CN118049367A - Water pump winding state monitoring method, medium and system of drainage pump station - Google Patents
Water pump winding state monitoring method, medium and system of drainage pump station Download PDFInfo
- Publication number
- CN118049367A CN118049367A CN202410190200.1A CN202410190200A CN118049367A CN 118049367 A CN118049367 A CN 118049367A CN 202410190200 A CN202410190200 A CN 202410190200A CN 118049367 A CN118049367 A CN 118049367A
- Authority
- CN
- China
- Prior art keywords
- dynamic image
- water pump
- winding
- noise
- gray
- 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.)
- Pending
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 87
- 238000004804 winding Methods 0.000 title claims abstract description 72
- 238000012544 monitoring process Methods 0.000 title claims abstract description 68
- 238000000034 method Methods 0.000 title claims abstract description 65
- 230000004927 fusion Effects 0.000 claims abstract description 41
- 238000013507 mapping Methods 0.000 claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000004458 analytical method Methods 0.000 claims abstract description 8
- 230000008859 change Effects 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 10
- 230000009466 transformation Effects 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 230000001133 acceleration Effects 0.000 claims description 5
- 238000009499 grossing Methods 0.000 claims description 5
- 238000005086 pumping Methods 0.000 claims description 5
- 230000003287 optical effect Effects 0.000 claims description 4
- 238000009434 installation Methods 0.000 claims description 3
- 230000016507 interphase Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 2
- 238000006073 displacement reaction Methods 0.000 claims 6
- 230000009286 beneficial effect Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000012549 training Methods 0.000 description 8
- 238000000605 extraction Methods 0.000 description 7
- 230000006872 improvement Effects 0.000 description 7
- 238000013527 convolutional neural network Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 4
- 238000007499 fusion processing Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000002604 ultrasonography Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000000704 physical effect Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000007789 sealing Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000009440 infrastructure construction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000010865 sewage Substances 0.000 description 1
- 238000004901 spalling Methods 0.000 description 1
- 230000035882 stress Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000008646 thermal stress Effects 0.000 description 1
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention provides a water pump winding state monitoring method, medium and system of a drainage pump station, belonging to the technical field of drainage pump stations, wherein the water pump winding state monitoring method of the drainage pump station comprises the following steps: acquiring water pump operation monitoring data, preprocessing to obtain preprocessed monitoring data, and establishing temperature, noise and vibration dynamic images; carrying out gray scale processing to obtain temperature, noise and vibration gray scale dynamic images, and registering to form a fusion dynamic image; establishing a fusion dynamic image, wherein the size and the resolution of the fusion dynamic image are consistent with those of three registered dynamic images, and R value, G value and B value are respectively mapping values of gray values of pixel points corresponding to the temperature, noise and vibration gray dynamic images; marking the ultrasonic wave position on the fusion dynamic image to form the fusion ultrasonic dynamic image; inputting a fused ultrasonic dynamic image by using a water pump winding state analysis model to obtain the winding state of the water pump to be detected; the invention can efficiently judge the winding state.
Description
Technical Field
The invention belongs to the technical field of drainage pumping stations, and particularly relates to a water pump winding state monitoring method, medium and system of a drainage pumping station.
Background
Along with the rapid development of urban infrastructure construction, various drainage pump stations are widely adopted, penetrate through urban underground pipe networks and bear the pumping tasks of rainwater, domestic sewage and the like. The long-time continuous running water pump winding is easy to generate faults under the actions of thermal stress, current voltage and mechanical stress. Therefore, monitoring the operation state of the water pump of the pump station is important to ensure the reliability of the drainage system.
In the prior art, the running state is determined through monitoring the load of the water pump motor, and when the detected current parameter is higher than the protection value, the fault state is determined. The method only monitors the electrical parameters, and cannot observe the physical state change of the water pump winding faults. Another method measures winding temperature by means of a temperature sensor, but the fault pattern cannot be determined exactly for temperature anomalies. The single parameter monitoring method has poor fault type discrimination capability.
In the prior art, the fault state is comprehensively judged by integrating multiple parameters such as temperature, vibration and the like, but the sensing parameters lack of a synchronous data integration processing mechanism, and the judgment is based on the lack of objective quantification standards. The multisource heterogeneous monitoring data are difficult to effectively fuse, and the accuracy of fault diagnosis is reduced. In addition, the establishment of the judgment rule depends on subjective judgment of expert experience, the reproducibility is poor, and popularization and application are difficult.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a water pump winding state monitoring method, medium and system of a drainage pump station.
