CN113820017A - Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method - Google Patents
Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method Download PDFInfo
- Publication number
- CN113820017A CN113820017A CN202110956798.7A CN202110956798A CN113820017A CN 113820017 A CN113820017 A CN 113820017A CN 202110956798 A CN202110956798 A CN 202110956798A CN 113820017 A CN113820017 A CN 113820017A
- Authority
- CN
- China
- Prior art keywords
- carbon brush
- temperature
- generator
- data
- infrared image
- 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.)
- Withdrawn
Links
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 title claims abstract description 152
- 229910052799 carbon Inorganic materials 0.000 title claims abstract description 152
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012544 monitoring process Methods 0.000 title claims abstract description 38
- 238000013528 artificial neural network Methods 0.000 claims abstract description 45
- 238000013523 data management Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 22
- 230000015654 memory Effects 0.000 claims description 20
- 238000003062 neural network model Methods 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 14
- 230000005284 excitation Effects 0.000 claims description 12
- 238000011156 evaluation Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 230000008901 benefit Effects 0.000 claims description 10
- 238000012360 testing method Methods 0.000 claims description 9
- 238000009529 body temperature measurement Methods 0.000 claims description 7
- 238000009826 distribution Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 claims description 4
- 238000013439 planning Methods 0.000 claims description 4
- 238000003708 edge detection Methods 0.000 claims description 3
- 238000010438 heat treatment Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 2
- 238000013500 data storage Methods 0.000 abstract description 6
- 238000011161 development Methods 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003331 infrared imaging Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J2005/0077—Imaging
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Radiation Pyrometers (AREA)
Abstract
The generator carbon brush temperature monitoring system based on the infrared image is based on industrial computer end and mobile end platform development monitoring software, and multi-platform remote temperature monitoring is realized by utilizing a 4G network; the cloud server and the local data management system are utilized to access data into the cloud database and the local database to realize data storage and management; the method for predicting the temperature of the carbon brush of the generator comprises the following steps: the method comprises the steps of collecting on-site infrared image data by using an on-line monitoring system of the carbon brush temperature of the generator, processing to obtain carbon brush temperature data, predicting and analyzing the development trend of the carbon brush temperature data by using a long-time memory-feedforward LSTM-BP neural network combined model, fusing infrared image characteristic information and carbon brush temperature information, improving the prediction precision of the carbon brush temperature data by using the long-time memory-feedforward neural network combined model, and realizing the prediction of the development trend of the carbon brush temperature.
Description
Technical Field
The invention belongs to the technical field of municipal public facility monitoring, and particularly relates to a generator carbon brush temperature monitoring system and a temperature prediction method based on infrared images.
Background
The electric power is related to the national civilization, and in 2018, the installed capacity of thermal power generating units in China reaches 11 hundred million kilowatts, and the thermal power generating units account for 70 percent of the total capacity. By the end of 2019, the installed capacity of power generation in China is expected to break through 20 hundred million kilowatts for the first time, and the total installed capacity stably occupies the first world. With the increasing of installed capacity, the safety problem of the system is closely concerned, and the reliability of the generator which is one of three main machines of the thermal power generating unit has a great influence on the operation of the thermal power generating unit. The excitation system is the key of safe and stable operation of the generator set.
The carbon brush and the slip ring are used as important components of a synchronous generator excitation system and are weak links of the generator operation. The fault statistics of the generator show that: the faults of the carbon brushes and the slip rings belong to multiple faults, when the carbon brushes jump, the contact resistance of the carbon brushes is increased, the unevenness of current distribution among the carbon brushes is increased, when the current in the carbon brushes exceeds 80A, a series of problems of unstable excitation voltage and current, system power fluctuation and the like can be generated, the heat effect of the current can further burn the carbon brushes, the temperature of the carbon brushes and the slip rings of the generator is abnormally increased, and in severe cases, the ring fire can even be caused, so that the slip rings of the generator are burnt out. Therefore, the quality of the carbon brush directly influences the production safety of the generator, a set of carbon brush temperature monitoring system is researched, and the carbon brush temperature monitoring system has great significance for improving the economic benefit of a power plant.
At present, two methods, namely a point temperature gun temperature measurement method and an infrared temperature sensor temperature measurement method, are mainly used for measuring the temperature of the carbon brush at home and abroad, the point temperature gun temperature measurement method is complex in steps, low in efficiency and prone to errors, and meanwhile, the site environment is dangerous. The infrared temperature sensor temperature measurement method is non-contact measurement, saves complicated wiring and tedious manual inspection, and has the defects that single-point or matrix sensors are mostly adopted for temperature measurement, the coverage area is limited, pixels are low, the quantity of acquired data is small, and meanwhile, the precision is poor, and the requirement of a large-scale power plant cannot be met. Therefore, a set of temperature monitoring system can be researched by the thermal infrared imager, so that the infrared imaging can be realized, the pixel is high, the coverage area is large, and the temperature data can be collected and processed.
