;
further, the vibration parameter and the temperature parameter of the middle section are detected by a deep learning model, including,
acquiring image data of an intermediate section and time sequence data of vibration and temperature parameters, and preprocessing;
extracting image features including color change, cracks, fine lines and edge contours through a CNN model;
it should be noted that the detection of vibration and temperature of the intermediate section of the casing by the deep learning model avoids destructive detection of the equipment, and can continuously and finely monitor the real-time state of the intermediate section of the casing. Among them, the present invention prefers color change and fine lines for the selection of input data, and increases problems of material degradation possibly caused by fine temperature change and possibility of corrosion by gas or objects in the pipe compared to general direct detection of edge profile or crack.
Inputting the image characteristics, vibration and temperature parameters into an LSTM network model for characteristic fusion to obtain characteristic vectors combining three data sources;
it should be noted that by fusing the features of different data sources, the LSTM network can better learn the inherent correlation between data and help to improve the predictive performance of the model. The feature fusion is beneficial to better capturing potential relations among data by the model, and the accuracy and the robustness of prediction are improved. The image characteristics, vibration and temperature parameters are different in physical characteristics and perceptual characteristics, and the fusion of the image characteristics, vibration and temperature parameters can provide richer and more comprehensive multidimensional information.
And inputting the feature vector into the CNN model again to perform convolutional neural network learning to form a first model, wherein the first model is a neural network model with vibration and temperature prediction.
It should be noted that, through multi-level feature extraction, the model can better understand the structure and rule of data, and provide richer information for vibration and temperature prediction.
S4: acquiring and monitoring current flow and pressure parameters of the connection section through an SVM algorithm;
further, current flow and pressure parameters of the joint section are detected and obtained through an SVM algorithm, including,
acquiring current flow and pressure parameters of the connection section as a sample set, extracting characteristic values and preprocessing;
extracting current flow characteristics comprises frequency distribution, current mean and variance and current difference value of preset time interval;
it should be noted that the frequency distribution is helpful to understand the frequency domain characteristics of the current, and can understand the change of the current peak value through the analysis of the frequency spectrum, so as to prevent surge and overload protection; examining the change in current over different time periods can help identify sudden events or changes.
Extracting pressure characteristics comprises the pressure difference value between the average value and the peak value of pressure and at a preset time interval;
it should be noted that the pressure difference value at the preset time interval can reveal a trend of abnormal or varying pressure.
Specifically, the preset time interval is selected within a range of 10-20 minutes.
And inputting the extracted characteristic values into an SVM model, taking RBF as a kernel function, performing parameter tuning, and forming a second model through cross verification, wherein the second model is a model with a current value and a shell pressure value.
It should be noted that RBF kernel functions are chosen to be able to handle non-linear data, and current flow and pressure parameters are often not simply linear relationships, but may include more complex non-linear relationships that may occur under specific pressure or current conditions, thus improving generalization and classification performance of the model through tuning of kernel function parameters.
S5: performing change analysis according to the acquired real-time data, adjusting the division of the first grid to increase the grid density of the area with abnormal change, and obtaining a second grid;
further, the method comprises performing variation analysis based on the obtained real-time data, adjusting the division of the first grid to increase the grid density of the region with abnormal variation to obtain a second grid, including,
acquiring real-time data through the first model and the second model, and locating a changed area through a feature vector when the acquired vibration value exceeds thirty percent of a historical mean value or a certain temperature value changes beyond variance in the past half hour;
it should be noted that, setting corresponding values for the abnormality or sudden change in the captured data, the method can directly locate by combining the earlier characteristic vector in a simpler comparison mode, which is helpful for locating the specific area where the change or abnormality occurs more accurately; meanwhile, the positioning mode can further provide the next action for solving the problem, reduce the possibility of false alarm and quickly identify the potential problem area.
When the acquired continuous current values or pressure values are continuously increased or decreased, marking as abnormal, and directly positioning the changed area;
specifically, it may be preferable that consecutive 5 current values or pressure values be marked as abnormal when they are continuously increased or decreased.
When the current value or the pressure value in a certain time period exceeds 50% of the historical average value, marking as abnormal, and directly positioning a changed area;
the grid density of the change area is increased to 1.5 times of the original density, and a second grid is obtained.
It should be noted that increasing the grid density for the change region alone can improve the resolution and detail capture capabilities of the change region, provide more reliable data prediction and monitoring functions, and facilitate more accurate analysis of the change region data to discover potential problems in time and take necessary actions. And the overall calculation amount is greatly reduced, and the labor cost is also reduced.