The invention is realized in the following way:
The first aspect of the invention provides a water pump winding state monitoring method of a drainage pump station, which comprises the following steps:
s10, continuously acquiring operation monitoring data of a water pump to be tested, wherein the operation monitoring data comprise temperature, noise, vibration and ultrasonic waves;
S30, preprocessing the monitoring data to obtain preprocessed monitoring data;
s30, establishing a temperature dynamic image, a noise dynamic image and a vibration dynamic image according to the preprocessing monitoring data;
s40, carrying out gray scale processing on the temperature dynamic image, the noise dynamic image and the vibration dynamic image to obtain a temperature gray scale dynamic image, a noise gray scale dynamic image and a vibration gray scale dynamic image;
s50, registering the temperature gray scale dynamic image, the noise gray scale dynamic image and the vibration gray scale dynamic image to ensure that the sizes and the resolutions of the three dynamic images are consistent;
s60, establishing a fusion dynamic image, wherein the size and the resolution of the fusion dynamic image are consistent with those of the temperature gray dynamic image after registration, the R value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the temperature gray dynamic image, the G value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the noise gray dynamic image, and the B value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the vibration gray dynamic image;
s70, marking the position of the ultrasonic wave on the fusion dynamic image according to the time and the position acquired by the ultrasonic wave to form the fusion ultrasonic dynamic image;
S80, inputting the fused ultrasonic dynamic image by adopting a pre-trained water pump winding state analysis model to obtain the winding state of the water pump to be tested, wherein the winding state of the water pump comprises the following steps: normal operation, winding turn-to-turn discharge, winding fusion short circuit, winding interphase short circuit, winding open circuit, winding grounding.
On the basis of the technical scheme, the water pump winding state monitoring method of the drainage pump station can be improved as follows:
the step of acquiring the operation monitoring data of the water pump to be tested comprises the following steps:
a plurality of temperature sensors are arranged and are in good heat conduction connection with the water pump winding wire, and the temperature change of the winding is detected in real time;
A plurality of noise sensors are arranged, and the operation noise information of the water pump is collected by adopting a mechanical vibration collection principle;
setting a plurality of vibration sensors, and collecting mechanical vibration information of the water pump through the acceleration sensors;
a plurality of ultrasonic sensors are arranged, and the positions of objects in the water pump are positioned by transmitting and receiving ultrasonic waves.
The beneficial effects of adopting above-mentioned improvement scheme are: the purpose of this step is to provide the required temperature, noise, vibration, and ultrasound multisource heterogeneous sensing data for subsequent monitoring.
Further, the step of preprocessing the monitoring data includes: removing abnormal values in the temperature data, filtering to remove external noise in the noise data, calibrating to remove installation errors in the vibration data, and mapping to obtain three-dimensional coordinates of the object in the ultrasonic data.
The beneficial effects of adopting above-mentioned improvement scheme are: the aim of the step is to remove noise and error components in the monitoring data, obtain accurate and reliable preprocessed monitoring data, and provide a reliable data source for subsequent analysis establishment.
Further, the step of creating a dynamic image includes: and respectively converting the temperature, noise and vibration data into gray dynamic images by adopting an encoding method.
The beneficial effects of adopting above-mentioned improvement scheme are: the purpose of this step is to uniformly convert different types of monitoring data into the form of a gray-scale dynamic image, and provide a data basis for subsequent image processing and analytical modeling.
Further, the step of performing gradation processing on the moving image includes: and normalizing the contrast of the temperature gray scale dynamic image, filtering to remove low-level noise of the noise image, and smoothing to treat random disturbance of the vibration image.
The beneficial effects of adopting above-mentioned improvement scheme are: the method aims to enhance the image quality of three dynamic images through an image processing method, improve the contrast and the identifiability of key information and provide an input image with better quality for carrying out feature extraction and state identification for subsequent image analysis modeling.
Further, the step of registering the images includes: the three gray scale dynamic images are subjected to strict geometric and optical registration by adopting a registration algorithm, a scaling method and matrix transformation.
The beneficial effects of adopting above-mentioned improvement scheme are: the method aims at enabling three gray scale dynamic images from different sources to achieve strict geometric and optical registration, and provides a basis for subsequent multi-source heterogeneous information fusion processing.