A large amount of experimental data show that the carbon brushes of the power plant have different temperature variation trends in different seasons. Meanwhile, the temperature change is not jump but slowly increases. The power plant generally adopts the drawing carbon brush temperature curve, judges the change of carbon brush temperature through the trend, and through the historical data of certain cycle, the analysis predicts the general future trend of whole incident. However, the method of prediction by rule has large error, and only the approximate trend of temperature can be obtained, and the quantitative analysis cannot be well performed. Through the trend prediction of the temperature data of the carbon brush of the generator, the unplanned shutdown accidents of the generator set caused by the faults of collector rings and the carbon brush of the generator can be effectively reduced, and meanwhile, through analysis, the method can provide guiding significance for operation and maintenance personnel on the running state of the carbon brush of the power plant and the load analysis of an excitation system.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a generator carbon brush temperature monitoring system and a temperature data trend prediction method based on infrared images, which can realize remote real-time monitoring of multiple devices, data storage of a database and analysis and prediction of temperature data, and can provide guiding significance for operation and maintenance personnel on analysis of the running state of the carbon brush of the power plant and the load of an excitation system on the basis of improving the detection efficiency.
In order to achieve the purpose, the invention adopts the technical scheme that: the generator carbon brush temperature on-line monitoring system based on the infrared image comprises a carbon brush group arranged on a generator collecting ring; an annular fixing support is arranged on the generator platform, the two thermal infrared imagers are fixed on two sides of the generator through the annular fixing support, and the visual angles of the thermal infrared imagers cover all carbon brushes of the generator; the thermal infrared imager is connected with the LAN port of the industrial 4G router through a twisted pair; the industrial 4G router uploads the on-site data to the cloud server through the 4G base station; the remote control end is provided with an industrial personal computer, generator carbon brush temperature online monitoring software is installed in the industrial personal computer, field data are processed, and the data are stored in a local database; the mobile terminal monitors the field condition by accessing the local database.
The thermal infrared imager adopts an infrared detector of an uncooled focal plane, the pixel value specification is 160 x 120, each infrared image contains 19200 pixels, and the temperature measurement range is-20-150 ℃; the thermal infrared imager is used for acquiring an on-site infrared image and transmitting the image to the cloud server through a network; the thermal infrared imager comprises a first thermal infrared imager and a second thermal infrared imager.
The cloud server accesses the infrared image data of the thermal infrared imager to a network through an industrial router loaded with a 4G internet of things card, and is wirelessly connected with the cloud server through a fixed IP address and the serial number of the thermal infrared imager; the server adopts a 4G industrial router, and the base station adopts a 4G base station.
The on-line monitoring software for the temperature of the carbon brush of the generator comprises upper computer software and database management software, infrared images and carbon brush temperature data measured in an industrial field are displayed in an industrial personal computer in a remote and real-time mode, the on-line monitoring software for the temperature of the carbon brush of the generator has the functions of overtemperature alarm, data recording, image recording, temperature curve drawing, temperature parameter correction and fault diagnosis, and a mobile terminal detection system of an operator is in wireless connection with a data management system through a fixed I P address.
The method for predicting the carbon brush temperature of the generator based on the infrared image comprises the following steps:
step 3, using the data filled with the missing values in the step 2, and aiming at the time sequence characteristics of the generator carbon brush temperature data and the characteristics of nonlinearity and influence of multiple factors of the generator carbon brush temperature data, giving play to the time sequence prediction advantages of the long-time memory network, and combining the nonlinear characteristics of the feedforward neural network to construct and train a long-time memory-feedforward neural network combination model;
and 4, predicting the trend of the carbon brush temperature data of the generator by using the long-time memory-feedforward neural network combined model, analyzing the running condition of the carbon brush and the load state of an excitation system according to the trend of the carbon brush temperature data of the generator, drawing an actual temperature curve chart by using the temperature data in the step 1, comparing the actual temperature curve chart with a prediction result, and verifying the prediction precision from the absolute average error, the average absolute percentage error and the root mean square error.
The infrared image data and the temperature data in the step 1 are collected by an on-line monitoring system for the temperature of the carbon brush of the generator and stored in a local database, the local database is accessed at a remote industrial control machine end, and 539 generator carbon brush infrared image data and temperature data which are collected by a power plant continuously for 10 days are selected randomly, wherein the number of training sets is 485, the training sets account for 90% of the total number of the infrared image data and the temperature data of the carbon brush of the generator, the number of test sets accounts for 54%, and the training sets account for 10% of the total number of the infrared image data and the temperature data of the carbon brush of the generator.