S6: and monitoring the parameters of the second grid area in real time, comparing the differences between the data, and carrying out early warning on the shell problem.
Further, the second grid region parameters are monitored in real time, the differences between the data are compared, fault prediction is performed, including,
acquiring parameters of a second grid region, extracting parameters of a change region, inputting the parameters into respective corresponding models for parameter value prediction, and obtaining first prediction data;
it should be noted that this step aims at analyzing the parameters of the variation region and predicting the parameter values using a pre-established model, which helps to see if the current parameters meet the expected values and are abnormal.
Taking the average value of the residual errors of the actually monitored data and the first prediction data as a reference, and if the residual errors exceed twice the standard deviation of the average value, carrying out shell problem early warning.
It should be noted that the residual is the difference between the actual observed value and the predicted value, and can be used to find anomalies.
The above is a schematic scheme of an online monitoring method for the integrity of the converter transformer valve side sleeve shell of the present embodiment. It should be noted that, the technical solution of the online monitoring system for the integrity of the converter transformer valve side sleeve shell and the technical solution of the online monitoring method for the integrity of the converter transformer valve side sleeve shell belong to the same concept, and in this embodiment, details of the technical solution of the online monitoring system for the integrity of the converter transformer valve side sleeve shell are not described in detail, and all reference may be made to the description of the technical solution of the online monitoring method for the integrity of the converter transformer valve side sleeve shell.
The online monitoring system for the integrity of the converter transformer valve side sleeve shell in the embodiment comprises:
the first grid construction module is used for acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
the first dividing module is used for expanding the first grid and dividing the middle section and the connecting section of the sleeve according to image recognition;
the first model monitoring module is used for acquiring and monitoring vibration parameters and temperature parameters of the middle section through the deep learning model;
the second model monitoring module is used for acquiring and monitoring current flow and pressure parameters of the connecting section through an SVM algorithm;
the second dividing module is used for carrying out change analysis according to the acquired real-time data, adjusting the division of the first grid to increase the grid density of the area with abnormal change and obtaining a second grid;
and the early warning module is used for monitoring the parameters of the second grid area in real time, comparing the differences between the data and carrying out early warning on the problems of the shell.
The embodiment also provides a computing device, which is suitable for the condition of on-line monitoring of the integrity of the converter transformer valve side sleeve shell, and comprises:
a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the on-line monitoring method for the integrity of the sleeve shell at the converter transformer valve side according to the embodiment.
The present embodiment also provides a storage medium having stored thereon a computer program which, when executed by a processor, implements the on-line monitoring method for achieving the integrity of the converter transformer valve side sleeve housing as proposed in the above embodiments.
The storage medium proposed in this embodiment belongs to the same inventive concept as the online monitoring method for realizing the integrity of the converter transformer valve side casing provided in the above embodiment, and technical details not described in detail in this embodiment can be seen in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.
From the above description of embodiments, it will be clear to a person skilled in the art that the present invention may be implemented by means of software and necessary general purpose hardware, but of course also by means of hardware, although in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk, a Read Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, etc., including several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to execute the method of the embodiments of the present invention.
Example 2
Referring to tables 1-3, for one embodiment of the present invention, an online monitoring method for implementing integrity of a casing of a converter transformer valve side casing is provided, and in order to verify beneficial effects, an exemplary application scenario and a comparison scheme are provided for scientific demonstration.
S1: acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size; s2: expanding a first grid, and dividing a sleeve middle section and a connecting section according to image recognition; s3: obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model; s4: acquiring and monitoring current flow and pressure parameters of the connection section through an SVM algorithm; s5: performing change analysis according to the acquired real-time data, adjusting the division of the first grid to increase the grid density of the area with abnormal change, and obtaining a second grid; s6: and monitoring the parameters of the second grid area in real time, comparing the differences between the data, and carrying out early warning on the shell problem.
According to the scheme, the invention provides a sleeve with a diameter of 3 meters and a diameter of 0.5 meter, the length of the middle section is 2 meters, and the two end connecting sections are respectively 0.5 meter according to the characteristics of the sleeve. Table 1 shows the detection parameters for a certain period of time.
Table 1: and a comparison parameter table.
As can be seen from Table 1, the invention has higher accuracy and smaller fluctuation range in the factors of vibration parameter, temperature parameter, current flow and pressure parameter, etc. in different time periods, and simultaneously, the invention can obtain corresponding feature vector to predict by combining CNN and LSTM network model to perform feature fusion, and SVM algorithm obtains and monitors the current flow and pressure parameter of the connecting section to achieve a certain prediction function, compared with the prediction effect of using common neural network, the comparison table of Table 2 and Table 3 is shown.