Further, the step of creating a fused image includes: initializing RGB images with the same resolution, traversing pixels, reading three image gray values, and respectively mapping and assigning the three image gray values to three color components of the RGB images.
The beneficial effects of adopting above-mentioned improvement scheme are: the method aims at constructing a dynamic image fused with temperature, noise and vibration three-dimensional information, and provides rich feature descriptors for image discrimination modeling of the water pump state.
Further, the step of obtaining a fused ultrasound dynamic image includes: and calculating data coordinates of ultrasonic positioning, and projecting the data coordinates to the fusion RGB image for visual marking.
The beneficial effects of adopting above-mentioned improvement scheme are: the purpose of this step is to project the positioning information of the ultrasonic wave under the image coordinate system, and realize the multi-source data fusion of counterpoint with other sensing information, so as to provide input data for the subsequent state detection task.
A second aspect of the present invention provides a computer readable storage medium, wherein the computer readable storage medium stores program instructions, and the program instructions are used to execute the water pump winding state monitoring method of the water pump station when the program instructions are executed.
A third aspect of the present invention provides a water pump station water pump winding condition monitoring system, comprising the computer readable storage medium described above.
The technical scheme of the invention adopts a plurality of heterogeneous parameters to synchronously collect and monitor the state of the water pump, and more comprehensively observe the change process of the physical effect of the winding. The temperature, noise and vibration data are mapped into dynamic gray images through an image processing technology. And creatively projecting and labeling the ultrasonic positioning result on the image. And realizing the data fusion of physical data and geometric space information alignment.
The effective coordination of the multi-source heterogeneous data is realized based on the operations of image registration, channel coding and the like, and the three-channel fusion image with rich information content is constructed. Compared with the prior art, the method simply superimposes the element signals, generates the fusion image as the judgment basis of state identification, provides the dynamic characterization of more sufficient and integrated space dimension and time dimension, and is more beneficial to the judgment of the physical condition of the winding.
Meanwhile, the scheme adopts a pretrained convolutional neural network model to realize image feature extraction and classification judgment. And judging the difference of the images in different fault states by utilizing the strong fitting capacity of the deep learning model. The judgment rule is obtained through a large amount of data training, and the defect that the prior art relies on subjective experience is overcome. Compared with the traditional machine learning algorithm, the deep learning method is more efficient and accurate. Compared with the prior art, the invention has obvious technical effects in the aspects of fault diagnosis accuracy, discrimination subdivision capability, automation level and the like.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention 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 flow chart of a method for monitoring the condition of a water pump winding of a drainage pump station.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
Referring to fig. 1, a flowchart of a first embodiment of a method for monitoring a water pump winding state of a drainage pump station according to a first aspect of the present invention includes the following steps:
s10, continuously acquiring operation monitoring data of a water pump to be tested, wherein the operation monitoring data comprise temperature, noise, vibration and ultrasonic waves;
S30, preprocessing the monitoring data to obtain preprocessed monitoring data;
S30, establishing a temperature dynamic image, a noise dynamic image and a vibration dynamic image according to the preprocessing monitoring data;
s40, carrying out gray scale processing on the temperature dynamic image, the noise dynamic image and the vibration dynamic image to obtain a temperature gray scale dynamic image, a noise gray scale dynamic image and a vibration gray scale dynamic image;
S50, registering the temperature gray scale dynamic image, the noise gray scale dynamic image and the vibration gray scale dynamic image to ensure that the sizes and the resolutions of the three dynamic images are consistent;
s60, establishing a fusion dynamic image, wherein the size and the resolution of the fusion dynamic image are consistent with those of the registered temperature gray dynamic image, the R value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the temperature gray dynamic image, the G value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the noise gray dynamic image, and the B value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the vibration gray dynamic image;
s70, marking the position of the ultrasonic wave on the fusion dynamic image according to the time and the position acquired by the ultrasonic wave to form the fusion ultrasonic dynamic image;
s80, inputting a fusion ultrasonic dynamic image by adopting a pre-trained water pump winding state analysis model to obtain the winding state of the water pump to be tested, wherein the winding state of the water pump comprises the following steps: normal operation, winding turn-to-turn discharge, winding fusion short circuit, winding interphase short circuit, winding open circuit, winding grounding.
The specific implementation manner of step S10 is:
(1) A plurality of temperature sensors are arranged, and the temperature sensors are in good heat conduction connection with the water pump winding metal wire through thermocouple wires, so that the temperature change of the winding can be detected in real time. The common temperature sensor can select a PT100 chip with higher precision and quick response time for detection;
(2) A plurality of noise sensors are arranged, and adopt a mechanical vibration collection principle, so that the noise sensors can collect various noise information in the operation process of the water pump through good mechanical connection. The commonly used noise sensor can be a microphone sensor with high sensitivity;
(3) A plurality of vibration sensors are arranged, and the vibration sensors collect mechanical vibration information of the water pump through the acceleration sensor. The common vibration sensor can be a MEMS chip vibration sensor with high sensitivity and wide frequency response range;
(4) A plurality of ultrasonic sensors are arranged, and the ultrasonic sensors can position an object in the water pump by emitting and receiving ultrasonic waves. The commonly used ultrasonic sensor can select a pulse ultrasonic module with high precision and accurate positioning.
The purpose of this step is to provide the required temperature, noise, vibration, and ultrasound multisource heterogeneous sensing data for subsequent monitoring.
The specific implementation manner of step S20 is:
(1) And (3) performing abnormal value removal processing on the temperature data, and eliminating obvious error data caused by sensor connection and the like. The common method for removing the abnormal value can adopt a 3 delta rule;
(2) And filtering the noise data to remove noise components introduced by other external noise sources and retain the running noise of the water pump. The common filtering method can adopt wavelet transformation to carry out smoothing treatment;
(3) And calibrating the vibration data, eliminating deviation caused by the position and the installation error of the sensor, and obtaining standardized data. The common calibration method can adopt a multi-sensor information fusion algorithm for processing;
(4) And carrying out positioning mapping on the ultrasonic data to obtain the accurate position of the object in the water pump in the three-dimensional space. The common positioning mapping method can adopt the triangulation principle to combine the quantity and the position of the sensors for data fusion processing.
The aim of the step is to remove noise and error components in the monitoring data, obtain accurate and reliable preprocessed monitoring data, and provide a reliable data source for subsequent analysis establishment.
The specific implementation manner of step S30 is:
(1) And converting the temperature monitoring data into a gray dynamic image by adopting a gray coding method, wherein the temperature value is mapped into a gray range, and the gray value corresponding to the higher temperature value is higher. The common coding method can adopt linear mapping relation to carry out simple conversion;
(2) And converting the noise monitoring data into a gray dynamic image by adopting a sound pressure coding method, wherein the noise decibel value is mapped into a gray range, and the gray value corresponding to the higher noise value is higher. The common coding method can also adopt linear mapping relation to carry out simple conversion;
(3) And converting the vibration monitoring data into a gray dynamic image by adopting an acceleration coding method, wherein the absolute value of the vibration acceleration is mapped into a gray range, and the larger the vibration value is, the higher the corresponding gray value is. The encoding method may also refer to a linear mapping of temperature and noise.
The purpose of this step is to uniformly convert different types of monitoring data into the form of a gray-scale dynamic image, and provide a data basis for subsequent image processing and analytical modeling.
The specific implementation manner of step S40 is:
(1) And carrying out normalization processing on the temperature gray scale dynamic image, so that the temperature value is mapped to the whole gray scale range of 0-255, and the image contrast is enhanced. The normalization method can adopt a minimum and maximum normalization algorithm, and occupies the whole gray scale range through linear mapping;
(2) And (3) carrying out threshold filtering on the noise gray scale dynamic image, removing low-level noise corresponding to slow gray scale change, and retaining sudden high-noise events corresponding to severe gray scale change. The common threshold filtering method can adopt a median filtering rule for processing;
(3) And the vibration gray scale dynamic image is subjected to smoothing processing, so that random disturbance caused by noise of the sensor is reduced. A common smoothing method may use gaussian filtering to smooth the varying gray scale curve.
The method aims to enhance the image quality of three dynamic images through an image processing method, improve the contrast and the identifiability of key information and provide an input image with better quality for carrying out feature extraction and state identification for subsequent image analysis modeling.
The specific implementation manner of step S50 is as follows:
(1) And performing geometric registration on the three gray dynamic images by adopting an image registration algorithm to ensure that the three gray dynamic images are overlapped under an image coordinate system. The common registration algorithm can select a matching method based on feature points, estimate a transformation model by extracting SIFT feature points on an image and calculating the distance of feature descriptors, eliminate mismatching by using a RANSAC method, and finally map image coordinates to a unified coordinate system by using the transformation model;
(2) And processing the three gray scale dynamic images after geometric registration into the same resolution by adopting an image scaling method. The common scaling method can select bicubic interpolation method, and can effectively keep more original image information. The target resolution may match the lowest resolution image in the dataset;
(3) Checking whether the three images are exactly coincident in the spatial coordinate system and resolution, if necessary, a matrix transformation can be used to fine tune the image position to eliminate any possible bias.
The method aims at enabling three gray scale dynamic images from different sources to achieve strict geometric and optical registration, and provides a basis for subsequent multi-source heterogeneous information fusion processing.
The specific implementation manner of step S60 is:
(1) Initializing an RGB image with the same size and resolution as the three registered gray dynamic images, wherein the color space of the image is 24-bit true color;
(2) Traversing all pixel coordinates of the image, and reading gray values of corresponding coordinates on the three gray dynamic images;
(3) Linearly mapping the gray value of the temperature gray dynamic image and assigning the gray value to the R component of the pixel of the current RGB image;
(4) Linearly mapping and assigning the gray value of the noise image to the G component of the pixel of the current RGB image;
(5) Linearly mapping and assigning the gray value of the vibration image to the B component of the pixel of the current RGB image;
(6) Repeating the steps (2) - (5) until the gray value information fused with the three sources is allocated for all pixels of the RGB image.
Specifically, an RGB image I RGB having the same size h×w and resolution as the three gray-scale moving images after registration is initialized:
wherein the image color space is 24-bit true color, H represents the image height (number of pixels), W represents the image width (number of pixels), Representing the product of the red channels of all pixels.
All pixel coordinates of the image are traversed (x, y), where x=1, 2, W; y=1, 2,..h. And reading gray values I T(x,y)、IN(x,y)、IV (x, y) of corresponding coordinates (x, y) on the temperature gray dynamic image, the noise gray dynamic image and the vibration gray dynamic image. Linear mapping of the read gray values into the range of [0,255 ]:
gray value of temperature gray dynamic image after mapping:
The gray value of the noise gray dynamic image after mapping:
vibration gray scale dynamic image gray scale value after mapping:
Assigning the mapped gray values to color channels of corresponding coordinates (x, y) on the RGB image:
IRGB(x,y,R)=I′T(x,y);
IRGB(x,y,G)=I′N(x,y);
IRGB(x,y,B)=I′V(x,y);
The traversal process is repeated until a color fused with the source image information is generated for all pixel coordinates of I RGB.
The method aims at constructing a dynamic image fused with temperature, noise and vibration three-dimensional information, and provides rich feature descriptors for image discrimination modeling of the water pump state.
The specific implementation manner of step S70 is:
(1) Reading positioning data acquired by an ultrasonic sensor, wherein the positioning data comprises information such as ultrasonic transmitting and receiving moments, sensor positions and the like;
(2) According to the propagation characteristics of the ultrasonic waves, calculating the three-dimensional coordinates of the internal target position of the water pump generated by the reflected waves;
(3) Projecting the target position to an image coordinate system of the fusion RGB dynamic image constructed in the previous step;
(4) In the RGB image coordinate system, the corresponding target position is marked and visualized by means of color mapping.
Specifically, the positioning data of the target k acquired by the ultrasonic sensor is read, which includes the transmitting and receiving time t k and the space coordinates (x k,yk,zk) of the ultrasonic sensor. Calculating the spatial position coordinates (x' k,y′k,z′k) of the object k according to the propagation velocity v of the ultrasonic wave in the medium:
x′k=xk+v×tk×cosα×cosβ;
y′k=yk+v×tk×cosα×sinβ;
z′k=zk+v×tk×sinα;
Wherein alpha and beta are the azimuth angle and the pitch angle of the ultrasonic signals. The pixel coordinates of object k on image I RGB are calculated by projective transformation (u k,vk). Color coding marking is carried out at the coordinates (u k,vk) to generate a fusion ultrasonic dynamic image.
The purpose of this step is to project the positioning information of the ultrasonic wave under the image coordinate system, and realize the multi-source data fusion of counterpoint with other sensing information, so as to provide input data for the subsequent state detection task.
The specific implementation modes of the establishment and training of the water pump winding state analysis model are as follows:
(1) Collecting a water pump fusion ultrasonic dynamic image containing each winding state, and collecting 2 ten thousand images in each state;
(2) The data set is divided into a training data set and a verification data set according to the proportion, and the proportion of the training data set is 80%;
(3) A convolutional neural network model is constructed, which comprises 1 convolutional layer and 2 fully connected layers. The input image size is 224 x 224 pixels;
(4) The number of the convolution layer filters is 32, the size is 3*3, the activation function adopts a ReLU, and the pooling mode is 2 x 2 max pooling;
(5) The first full-connection layer is 128 nodes in size, and the activation function adopts a ReLU;
(6) The second full-connection layer is 6 nodes in size, the activation function adopts Softmax, and 6 classification labels are corresponding to the activation function;
(7) Adopting an Adam algorithm optimizer, wherein the initial value of the learning rate is 0.001;
(8) The maximum iteration number of model training is 30, and model parameters are trained by using a training data set;
(9) Each training 1 round evaluates the model by using the verification data set, and records the loss function value;
(10) Selecting a model with the minimum verification loss function as a final model;
(11) The softmax probability values of the 6 output nodes of the final model are the predicted probabilities of the identified water pump states.
The model aims to serve as an end-to-end image classification method, performs feature extraction and state classification on the input fused ultrasonic dynamic image, and outputs a result to a water pump state monitoring system so as to accurately monitor the state of a water pump winding.
As shown in fig. 1, a flow chart of a second embodiment of a water pump winding status monitoring method of a drainage pump station according to a first aspect of the present invention includes the following steps:
The water pump winding state monitoring system is arranged in a water drainage pump station in a certain city. 4 temperature sensors are arranged on the system and are arranged on the surface of a water inlet and outlet pipeline of 2 driving water pumps, and the measuring range is 0-100 ℃;6 vibration sensors are arranged on the surface of the water pump shell, and the measuring range is 0-20m/s 2; 8 noise sensors are arranged around the inner periphery of the pump station, and the measuring range is 30-120 dB; and 2 ultrasonic sensors, the emission frequency is 40kHz, and the range is 0.2-2 meters. The sensor collects real-time data and sends the data to the monitoring host.
In the test process, after the water pump A runs for 5 hours, the monitoring platform detects that the measured value of the 3 rd path of temperature sensor is gradually increased, and other temperature sensors have no obvious change. In the same period, the effective values (RMS) of the 2 vibration sensors corresponding to the surface of the pump body have larger fluctuation which is higher than 0.2m/s 2 in normal operation and reach 0.8m/s 2. Noise and ultrasonic waves are not obviously abnormal. And judging that the local area of the water pump A winding is over-heated through analysis, and judging that the turn-to-turn short circuit fault of the water pump A winding is caused based on the temperature and the vibration signal.
In order to verify the judgment result, the water pump A is overhauled and maintained, and the disassembly finds that 2 adjacent coils of the stator winding are burnt and shorted, and the judgment is consistent with the judgment of the monitoring system. And finally, after the fault winding is replaced, the water pump is operated for 5 hours again, and the reading values of the sensors are recovered to be normal, so that the application inspection of the system is completed.
During maintenance, the water pump B continues to operate normally. After the system is operated for 12 hours, no abnormality is detected in the temperature of the water pump B, the effective value of the vibration signal is maintained at 0.15m/s 2, and the ultrasonic signal shows no obvious change in the position of the rotor in the operation process. But 2 detected sound pressure levels in 8 noise sensors are increased in short time to more than 100 dB, and other noise signals are normal. Therefore, the system judges that the water pump B is increased in noise caused by impeller spalling, and normal operation is not affected.
When the water pump B is inspected, an maintainer does not find the abrasion condition of the rotor, and preliminarily judges that the impeller has no fault sign. However, after the impeller is disassembled, the sealing ring material at the inner side of the impeller is found to be damaged and fall off, and the characteristics of the sealing ring material are consistent with noise abnormal signals. The system can accurately position the internal faults of the impeller, and the strong detection and diagnosis capability of the monitoring system is verified.
Through continuous monitoring for a plurality of months, the average fault detection rate of the system reaches 95%, and the false alarm rate is 3%. Compared with the method of manual experience judgment, the system realizes full-automatic monitoring and intelligent diagnosis for 7×24 hours. Remarkably lightens the overhaul working intensity and greatly improves the efficiency of fault prediction and positioning. The system provides a reliable solution for health state evaluation of other similar pump station equipment.
A second aspect of the present invention provides a computer readable storage medium, where the computer readable storage medium stores program instructions, and the program instructions are configured to execute the method for monitoring a water pump winding state of a drainage pump station.
A third aspect of the present invention provides a water pump station water pump winding condition monitoring system, comprising the computer readable storage medium described above.
The principle of constructing a convolutional neural network model for identifying the state of a water pump winding in the technical scheme of the invention is as follows: the convolutional neural network extracts the spatial domain and frequency domain characteristics of the input image through the convolutional layer, the network parameters are continuously updated and extracted to extract the main distinguishing characteristics of the images in different states in the training process, the characteristics which are significant for state identification are reserved, and unnecessary characteristics are removed. And the full connection layer performs nonlinear combination on the features, highly compresses the information into state category information, and realizes end-to-end feature extraction and multi-classification tasks.
Compared with the traditional method which relies on a preset extraction rule, the convolutional neural network can autonomously learn a decision surface for feature extraction and classification judgment. And obtaining stronger supervision information under the drive of a large number of samples, and automatically inducing a decision mode from the data. Meanwhile, the backward propagation algorithm based on gradient descent can effectively optimize the loss function, adjust parameters of the rolling and full-connection layers, and continuously enhance the fitting and distinguishing capabilities of the network. The convolutional neural network can fundamentally solve the problem that the prior art depends on subjective experience.
By constructing a deep network structure, the number of network layers is increased, and the neural network has the capability of fitting various complex nonlinear functions. The characteristic dimension in the method is fully developed, the information loss is reduced to the minimum, and the method is more sensitive to the tiny change of the physical effect of the winding, so that the difference mapping of the images under different fault types to the category is more accurately obtained, and the accurate identification of the state is realized. The technical scheme of the invention is a principle basis of higher efficiency and accuracy than the prior art.
Claims (10)
1. A water pump winding state monitoring method of a drainage pump station is characterized by comprising the following steps:
s10, continuously acquiring operation monitoring data of a water pump to be tested, wherein the operation monitoring data comprise temperature, noise, vibration and ultrasonic waves;
S30, preprocessing the monitoring data to obtain preprocessed monitoring data;
s30, establishing a temperature dynamic image, a noise dynamic image and a vibration dynamic image according to the preprocessing monitoring data;
s40, carrying out gray scale processing on the temperature dynamic image, the noise dynamic image and the vibration dynamic image to obtain a temperature gray scale dynamic image, a noise gray scale dynamic image and a vibration gray scale dynamic image;
s50, registering the temperature gray scale dynamic image, the noise gray scale dynamic image and the vibration gray scale dynamic image to ensure that the sizes and the resolutions of the three dynamic images are consistent;
s60, establishing a fusion dynamic image, wherein the size and the resolution of the fusion dynamic image are consistent with those of the temperature gray dynamic image after registration, the R value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the temperature gray dynamic image, the G value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the noise gray dynamic image, and the B value of each pixel point on the fusion dynamic image is a mapping value of the gray value of the pixel point corresponding to the vibration gray dynamic image;
s70, marking the position of the ultrasonic wave on the fusion dynamic image according to the time and the position acquired by the ultrasonic wave to form the fusion ultrasonic dynamic image;
S80, inputting the fused ultrasonic dynamic image by adopting a pre-trained water pump winding state analysis model to obtain the winding state of the water pump to be tested, wherein the winding state of the water pump comprises the following steps: normal operation, winding turn-to-turn discharge, winding fusion short circuit, winding interphase short circuit, winding open circuit, winding grounding.
2. The method for monitoring the condition of a water pump winding of a water pump station according to claim 1, wherein the step of acquiring operation monitoring data of the water pump to be tested comprises:
setting a plurality of temperature sensors for detecting the temperature change of the winding in real time;
A plurality of noise sensors are arranged, and the operation noise information of the water pump is collected by adopting a mechanical vibration collection principle;
setting a plurality of vibration sensors, and collecting mechanical vibration information of the water pump through the acceleration sensors;
a plurality of ultrasonic sensors are arranged, and the positions of objects in the water pump are positioned by transmitting and receiving ultrasonic waves.
3. A method of monitoring the condition of a water pump winding in a displacement pump station according to claim 2, wherein the step of preprocessing the monitored data comprises: removing abnormal values in the temperature data, filtering to remove external noise in the noise data, calibrating to remove installation errors in the vibration data, and mapping to obtain three-dimensional coordinates of the object in the ultrasonic data.
4. A method of monitoring the condition of a water pump winding in a displacement pump station according to claim 3, wherein the step of creating a dynamic image comprises: and respectively converting the temperature, noise and vibration data into gray dynamic images by adopting an encoding method.
5. The method for monitoring the state of a water pump winding of a drainage pumping station according to claim 4, wherein the step of gray-scale processing the dynamic image comprises the steps of: and normalizing the contrast of the temperature gray scale dynamic image, filtering to remove low-level noise of the noise image, and smoothing to treat random disturbance of the vibration image.
6. A method of monitoring the condition of a water pump winding in a displacement pump station according to claim 5, wherein the step of registering the images comprises: the three gray scale dynamic images are subjected to strict geometric and optical registration by adopting a registration algorithm, a scaling method and matrix transformation.
7. A method of monitoring the condition of a water pump winding in a displacement pump station according to claim 6, wherein the step of creating a fused image comprises: initializing RGB images with the same resolution, traversing pixels, reading three image gray values, and respectively mapping and assigning the three image gray values to three color components of the RGB images.
8. The method for monitoring the state of a water pump winding of a drainage pumping station according to claim 7, wherein the step of obtaining a fused ultrasonic dynamic image comprises: and calculating data coordinates of ultrasonic positioning, and projecting the data coordinates to the fusion RGB image for visual marking.
9. A computer readable storage medium having stored therein program instructions which, when executed, are adapted to carry out a method of monitoring the condition of a water pump winding of a displacement pump station according to any one of claims 1 to 8.
10. A water pump winding condition monitoring system for a displacement pump station, comprising the computer readable storage medium of claim 9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410190200.1A CN118049367A (en) | 2024-02-21 | 2024-02-21 | Water pump winding state monitoring method, medium and system of drainage pump station |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410190200.1A CN118049367A (en) | 2024-02-21 | 2024-02-21 | Water pump winding state monitoring method, medium and system of drainage pump station |
Publications (1)
Publication Number | Publication Date |
---|---|
CN118049367A true CN118049367A (en) | 2024-05-17 |
Family
ID=91049743
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410190200.1A Pending CN118049367A (en) | 2024-02-21 | 2024-02-21 | Water pump winding state monitoring method, medium and system of drainage pump station |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN118049367A (en) |
-
2024
- 2024-02-21 CN CN202410190200.1A patent/CN118049367A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Lei et al. | New crack detection method for bridge inspection using UAV incorporating image processing | |
CN111094956B (en) | Processing thermal imaging images with neural networks to identify subsurface erosion on insulation (CUI) | |
CN113592828B (en) | Nondestructive testing method and system based on industrial endoscope | |
Nayyeri et al. | Foreground–background separation technique for crack detection | |
CN110473187B (en) | Object-oriented line scanning three-dimensional pavement crack extraction method | |
CN115187527B (en) | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum | |
CN114445366A (en) | Intelligent long-distance pipeline radiographic image defect identification method based on self-attention network | |
CN113075065B (en) | Deep sea pipeline crack propagation monitoring and reliability evaluation system based on image recognition | |
CN106951863B (en) | Method for detecting change of infrared image of substation equipment based on random forest | |
CN113436157A (en) | Vehicle-mounted image identification method for pantograph fault | |
CN116704733B (en) | Aging early warning method and system for aluminum alloy cable | |
CN115751203A (en) | Natural gas pipeline leakage monitoring system based on thermal infrared imager | |
CN110660049A (en) | Tire defect detection method based on deep learning | |
CN113530850B (en) | Centrifugal pump fault diagnosis method based on ESA and stacked capsule self-encoder | |
CN112734637B (en) | Thermal infrared image processing method and system for monitoring temperature of lead | |
CN118049367A (en) | Water pump winding state monitoring method, medium and system of drainage pump station | |
CN112525923A (en) | Shared juice platform pipeline inner wall detection method and system and readable storage medium | |
CN116824479A (en) | Intelligent inspection system for factory | |
CN113781513B (en) | Leakage detection method and system for water supply pipeline of power plant | |
CN115375925A (en) | Underwater sonar image matching algorithm based on phase information and deep learning | |
Dunphy et al. | Autonomous crack detection approach for masonry structures using artificial intelligence | |
Harshini et al. | Sewage Pipeline Fault Detection using Image Processing | |
Shin et al. | Visualization for explanation of deep learning-based defect detection model using class activation map | |
CN117593300B (en) | PE pipe crack defect detection method and system | |
CN113591251B (en) | Equipment fault temperature analysis and diagnosis method |
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 |