1) carrying out image binarization; 2) searching a four-connected area by using the bwleabel, wherein the four-connected area refers to a combination which can move in four directions of up, down, left and right from one point on the area, and reaches any pixel in the area on the premise of not exceeding the area; 3) marking the areas judged to be four connected; 4) selecting regions, and only reserving interested regions; 5) calculating the area of the reserved area by using regionprops;
and acquiring a corresponding infrared image in the online monitoring software of the temperature of the carbon brush of the generator, and filling the missing value through the corresponding relation between the temperature data and the area of the reserved area so as to ensure the integrity of the data.
The long-short-term memory-feedforward neural network combination model in the step 3 determines the weight of the long-short-term memory neural network model and the feedforward neural network model by a linear programming method in a series combination mode, and the parameters are set as follows:
long and short time memory neural network model: the input layer and the output layer are all 1 neuron, the hidden layer structure unit is set to be 3 layers, the number of the neurons is 60, 200 and 60, the maximum iteration number is 400, and when the error is less than 10-5When the time comes, the circulation is jumped out;
feedforward neural network model: the input layer and the output layer are all 1 neuron, the feedforward neural network is a double hidden layer and respectively comprises 8 and 4 neurons, the learning coefficient is 0.01, and the error control rate is 10-5Maximum training frequency is 5000 times, training function is train lm, and relative error distribution interval is tested [0.01,0.08]。
In the step 4, a long-time memory-feedforward neural network combined model is used for predicting the trend of the carbon brush temperature data of the generator, the carbon brush temperature data selected in the step 1 is used for drawing an actual temperature curve chart, the result is compared with the actual temperature curve, the evaluation index is used for verifying the prediction result, and the method is implemented according to the following steps:
the evaluation indexes used comprise absolute average error, average absolute percentage error and root mean square error, and the calculation formula is as follows:
in the formula yiThe actual carbon brush temperature value/DEG C of the generator;
is the predicted generator carbon brush temperature value/° c; n is the number of test sample sets; 1, 2, …, N;
the experimental test is carried out, the result shows that the method can well predict the data trend of the carbon brush temperature of the generator, the prediction error is mostly stabilized within 0.3 ℃, the running condition of the carbon brush and the load state of the excitation system are analyzed through the temperature data trend by predicting the carbon brush temperature of the generator, and meanwhile, the combined model can provide reasonable reference for the data prediction and decision planning of relevant types.
The invention has the beneficial effects that:
aiming at the problem that the normal operation of equipment is influenced by abnormal temperature rise of a carbon brush of a generator and a slip ring connected with the carbon brush of the generator in the flow guiding process due to the reasons of overhigh current, abnormal cooling device and the like in the carbon brush of a large generator, the invention designs the on-line monitoring system and the method which can remotely monitor operation data in real time and analyze and predict the temperature of the carbon brush;
the method can be used for well predicting the data trend of the carbon brush temperature of the generator, the prediction error is mostly stabilized within 0.3 ℃, and the running condition of the carbon brush and the load state of an excitation system are analyzed through the temperature trend by predicting the carbon brush temperature of the generator. Meanwhile, the combined model can provide reasonable reference for data prediction and decision planning of relevant types.
The device has the advantages that the real-time monitoring of the carbon brush temperature data is accurate due to the adoption of the high-precision thermal infrared imager; the data is accessed to the cloud database and the local database by adopting the 4G network to cooperate with the cloud server to realize data storage and management, so that the system has the advantages of remote monitoring and real-time synchronization; because multi-platform development multifunctional upper computer software is adopted, the system has the advantages of multi-equipment remote temperature monitoring, and also has the functions of overtemperature alarm, data recording, image recording, temperature curve drawing, temperature parameter correction, fault diagnosis and the like; the method of the invention adopts the steps of predicting the data trend of the carbon brush temperature of the generator by the long-time memory-feedforward neural network combined model and verifying the prediction result by using the evaluation index, so the method has the advantages of high prediction precision, nonlinearity and good time sequence characteristic.
Drawings
FIG. 1 is a schematic diagram of the system of the present invention.
Fig. 2 is a software flow chart of the carbon brush temperature analysis system according to the present invention.
Fig. 3 is a diagram of the field collected image and the result (experiment) extracted based on the sobel profile of the generator of the present invention.
FIG. 4 is a system diagram of a long-short-term memory-feedforward neural network combination model according to the present invention, wherein the combination model adopts a flow chart of a series combination mode.
Fig. 5 is a comparison graph (simulation) of the long-time and short-time memory-feedforward neural network prediction result, the long-time and short-time memory neural network prediction result and actually acquired temperature data.
Fig. 6 is a graph of predicted network error (simulation) according to the present invention.