Wherein: CNN parameter setting in the invention: a convolution layer with 64 filters, a kernel size of 3x3, a step size of 1, and an activation function of 'ReLU'; a maximum pooling layer, pool size 2x2, step length 2; the convolution layer is provided with 128 filters, the kernel size is 3x3, the step length is 1, and the activation function is ReLU; a maximum pooling layer, pool size 2x2, step length 2; the full connection layer has 1024 neurons, and the activation function is ReLU. Output layer: a fully connected layer, the number of neurons is equal to the number of categories, namely: the activation function uses softmax for a discrete range of current and pressure values. LSTM parameter settings: an LSTM layer with 128 neurons and setting a return sequence as True; LSTM layer has 64 neurons. A fully connected layer, the number of neurons being equal to the number of categories.
Table 2: and predicting comparison data.
Table 3: and (5) comparing effects.
As can be seen from tables 2 and 3, different features of the data can be extracted and understood by feature type combination and selection. CNN can effectively process spatial features of data, such as color change in the present invention, while LSTM can process time-series data to find long-term dependency information. The characteristics of the two networks are combined, and the space and time modes can be simultaneously utilized, so that more accurate and comprehensive prediction information is obtained. The SVM is used for acquiring and monitoring current flow and pressure parameters, so that the state and trend of the machine can be determined, and more accurate prediction can be fed back.
Meanwhile, the LSTM and SVM models can avoid over fitting to a certain extent. LSTM is not easily overfitted due to its special design, being able to process long sequence data; the SVM algorithm particularly effectively avoids overfitting, and an optimal decision boundary is found to separate various data to the greatest extent.
Compared with the common neural network prediction, the method and the device aim at daily monitoring and detection requirements, and can achieve accurate integrity check and a certain prediction function under the condition of reducing data calculation after classifying the data characteristics according to the monitoring parts.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered by the scope of the claims of the present invention.
Claims (10)
1. An on-line monitoring method for the integrity of a converter transformer valve side sleeve shell is characterized by comprising the following steps:
acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
expanding a first grid, and dividing a sleeve middle section and a connecting section according to image recognition;
obtaining and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model;
acquiring and monitoring current flow and pressure parameters of the connection section through an SVM algorithm;
performing change analysis according to the acquired real-time data, and adjusting the division of the first grid, wherein the adjustment is to increase the grid density of the area with abnormal change to obtain a second grid;
and monitoring the parameters of the second grid area in real time, comparing the differences between the data, and carrying out early warning on the shell problem.
2. The on-line monitoring method of converter valve side casing integrity according to claim 1, wherein said establishing a first grid on the casing surface based on dimensions comprises:
taking the height direction of the sleeve as an X axis and taking the direction perpendicular to the height direction as a Y axis;
taking the midpoint of the height direction as the origin of a coordinate system;
the sleeve surface is uniformly divided into rectangular block areas of the same size.
3. The on-line monitoring method for the integrity of a casing on a converter transformer valve side according to claim 1 or 2, wherein the expanding the first grid and dividing the middle section and the connecting section of the casing according to image recognition comprises: expanding the first grid along the Y axis, and if the sleeve structure has a convex structure, continuing to expand along the X axis until the sleeve structure is expanded into a two-dimensional plane;
on the unfolded two-dimensional plane, identifying and determining the area where the convex structure is located according to the original structure of the sleeve, and expanding according to an expanding rule;
the first expansion rule is that the edge of the area where the convex structure is located is continuously expanded by 2 rectangular block areas along the X axis and expanded to two-dimensional plane edge rectangular blocks along the Y axis so as to obtain a first plane, the first plane is marked as a connecting section, and the rest plane parts are middle sections;
the second expansion rule is that if a part of the convex structure has two-dimensional plane edges and cannot be expanded outwards, the expandable edges are expanded according to the first expansion rule to obtain a first plane, the first plane is marked as a joint section, and the rest plane parts are middle sections;
the third expansion rule is that if only one end of the sleeve has a convex structure, after expansion according to the first or second expansion rule, a second plane symmetrical to the first plane is obtained by taking the Y axis as a symmetry axis, and the second plane is marked as a connecting section, and the rest plane parts are middle sections;
if the convex structure does not exist, the Y axis is taken as a symmetry axis, two thirds of the length of the two-dimensional plane are respectively extended to the two ends of the X axis, a third plane is obtained, the mark is a middle section, and the rest plane parts are joint sections.
4. A method for on-line monitoring of converter valve side casing integrity according to claim 3, wherein obtaining and monitoring vibration parameters and temperature parameters of the intermediate section by means of a deep learning model comprises:
acquiring image data of an intermediate section and time sequence data of vibration and temperature parameters, and preprocessing;
extracting image features through a CNN model, wherein the image features comprise color changes, cracks, fine lines and edge contours;
inputting the image characteristics and vibration and temperature parameters into an LSTM network model for characteristic fusion to obtain characteristic vectors combining three data sources;
and inputting the feature vector into the CNN model again for convolutional neural network learning to form a first model, wherein the first model is a neural network model with vibration and temperature prediction.
5. The on-line monitoring method for converter transformer valve side casing integrity according to claim 4, wherein obtaining and monitoring current flow and pressure parameters of the junction segment by an SVM algorithm comprises:
acquiring current flow and pressure parameters of the connection section as a sample set, extracting characteristic values and preprocessing;
extracting current flow characteristics comprises frequency distribution, current mean and variance and current difference value of preset time interval;
extracting pressure characteristics comprises the average value and peak value of pressure and the current difference value of a preset time interval;
the selection range of the preset time interval is 10-20 minutes;
and inputting the extracted characteristic values into an SVM model, taking RBF as a kernel function, performing parameter tuning, and forming a second model through cross verification, wherein the second model is provided with a model of a current value and a shell pressure value.
6. The on-line monitoring method for converter transformer valve side casing integrity according to claim 5, wherein the performing a change analysis according to the acquired real-time data adjusts the division of the first grid, the adjusting to increase the grid density of the region with abnormal change, and obtaining the second grid comprises:
acquiring real-time data through the first model and the second model, and locating a changed area through the feature vector when the acquired vibration value exceeds thirty percent of the historical mean value or a certain temperature value changes beyond the variance within the past half hour;
when the acquired continuous current values or pressure values are continuously increased or decreased, marking as abnormal, and directly positioning the changed area;
when the current value or the pressure value in a certain time period exceeds 50% of the historical average value, marking as abnormal, and directly positioning a changed area;
and increasing the grid density of the change area to 1.5 times of the original density to obtain a second grid.
7. The on-line monitoring method for converter transformer valve side casing integrity according to claim 1 or 6, wherein the second grid region parameters are monitored in real time, and differences between the data are compared to perform fault prediction, comprising:
acquiring parameters of a second grid region, extracting parameters of a change region, inputting the parameters into respective corresponding models for parameter value prediction, and obtaining first prediction data;
taking the average value of the residual errors of the actually monitored data and the first prediction data as a reference, and if the residual errors exceed twice the standard deviation of the average value, carrying out shell problem early warning.
8. An on-line monitoring system for converter transformer valve side casing integrity, comprising:
the first grid construction module is used for acquiring the size and the image of the sleeve, and establishing a first grid on the surface of the sleeve according to the size;
the first dividing module is used for expanding the first grid and dividing the middle section and the connecting section of the sleeve according to image recognition;
the first model monitoring module is used for acquiring and monitoring vibration parameters and temperature parameters of the middle section through a deep learning model;
the second model monitoring module is used for acquiring and monitoring current flow and pressure parameters of the connecting section through an SVM algorithm;
the second dividing module is used for carrying out change analysis according to the acquired real-time data, adjusting the division of the first grid, wherein the adjustment is to increase the grid density of the area with abnormal change to obtain a second grid;
and the early warning module is used for monitoring the parameters of the second grid area in real time, comparing the differences between the data and carrying out early warning on the problems of the shell.