FIG. 7 is a graph of the predicted relative error (simulation) of the present invention.
Detailed Description
In order that the objects, technical solutions and advantages of the present invention will become more apparent, the present invention will be further described in detail with reference to the accompanying drawings in conjunction with the following specific embodiments.
As figure 1 generator carbon brush temperature on-line monitoring system based on infrared image includes: 2 thermal infrared imagers, 4G industrial routers, GPRS/4G base stations, cloud servers, carbon brush temperature analysis systems, data management systems and mobile terminal monitoring systems;
the two thermal infrared imagers are respectively arranged on two sides of the generator, so that the detection range of the thermal infrared imagers is ensured to include the main part of the slip ring of the generator and the whole range of the carbon brush; infrared image data of the thermal imager is transmitted to an industrial 4G router through a twisted pair, a real-time measured data image is transmitted to a cloud server through a 4G/GPRS base station, fig. 2 is a flow chart of on-line monitoring software for the temperature of a carbon brush of a generator, an upper computer carbon brush temperature analysis system developed at a remote industrial computer end acquires the infrared image data on the cloud server for infrared thermal imaging and data analysis, and the upper computer is matched with a MYSQL database to realize the functions of over-temperature alarm, data recording, image recording, temperature curve drawing, temperature parameter correction, fault diagnosis and the like;
as shown in fig. 2, when the carbon brush temperature analysis system is used, each function is initialized according to the default parameter, and each file information configured currently is read. Meanwhile, the software can automatically store the relevant parameters of the generator setting to a software directory for reading and using when the software is opened next time. If the configuration information of each file is normal, clicking ' connect ' to automatically connect the program with the equipment, if not, prompting ' failure to open the configuration file, please reinitialize the parameters! And at the moment, initializing monitoring parameters by software to be default settings, entering a main dialog box interface, clicking a 'play' button on the dialog box interface, creating a display thread, enabling a picture transmitted by the thermal infrared imager and captured specified carbon brush temperature information to be visible in the display thread, and displaying data in real time on a right list.
The method comprises the steps of obtaining carbon brush temperature, selecting whether to draw a real-time curve of a certain point or not, simultaneously clicking to start unit data storage, starting a temperature calculation thread and a data storage thread, calculating temperature data, storing the temperature data into a local database through the data storage thread, automatically judging whether the temperature exceeds the limit or not by a system, recording temperature and time data into the database if the temperature and the time data exceed the limit, starting an overtemperature index display, calling an overtemperature screenshot program to capture and store infrared images, and obtaining the temperature of the generator carbon brush again. And when the carbon brush of the generator fails to operate, an alarm signal is immediately sent to the background system. If the temperature is not over-limit, whether to draw a historical curve or not can be selected, and if a button of drawing the historical curve is clicked, a historical temperature curve graph in the period can be seen after the starting time and the ending time are set. The infrared thermal image can be turned off by clicking a 'stop' button, and the acquisition of the carbon brush temperature is suspended, otherwise the functions of temperature acquisition and monitoring are executed in a circulating mode all the time. Clicking a 'closing' button to release all threads, closing a dialog box to finish software operation, and storing the carbon brush temperature data and the picture obtained in the monitoring process in a software catalogue in a Word report form.
The mobile terminal monitoring system is based on mobile equipment monitoring software such as a mobile phone, and the terminal software realizes wireless connection with the background data management system by accessing the fixed IP address of the management system.
Based on the principle, the invention provides a generator carbon brush temperature data trend prediction method based on infrared images, which comprises the following steps:
the infrared image data and the temperature data are collected and stored in a local database by an on-line monitoring system for the temperature of the carbon brush of the generator, 539 generator carbon brush infrared image data and temperature data which are continuously collected for 10 days by a power plant are randomly selected by accessing the local database at a remote industrial control machine end, wherein the number of training sets is 485 and accounts for 90% of the total number of the infrared image data and the temperature data of the carbon brush of the generator, the number of test sets is 54 and accounts for 10% of the total number of the infrared image data and the temperature data of the carbon brush of the generator;
the specific operation of contour extraction is: 1) carrying out image binarization; 2) searching a four-connected area by using the bwleabel, wherein the four-connected area refers to a combination which can move in four directions of up, down, left and right from one point on the area, and reaches any pixel in the area on the premise of not exceeding the area; 3) marking the areas judged to be four connected; 4) selecting regions, and only reserving interested regions; 5) calculating the area of the reserved area by using regionprops; FIG. 3 is a field collected image of the generator and a result extracted based on a sobel profile;
and acquiring a corresponding infrared image in the online monitoring software of the temperature of the carbon brush of the generator, and filling the missing value through the corresponding relation between the temperature data and the area of the reserved area so as to ensure the integrity of the data.
Step 3, using the data filled with the missing values in the step 2, and aiming at the time sequence characteristics of the generator carbon brush temperature data and the characteristics of nonlinearity and influence of multiple factors of the generator carbon brush temperature data, giving play to the time sequence prediction advantages of the long-time memory network, and combining the nonlinear characteristics of the feedforward neural network to construct and train a long-time memory-feedforward neural network combination model;
fig. 4 is a system block diagram of a long-and-short-term memory-feedforward neural network combination model, the combination model adopts a series combination mode, the long-and-short-term memory neural network model and the feedforward neural network model determine weights by a linear programming method, and the parameters are set as follows:
the long-time memory neural network model comprises an input layer and an output layer which are all 1 neuron, and a hidden layer structure unit is set to be 3 layers and respectively comprises 60, 200 and 60 neurons, the maximum iteration number is 400, and when the error is less than 10-5When the time comes, the circulation is jumped out;
the feedforward neural network model comprises an input layer and an output layer which are all 1 neuron, the feedforward neural network is a double hidden layer and respectively comprises 8 and 4 neurons, the learning coefficient is 0.01, and the error control rate is 1 multiplied by 10-5Maximum training frequency is 5000 times, training function is train lm, and relative error distribution interval is tested [0.01,0.08]；
And 4, predicting the trend of the carbon brush temperature data of the generator by using the long-time memory-feedforward neural network combined model, analyzing the running condition of the carbon brush and the load state of an excitation system according to the trend of the carbon brush temperature data of the generator, drawing an actual temperature curve chart by using the temperature data in the step 1, comparing the actual temperature curve chart with a prediction result, and verifying the prediction precision from the absolute average error, the average absolute percentage error and the root mean square error.
Further, in the step 4, a long-time memory-feedforward neural network combination model is used for predicting the temperature trend of the carbon brush of the generator, the carbon brush temperature data selected in the step 1 is used for drawing an actual temperature curve graph, the result is compared with the actual temperature curve, the predicted result is verified by using an evaluation index, and the used evaluation index is as follows:
in the formula yiThe actual carbon brush temperature value/DEG C of the generator;
is the predicted generator carbon brush temperature value/° c; n is the number of test sample sets; i is 1, 2, …, N,
in order to verify the pair prediction effect of the combined model, the data in the step 1 is filled with the missing value by using the method in the step 2, and then the long-time and short-time memory-feedforward neural network combined model established in the step 3 is used for predicting the trend of the temperature, referring to fig. 4, the combined prediction in the figure adopts a series combination mode, and the long-time and short-time memory model and the feedforward model determine the weight by using a linear programming method, and the method mainly comprises the following steps:
firstly, a long-time memory neural network model is used for prediction: 1) loading carbon brush temperature data, and dividing the data into a training set and a test set; 2) normalizing the data, preparing variables for training using the normalized processed data; 3) training a long-time memory neural network by using a train network according to specified training Options, calculating an output value of a long-time memory neural network storage unit and an error item of each long-time memory neural network storage unit according to parameters set in Options, calculating the gradient of each weight by combining the corresponding error items, and updating the weight by adopting a gradient optimization algorithm; 4) when the training is long, the neural network is remembered to be trained well, and the output training effect index jumps out of the training process; 5) predicting by using the trained network;
then, training and predicting the result of the long-time memory neural network prediction by using a feedforward neural network: 6) assigning random numbers in an interval (-1,1) to each connection weight, setting an error function e, giving a calculation precision value beta and a maximum learning frequency M, and randomly selecting a kth input sample and a corresponding expected output; 7) calculating the input and the output of each neuron of the hidden layer, calculating the partial derivative of an error function to each neuron of the output layer, correcting a connection weight omega (k) and a connection weight, and using an optimized feedforward neural network model; 8) calculating a global error, judging whether the network error meets the requirement or not, and finishing training after the error meets a target error when the error meets the preset precision;
calculating a prediction error after model training is completed, and evaluating a prediction result: 9) predicting by using a trained feedforward neural network model, and performing inverse normalization on the obtained result data to obtain a predicted value; 10) and calculating the prediction condition of the model through the evaluation index, and analyzing the prediction precision of the long-time memory-feedforward neural network combined model in comparison with other models.
Fig. 5 is a comparison graph of the long-and-short-term memory-feedforward neural network combined model prediction result, the long-and-short-term memory neural network prediction result and actually acquired temperature data, the training data is arranged at the front section of the curve, the prediction data is arranged at the rear section of the curve, the trend of the carbon brush temperature data of the generator can be well predicted by the long-and-short-term memory-feedforward neural network combined model, and the prediction result is closer to the actual data than the long-and-short-term memory neural network.
FIG. 6 is a predicted network error, and FIG. 6 illustrates that most of the temperature prediction errors are stabilized within 0.3 ℃, and the overall error distribution of the long-time memory-feedforward neural network combined model is superior to the error of the long-time memory neural network.
Fig. 7 shows the predicted relative error, and fig. 7 illustrates that the relative error distribution interval of the long and short term memory neural network is [ -0.008,0.008], and the relative error distribution interval of the long and short term memory-feedforward neural network combined model is [ -0.006,0.008 ].
Table 1 shows performance evaluation of the long-short time memory neural network and the long-short time memory-feedforward neural network combined model for predicting the temperature of the carbon brush of the generator.
TABLE 1
| Evaluation index | LSTM network | BP-LSTM network |
| MAE | 0.1916 | 0.1601 |
| MAPE | 0.0039 | 0.0033 |
| RMSE | 0.2556 | 0.2380 |
Table 1 shows that the three evaluation index values of the long-time and short-time memory-feedforward neural network combined model are better than those of the single model, wherein the average absolute error MAE of the combined model is 0.1610, the root mean square error RMSE is 0.238, and both are smaller than those of the single model, which indicates that the combined model has higher prediction accuracy, the performance of the combined model in practical application is better than that of the single model, the combined prediction can more fully exert the advantages of the single model, and the rationality and effectiveness of applying the long-time and short-time memory-feedforward neural network combined model to the carbon brush temperature prediction of the generator are shown.
Experimental tests show that the method can well predict the temperature trend of the carbon brush of the generator, most of prediction errors are stabilized within 0.3 ℃, and the running condition of the carbon brush and the load state of an excitation system are analyzed through the temperature trend by predicting the temperature of the carbon brush of the generator.
Meanwhile, the long-time memory-feedforward neural network combined model can provide reasonable reference for relevant types of data prediction and decision planning.
The foregoing is a preferred embodiment of the present invention, and it will be apparent to those skilled in the art that variations, modifications, substitutions and alterations can be made in the embodiment without departing from the principles and spirit of the invention.
Claims (9)
1. A generator carbon brush temperature monitoring system based on infrared images is characterized by comprising a carbon brush group (1) arranged on a generator collecting ring (3); an annular fixing support (5) is arranged on the generator platform (2), the two thermal infrared imagers (4) are fixed on two sides of the generator through the annular fixing support, and the visual angles of the thermal infrared imagers (4) cover all carbon brushes of the generator; the thermal infrared imager (4) is connected with an LAN port of the industrial 4G router (6) through a twisted pair; the industrial 4G router (6) uploads the field data to the cloud server (8) through the 4G base station (7); the remote control end is provided with an industrial personal computer (9), generator carbon brush temperature online monitoring software is installed in the industrial personal computer (9), field data are processed, and the data are stored in a local database (10); the mobile terminal (11) monitors the field condition by accessing the local database (10).
2. The system for monitoring the carbon brush temperature of the generator based on the infrared image as claimed in claim 1, wherein the thermal infrared imager adopts an infrared detector with an uncooled focal plane, the pixel value specification is 160 x 120, each infrared image contains 19200 pixels, and the temperature measurement range is-20 ℃ to 150 ℃; the thermal infrared imager is used for acquiring an on-site infrared image and transmitting the image to the cloud server through a network; the thermal infrared imager comprises a first thermal infrared imager and a second thermal infrared imager.
3. The system for monitoring the carbon brush temperature of the generator based on the infrared images as claimed in claim 1, wherein the cloud server accesses the infrared image data of the thermal infrared imager to a network through an industrial router loaded with a 4G Internet of things card, and is wirelessly connected with the cloud server through a fixed IP address and the serial number of the thermal infrared imager.
4. The system for monitoring the temperature of the carbon brush of the generator based on the infrared image according to claim 1, wherein the software for monitoring the temperature of the carbon brush of the generator on line comprises upper computer software and database management software, the infrared image and the carbon brush temperature data measured in an industrial field are remotely displayed in an industrial personal computer in real time, the software for monitoring the temperature of the carbon brush of the generator has the functions of over-temperature alarm, data recording, image recording, temperature curve drawing, temperature parameter correction and fault diagnosis, and a mobile terminal detection system of an operator is in wireless connection with the data management system through a fixed IP address.
5. The method for predicting the carbon brush temperature of the generator based on the infrared image is characterized by comprising the following steps of:
the method comprises the following steps that 1, an infrared image-based generator carbon brush temperature monitoring system is adopted, an on-site monitoring end acquires infrared image data of an on-site carbon brush group (1) through an infrared thermal imager (4) and uploads the infrared image data to a cloud server (8) through an industrial router (6) and a 4G base station (7), a remote industrial control end (9) processes the infrared image data through upper computer software to obtain temperature data and stores the temperature data to a local database (10), and then the infrared image data and the carbon brush temperature data in the local database for a period of time are randomly selected;
step 2, performing edge detection on the infrared image data in the step 1 by using a sobel operator, extracting a carbon brush outline in the infrared image data, and filling a missing value by using a corresponding relation between the temperature data of the carbon brush of the generator and the infrared image area of the heating carbon brush;
step 3, using the data filled with the missing values in the step 2, and aiming at the time sequence characteristics of the generator carbon brush temperature data and the characteristics of nonlinearity and influence of multiple factors of the generator carbon brush temperature data, giving play to the time sequence prediction advantages of the long-time memory network, and combining the nonlinear characteristics of the feedforward neural network to construct and train a long-time memory-feedforward neural network combination model;
and 4, predicting the trend of the carbon brush temperature data of the generator by using the long-time memory-feedforward neural network combined model, analyzing the running condition of the carbon brush and the load state of an excitation system according to the trend of the carbon brush temperature data of the generator, drawing an actual temperature curve chart by using the temperature data in the step 1, comparing the actual temperature curve chart with a prediction result, and verifying the prediction precision from the absolute average error, the average absolute percentage error and the root mean square error.
6. The method for predicting the carbon brush temperature of the generator according to claim 5, wherein the infrared image data and the temperature data in the step 1 are collected and stored in a local database by an online generator carbon brush temperature monitoring system, the local database is accessed at a remote industrial computer terminal, and the infrared image data and the temperature data of the generator carbon brush collected continuously for 10 days by the power plant are randomly selected to be 539, wherein the number of training sets is 485 and accounts for 90% of the total number of the infrared image data and the temperature data of the generator carbon brush, and the number of test sets is 54 and accounts for 10% of the total number of the infrared image data and the temperature data of the generator carbon brush.
7. The method for predicting the carbon brush temperature of the generator based on the infrared image as claimed in claim 5, wherein the step 2 is to extract the carbon brush outline in the infrared image, and the specific method is as follows:
1) carrying out image binarization; 2) searching a four-connected area by using the bwleabel, wherein the four-connected area refers to a combination which can move in four directions of up, down, left and right from one point on the area, and reaches any pixel in the area on the premise of not exceeding the area; 3) marking the areas judged to be four connected; 4) selecting regions, and only reserving interested regions; 5) calculating the area of the reserved area by using regionprops;
and acquiring a corresponding infrared image in the online monitoring software of the temperature of the carbon brush of the generator, and filling the missing value through the corresponding relation between the temperature data and the area of the reserved area so as to ensure the integrity of the data.
8. The method for predicting the carbon brush temperature of the generator based on the infrared image as claimed in claim 5, wherein the long-short time memory-feedforward neural network combination model in the step 3 determines the weight of the long-short time memory neural network model and the feedforward neural network model by a linear programming method in a series combination mode, and the parameters are set as follows:
long and short time memory neural network model: the input layer and the output layer are all 1 neuron, the hidden layer structure unit is set to be 3 layers, the number of the neurons is 60, 200 and 60, the maximum iteration number is 400, and when the error is less than 10-5When the time comes, the circulation is jumped out;
feedforward neural network model: the input layer and the output layer are all 1 neuron, the feedforward neural network is a double hidden layer and respectively comprises 8 and 4 neurons, the learning coefficient is 0.01, and the error control rate is 10-5Maximum training frequency is 5000 times, training function is train lm, and relative error distribution interval is tested [0.01,0.08]。
9. The method for predicting the temperature data of the carbon brush of the generator based on the infrared image as claimed in claim 5, wherein a long-time memory-feedforward neural network combined model is used for predicting the trend of the temperature data of the carbon brush of the generator in the step 4, an actual temperature curve is drawn by using the temperature data of the carbon brush selected in the step 1, the result is compared with the actual temperature curve, an evaluation index is used for verifying the prediction result, and the method is implemented according to the following steps:
the evaluation indexes used comprise absolute average error, average absolute percentage error and root mean square error, and the calculation formula is as follows:
in the formula yiThe actual carbon brush temperature value/DEG C of the generator;
is the predicted generator carbon brush temperature value/° c; n is the number of test sample sets; 1, 2, …, N;
the method can well predict the data trend of the carbon brush temperature of the generator, the prediction error is mostly stabilized within 0.3 ℃, the running condition of the carbon brush and the load state of an excitation system are analyzed through the temperature data trend by predicting the carbon brush temperature of the generator, and meanwhile, the combined model can provide reasonable reference for the data prediction and decision planning of relevant types.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110956798.7A CN113820017A (en) | 2021-08-19 | 2021-08-19 | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110956798.7A CN113820017A (en) | 2021-08-19 | 2021-08-19 | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113820017A true CN113820017A (en) | 2021-12-21 |
Family
ID=78913337
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110956798.7A Withdrawn CN113820017A (en) | 2021-08-19 | 2021-08-19 | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113820017A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113836816A (en) * | 2021-09-30 | 2021-12-24 | 华能威海发电有限责任公司 | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method |
| CN115015757A (en) * | 2022-08-09 | 2022-09-06 | 天津九信科技有限公司 | Risk assessment method and system for motor running state |
| CN115248873A (en) * | 2022-09-22 | 2022-10-28 | 国网山西省电力公司太原供电公司 | Cable tunnel safety monitoring method and system based on data fusion |
| CN117781896A (en) * | 2024-02-23 | 2024-03-29 | 天生桥二级水力发电有限公司 | Infrared detection processing method and device suitable for carbon brush of unit and storage medium |
| CN117848515A (en) * | 2024-03-07 | 2024-04-09 | 国网吉林省电力有限公司长春供电公司 | Switch cabinet temperature monitoring method and system |
-
2021
- 2021-08-19 CN CN202110956798.7A patent/CN113820017A/en not_active Withdrawn
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113836816A (en) * | 2021-09-30 | 2021-12-24 | 华能威海发电有限责任公司 | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method |
| CN115015757A (en) * | 2022-08-09 | 2022-09-06 | 天津九信科技有限公司 | Risk assessment method and system for motor running state |
| CN115248873A (en) * | 2022-09-22 | 2022-10-28 | 国网山西省电力公司太原供电公司 | Cable tunnel safety monitoring method and system based on data fusion |
| CN117781896A (en) * | 2024-02-23 | 2024-03-29 | 天生桥二级水力发电有限公司 | Infrared detection processing method and device suitable for carbon brush of unit and storage medium |
| CN117781896B (en) * | 2024-02-23 | 2024-05-10 | 天生桥二级水力发电有限公司 | Infrared detection processing method and device suitable for carbon brush of unit and storage medium |
| CN117848515A (en) * | 2024-03-07 | 2024-04-09 | 国网吉林省电力有限公司长春供电公司 | Switch cabinet temperature monitoring method and system |
| CN117848515B (en) * | 2024-03-07 | 2024-05-07 | 国网吉林省电力有限公司长春供电公司 | Switch cabinet temperature monitoring method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113820017A (en) | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method | |
| Jaffery et al. | Scheme for predictive fault diagnosis in photo-voltaic modules using thermal imaging | |
| CN103901072B (en) | Method for diagnosing equipment overheating defects by utilizing infrared spectrum analysis | |
| CN109784348B (en) | Infrared power equipment identification and online diagnosis method and system | |
| CN113836816A (en) | Generator carbon brush temperature monitoring system based on infrared image and temperature prediction method | |
| JP2023508759A (en) | Method and Apparatus, Device and Storage Medium for Determining Operational State of Photovoltaic Array | |
| CN107807860B (en) | Power failure analysis method and system based on matrix decomposition | |
| Fang et al. | Fault diagnosis of electric transformers based on infrared image processing and semi-supervised learning | |
| Ding et al. | Fault detection of photovoltaic array based on Grubbs criterion and local outlier factor | |
| CN117783769B (en) | Power distribution network fault positioning method, system, equipment and storage medium based on visual platform | |
| CN118155047B (en) | Small battery management method suitable for alpine region | |
| Zheng et al. | Intelligent diagnosis method of power equipment faults based on single‐stage infrared image target detection | |
| CN114821852A (en) | Power grid defect depth identification inspection robot control system based on characteristic pyramid | |
| CN116227676A (en) | Production equipment energy consumption analysis method and device, electronic equipment and storage medium | |
| CN113284103B (en) | Substation equipment defect online detection method based on space transformation fast R-CNN model | |
| CN113920449A (en) | Photovoltaic power station inspection method and system, computer equipment and storage medium | |
| CN117972419A (en) | Anomaly-transporter photovoltaic power generation abnormality detection method | |
| CN107505049A (en) | A kind of transformer station's infrared detection method and system | |
| CN110262313A (en) | Power Material key point remote supervision system based on technology of Internet of things | |
| Jacome et al. | Real-Time Fault Identification of Photovoltaic Systems Based on Remote Monitoring with IoT | |
| CN117764547A (en) | Photovoltaic string fault diagnosis method and system | |
| CN115146715A (en) | Power utilization potential safety hazard diagnosis method, device, equipment and storage medium | |
| CN110318953A (en) | Temperature monitoring method and device for wind turbine generator electric control system | |
| Zhao et al. | [Retracted] Working Condition Monitoring System of Substation Robot Based on Video Monitoring | |
| LU500843B1 (en) | A monitoring system and prediction method of temperature for generator carbon brush based on an infrared image |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20211221 |