9. An electronic device, comprising:
a memory and a processor;
the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, which when executed by the processor, implement the steps of the method for on-line monitoring of converter valve side casing integrity of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor perform the steps of the method for on-line monitoring of converter valve side casing integrity of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410257398.0A CN117848423B (en) | 2024-03-07 | 2024-03-07 | On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410257398.0A CN117848423B (en) | 2024-03-07 | 2024-03-07 | On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117848423A true CN117848423A (en) | 2024-04-09 |
| CN117848423B CN117848423B (en) | 2024-05-17 |
Family
ID=90546861
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410257398.0A Active CN117848423B (en) | 2024-03-07 | 2024-03-07 | On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117848423B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006132194A1 (en) * | 2005-06-07 | 2006-12-14 | Sony Corporation | Image processing device and image processing method and computer program |
| GB0809588D0 (en) * | 2008-05-27 | 2008-07-02 | Simpleware Ltd | Image processing method |
| CN102866983A (en) * | 2012-08-09 | 2013-01-09 | 同济大学 | Finite difference method of fine simulation tubular well structure |
| CN106887000A (en) * | 2017-01-23 | 2017-06-23 | 上海联影医疗科技有限公司 | The gridding processing method and its system of medical image |
| CN114325274A (en) * | 2022-01-10 | 2022-04-12 | 西南交通大学 | Converter transformer valve side sleeve partial discharge detection system and method |
| CN114638784A (en) * | 2022-02-17 | 2022-06-17 | 中南大学 | Method and device for detecting surface defects of copper pipe based on FE-YOLO |
| CN115471482A (en) * | 2022-09-20 | 2022-12-13 | 重庆理工大学 | Small-caliber container inner wall defect detection method based on computer vision |
-
2024
- 2024-03-07 CN CN202410257398.0A patent/CN117848423B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006132194A1 (en) * | 2005-06-07 | 2006-12-14 | Sony Corporation | Image processing device and image processing method and computer program |
| GB0809588D0 (en) * | 2008-05-27 | 2008-07-02 | Simpleware Ltd | Image processing method |
| CN102866983A (en) * | 2012-08-09 | 2013-01-09 | 同济大学 | Finite difference method of fine simulation tubular well structure |
| CN106887000A (en) * | 2017-01-23 | 2017-06-23 | 上海联影医疗科技有限公司 | The gridding processing method and its system of medical image |
| CN114325274A (en) * | 2022-01-10 | 2022-04-12 | 西南交通大学 | Converter transformer valve side sleeve partial discharge detection system and method |
| CN114638784A (en) * | 2022-02-17 | 2022-06-17 | 中南大学 | Method and device for detecting surface defects of copper pipe based on FE-YOLO |
| CN115471482A (en) * | 2022-09-20 | 2022-12-13 | 重庆理工大学 | Small-caliber container inner wall defect detection method based on computer vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117848423B (en) | 2024-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111210024B (en) | Model training method, device, computer equipment and storage medium | |
| Yang et al. | Remaining useful life prediction based on a double-convolutional neural network architecture | |
| Cai et al. | Wide-area monitoring of power systems using principal component analysis and $ k $-nearest neighbor analysis | |
| WO2020049087A1 (en) | Computer-implemented method, computer program product and system for anomaly detection and/or predictive maintenance | |
| Maurya et al. | Condition monitoring of machines using fused features from EMD-based local energy with DNN | |
| WO2022160413A1 (en) | Electric power production anomaly monitoring method and apparatus, and computer device and storage medium | |
| CN112734692A (en) | Transformer equipment defect identification method and device | |
| CN115358281B (en) | Machine learning-based cold and hot all-in-one machine monitoring control method and system | |
| CN113191429A (en) | Power transformer bushing fault diagnosis method and device | |
| WO2019071438A1 (en) | Method and apparatus for monitoring state of device in process industry and medium | |
| Fang et al. | Fault diagnosis of electric transformers based on infrared image processing and semi-supervised learning | |
| CN113554526A (en) | Fault early warning method and device for power equipment, storage medium and processor | |
| CN115861210B (en) | Transformer substation equipment abnormality detection method and system based on twin network | |
| CN115618719A (en) | Fire smoke monitoring and early warning method and system in industrial building and storage medium | |
| Hauser et al. | Real-time combustion state identification via image processing: A dynamic data-driven approach | |
| CN117848423B (en) | On-line monitoring method, system, equipment and medium for integrity of converter transformer valve side sleeve shell | |
| CN116823233B (en) | User data processing method and system based on full-period operation and maintenance | |
| CN117333821A (en) | Valve base electronic equipment board aging state evaluation method and system based on image recognition | |
| CN115965625B (en) | Instrument detection device based on visual recognition and detection method thereof | |
| Huang et al. | Machining surface roughness detection by adaptive deep fusion capsule network with low illumination and noise robustness | |
| CN110471279B (en) | Vine-copulas-based industrial production simulation scene generator and scene generation method | |
| CN115329493B (en) | Impeller machinery fault detection method based on digital twin model of centrifugal pump | |
| Qi et al. | Rapid transformer health state recognition through canopy cluster-merging of dissolved gas data in high-dimensional space | |
| Lu et al. | Anomaly detection method for substation equipment based on feature matching and multi-Semantic classification | |
| Tosyali et al. | Data-driven gantry health monitoring and process status identification based on texture extraction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |