CN115342680A - Intelligent method for identifying abnormal state of indirect air cooling system - Google Patents
Intelligent method for identifying abnormal state of indirect air cooling system Download PDFInfo
- Publication number
- CN115342680A CN115342680A CN202210915864.0A CN202210915864A CN115342680A CN 115342680 A CN115342680 A CN 115342680A CN 202210915864 A CN202210915864 A CN 202210915864A CN 115342680 A CN115342680 A CN 115342680A
- Authority
- CN
- China
- Prior art keywords
- model
- radiator
- data
- cooling system
- temperature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F28—HEAT EXCHANGE IN GENERAL
- F28F—DETAILS OF HEAT-EXCHANGE AND HEAT-TRANSFER APPARATUS, OF GENERAL APPLICATION
- F28F27/00—Control arrangements or safety devices specially adapted for heat-exchange or heat-transfer apparatus
- F28F27/003—Control arrangements or safety devices specially adapted for heat-exchange or heat-transfer apparatus specially adapted for cooling towers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Thermal Sciences (AREA)
- Mechanical Engineering (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses an intelligent method for identifying abnormal states of an indirect air cooling system, which comprises the following steps: s1, arranging an m multiplied by n fiber bragg grating temperature sensor array on the surface of a radiator of a cooling triangle, and acquiring temperature field distribution data of the radiator and the cooling triangle; s2, judging the accuracy of the running state of the radiating surface according to an intelligent algorithm, and providing a quantitative index for identifying the accuracy; s3, acquiring the operation state of the sector according to the quantification index and by combining the actual operation state of the radiator; and S4, forming a decision tree model through the combination of the running states of the row actuators, and judging the running state of the sector by combining the temperature field distribution data. The state of the radiating surface is judged by adopting the XGboost extreme gradient promotion, the method is suitable for the conditions of multiple data types and multiple variables, and the provided abnormal state identification algorithm is favorable for an air cooling system to meet the requirement of freezing prevention in winter.
Description
Technical Field
The invention relates to the technical field of identification of abnormal states of indirect air cooling systems, in particular to an intelligent method for identifying abnormal states of indirect air cooling systems.
Background
At present, the surface condensing type indirect air cooling system has wide application in the field of thermal power generation, an indirect cooling tower is an important component of the indirect cooling system, the indirect cooling tower mainly exchanges heat through a plurality of cooling triangles which enclose a circle at the bottom of the tower, and the cooling triangles are composed of a louver surface and two radiator columns. The radiator is generally composed of circular tubes and rectangular fins for conveying circulating cooling water, and the water temperature in the tubes is reduced by the convective heat transfer of air flowing outside the tubes. However, the operating environment is extremely harsh due to long term exposure to air, and the radiator is prone to malfunction and abnormality. In winter, due to the fact that the temperature of the sector temperature field is too low, abnormal conditions that the radiator is frozen or even frozen and cracked can be caused. Suspended matters in the air can be attached to the surface of the radiator, so that the scaling of the radiator is caused, and the heat exchange performance of the radiator is influenced. The working state of the cooling triangle greatly affects the working performance of the air cooling tower, and a large amount of manpower and material resources are consumed for overhauling and maintaining the radiator and the shutter.
In order to judge the dust deposition pollution degree of a cooling triangle and monitor the possibility that each sector of the air cooling tower is frozen, for example, the utility model CN201220700041.8 provides an indirect air cooling tower radiator temperature field online monitoring system, which uses a DSC bus controller to collect real-time temperature and monitor the temperature field of the radiator online by arranging intelligent temperature sensors at the air inlet and the air outlet of each radiator;
for example, the patent of invention cn202110470550.x proposes a steel structure indirect cooling tower temperature field monitoring system under 5G mode, which utilizes the high-speed, low-delay and large-data transmission characteristics of a 5G network, obtains temperature measurement data through a radiator wall surface fiber grating temperature sensor, and monitors and manages the working state of an air cooling tower radiator by using an air cooling island monitoring and management platform;
for example, the patent of utility model CN201420043052.2 proposes an indirect air cooling system performance test platform, which monitors the air temperature and other related parameters at different positions of the air inlet and outlet of the radiator by building an experiment platform, so as to facilitate the implementation of quantitative experiments and analyses related to the service life of other devices such as radiators, shutters and the like under different working conditions and different design sizes;
therefore, the work of the cooling triangle which can not be automatically operated by the army in the three comparison files can be obtained, so that how to monitor the working performance of the indirect cooling triangle and how to automatically judge the abnormal states of the radiating surface and the shutter are necessary conditions and key factors for realizing unattended operation and complete automatic operation of the intelligent air cooling system, and an effective solution is not provided at present aiming at the problems in the related technology.
Disclosure of Invention
The invention provides an intelligent method for identifying abnormal states of an indirect air cooling system, aiming at the problems in the related art, so as to overcome the technical problems in the prior related art.
Therefore, the invention adopts the following specific technical scheme:
an intelligent method for identifying abnormal states of an indirect air cooling system comprises the following steps:
s1, arranging an m multiplied by n fiber bragg grating temperature sensor array on the surface of a radiator of a cooling triangle, and acquiring temperature field distribution data of the radiator and the cooling triangle;
s2, judging the accuracy of the running state of the radiating surface according to an intelligent algorithm, and providing a quantitative index for identifying the accuracy;
s3, acquiring the operation state of the sector according to the quantitative index and by combining the actual operation state of the radiator;
and S4, forming a decision tree model by listing the combination of the operation states of the actuators, and judging the operation state of the sector by combining the temperature field distribution data.
Further, the step of arranging the mxn fiber bragg grating temperature sensor array on the surface of the heat sink of the cooling triangle and acquiring the temperature field distribution data of the heat sink and the cooling triangle further includes the steps of:
s11, constructing a fiber bragg grating temperature sensor array by constructing an mxn force measurement sensor array and an mxn temperature sensor;
s12, using optical fibers for the fiber grating temperature sensor array, and influencing the Bragg wavelength movement amount of the fiber grating sensor through the temperature change and the mechanical strain change;
s13, temperature compensation is carried out on the Bragg wavelength movement amount by adopting an external force method, the change of the piezoelectric sensor is used for detecting the change of the optical fiber and is reduced and transferred to the change of the piezoelectric sensor for detecting the temperature, the sum of the Bragg wavelength movement amounts is detected, and the temperature field distribution data of the radiator and the cooling triangle are obtained.
Further, the method for judging the accuracy of the running state of the radiating surface according to the intelligent algorithm and providing the quantitative index for identifying the accuracy further comprises the following steps:
s21, screening temperature field data measured by a radiating surface and temperature measurement data of an inlet and an outlet of a radiator, and eliminating abnormal data;
s22, establishing feature sets for the radiators and corresponding temperature fields in different running states;
s23, dividing the data subjected to feature selection into a training sample and a test sample;
s24, training the model by using the training samples, and performing parameter optimization on the XGboost algorithm model through grid search and cross validation to obtain a GC-XGboost algorithm model;
s25, testing the model obtained in the step S24 by using the test sample;
s26, transversely comparing the calculation results of the GC-XGboost algorithm model with a support vector machine, and checking the superiority of the used model through calculating and identifying accuracy.
Further, the dividing the data after feature selection into the training samples and the testing samples further includes the following steps:
s231, acquiring temperature field creation feature set data corresponding to the radiator, and performing data preprocessing of cleaning interpolation on the temperature field creation feature set data;
s232, modeling an XGboost model, training the XGboost model and the data, and recording a predicted evaluation index value;
and S233, enumerating all the parameters of the XGboost model, forming a grid, scoring all the models of the grid based on all the parameter combinations by using cross validation, scoring the XGboost model of the parameter combination with the highest score, and obtaining the GC-XGboost model.
Further, the feature set data includes a training set and a testing set.
Further, the optimized objective function modeled by the XGBoost model is as follows:
in the formula, obj (t) Representing the objective function of the t-th training, y i The true value of the ith sample is represented,
is the predicted value of the model of the t-1 st round, l represents the loss function, f i (x i ) Representing the input as x i Function value of time t-th round, Ω (f) i ) Denotes a regularization term, C denotes a constant, x i Data processed in step S231 is shown.
Further, the transverse comparison of the calculation results of the support vector machine and the GC-XGboost algorithm model is carried out, and the superiority of the used model under the application is checked through the calculation and identification accuracy, and the method further comprises the following steps:
s261, dividing the support vector machine into two types of modes A 1 Training set and A 2 Training set, T = { (x) 1 ，y 1 )(x 2 ，y 2 )…(x n ，y n ) Is from mode A 1 And A 2 From a training set obtained by sampling, where x n ∈R M 、y n E {1, -1}, wherein R M Comprises A 1 Class A and A 2 Class;
s262, if x n Belong to A 1 Class, then correspond to y n ＝1；
S263, if x n Belong to A 2 Class, then correspond to y n ＝-1；
S264, seeking R M Transversely comparing the real function g (x) with the parameters of the GC-XGboost algorithm model, wherein the comparison result is as follows;
Wherein sgn { } represents a sign function, and g (x) represents a decision classification function;
and S265, checking the superiority of the used model by calculating the identification accuracy.
Further, the actual operating states of the radiator include the following operating states: the two radiators work normally, the two radiators are abnormal, one radiator is abnormal, and the other radiator works normally;
wherein the abnormal state comprises isolation, freezing and stopping.
Further, the step of forming a decision tree model by listing the combination of the operating states of the actuators and determining the operating state of the sector by combining the temperature field distribution data further includes the steps of:
s41, determining all segmentation points of the sample characteristics through the combination of the operation states of the actuators in a list, and segmenting each determined segmentation point;
s42, selecting an optimal segmentation point through a scoring function;
and S44, comparing the cutting point temperature of the optimal cutting point with the temperature of each point of the sector, and judging the running state of each point of the sector according to the comparison result.
Further, the cutting standard formula of the cutting point is as follows:
in the formula (I), the compound is shown in the specification,gain represents the difference between a single node obj and a tree obj of two nodes after segmentation, the segmentation points of all the characteristics are traversed, the segmentation point of the maximum Gain is found, namely the optimal segmentation point, wherein obj represents an objective function, G L 、G R 、H L And H R Both represent first-order second derivatives of left and right subtrees of the current node, L and R both represent constants, and lambda represents a regularization coefficient;
if the gamma value is set to be too large, the Gain is negative, and the node is not segmented, because the segmented tree structure is deteriorated; the larger the value of γ, the more stringent the requirement for the descending amplitude of obj after slicing.
The beneficial effects of the invention are as follows: the state of the radiating surface is judged by adopting XGboost extreme gradient lifting, the improvement is based on GBDT, the XGboost algorithm is an efficient parallel machine learning algorithm, and the XGboost algorithm is suitable for the conditions of multiple data types and multiple variables; compared with the traditional fault diagnosis method, the XGboost algorithm overcomes the problem that the traditional method is not high enough in efficiency under the condition of large data scale, the precision of fault diagnosis is improved by effectively utilizing large-scale data, the phenomenon that a radiator is easily frozen due to too low sector temperature in winter, and meanwhile, the temperature field distribution state of a radiating surface which is not operated is different from the temperature distribution state of the radiating surface which is operated due to lower environment temperature in winter, so that the possibility of wrong judgment is avoided, and therefore, the abnormal state identification algorithm provided by the invention is particularly favorable for an air cooling system to meet the requirement of freezing prevention in winter.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flowchart of an intelligent method for recognizing abnormal conditions of an indirect air cooling system according to an embodiment of the present invention
Fig. 2 is a sector state diagram of an intelligent method for recognizing abnormal states of an indirect air-cooling system according to an embodiment of the present invention;
fig. 3 is a decision algorithm diagram of XGBoost in an intelligent method for identifying an abnormal state of an indirect air cooling system according to an embodiment of the present invention.
Detailed Description
For further explanation of the various embodiments, the drawings which form a part of the disclosure and which are incorporated in and constitute a part of this specification, illustrate embodiments and, together with the description, serve to explain the principles of operation of the embodiments, and to enable one skilled in the art to understand the embodiments and advantages of the disclosure for reference and without scale, wherein elements are not shown in the drawings and like reference numerals are used to refer to like elements generally.
According to the embodiment of the invention, an intelligent method for identifying the abnormal state of the indirect air cooling system is provided.
Referring to the drawings and the detailed description, the present invention will be further described, as shown in fig. 1, an intelligent method for recognizing an abnormal state of an indirect air cooling system according to an embodiment of the present invention includes the following steps:
s1, arranging an m multiplied by n fiber bragg grating temperature sensor array on the surface of a radiator of a cooling triangle, and acquiring temperature field distribution data of the radiator and the cooling triangle;
in one embodiment, the arranging of the mxn fiber grating temperature sensor array on the surface of the heat sink of the cooling triangle and acquiring the temperature field distribution data of the heat sink and the cooling triangle further comprises the following steps:
s11, constructing a fiber bragg grating temperature sensor array by constructing an mxn force measurement sensor array and an mxn temperature sensor;
s12, using optical fibers for the fiber grating temperature sensor array, and influencing the Bragg wavelength movement amount of the fiber grating sensor through the temperature change and the mechanical strain change;
and S13, carrying out temperature compensation on the Bragg wavelength movement amount by adopting an external force method, replacing the change of the piezoelectric sensor for detecting the change of the optical fiber reduction transfer to the piezoelectric sensor change for detecting the temperature, and acquiring the temperature field distribution data of the radiator and the cooling triangle.
S2, judging the accuracy of the running state of the radiating surface according to an intelligent algorithm, and providing a quantitative index for identifying the accuracy;
in one embodiment, as shown in fig. 2 to fig. 3, the determining the accuracy of the operating state of the cooling surface according to the intelligent algorithm and providing the quantitative index for identifying the accuracy further includes the following steps:
s21, screening temperature field data measured by a radiating surface and temperature measurement data of an inlet and an outlet of a radiator, and eliminating abnormal data;
s22, establishing feature sets for the radiators and corresponding temperature fields in different operating states;
s23, dividing the data subjected to feature selection into a training sample and a test sample;
s24, training the model by using the training samples, and performing parameter optimization on the XGboost algorithm model through grid search and cross validation to obtain a GC-XGboost algorithm model;
s25, testing the model obtained in the step S24 by using the test sample;
s26, transversely comparing the calculation results of the GC-XGboost algorithm model by using a support vector machine, and checking the superiority of the used model through calculating the identification accuracy;
in one embodiment, the dividing the feature-selected data into training samples and test samples further comprises:
s231, acquiring temperature field creation characteristic set data corresponding to the radiator, and performing data preprocessing of cleaning interpolation on the temperature field creation characteristic set data;
s232, modeling an XGboost model, training the XGboost model and the data, and recording a predicted evaluation index value;
s233, forming a grid after listing the permutation and combination of all the parameters of the XGboost model, scoring all the models of the grid based on all the parameter combinations by using cross validation, scoring the XGboost model of the parameter combination with the highest score, and obtaining the GC-XGboost model;
in specific application, a GC-XGboost model is established, the performance of the model needs to be evaluated, and a plurality of indexes capable of evaluating the performance of a prediction model of a radiator are provided, wherein the three indexes are common: respectively mean square error MSE (mean squared error) for measuring the deviation between the observed value and the true value, mean absolute error MAE (nmeaballolute error) for reflecting the error magnitude generated by the predicted value and R-square value R for comparing the model performance in different dimensions 2 In the embodiment of the invention, the three indexes are selected to comprehensively evaluate and compare the models;
MSE is obtained by dividing the error sum of squares by the sample size, the error sum of squares is the sum of squares of errors between a predicted value and a real value in the fitting process of the linear regression model, the smaller the value is, the closer the predicted value and the real value is, the better the fitting effect is, and the calculation formula of MSE is as follows:
in order to avoid the occurrence of the situation of positive and negative offset of the error, another method is to calculate by using an absolute value, so that the situation of the error magnitude between the real value and the predicted value can be better reflected, the smaller the value is, the better the prediction effect is, and the calculation formula is as follows:
R 2 the fitting effect of the model is represented, and the larger the value is, the better the fitting effect is represented; the smaller the value is, the worse the fitting effect is, and even the model is judged to be improper, and the calculation formula is as follows:
wherein i represents the i-th sample, y i The true value of the i-th sample is represented,
represents the predicted value of the ith sample, n is the total sample size,
represents the mean value;
in one embodiment, the feature set data comprises a training set and a test set;
in one embodiment, the optimized objective function modeled by the XGBoost model is as follows:
in the formula (I), the compound is shown in the specification,
representing the objective function of the t-th training, y i The true value of the ith sample is represented,
is the predicted value of the model of the t-1 st round, l represents the loss function, f i (x i ) Representing the input as x i Function value of time t-th round, Ω (f) i ) Denotes a regularization term, C denotes a constant, x i Data after processing in step S231;
in one embodiment, the transversely comparing the calculation results of the GC-XGBoost algorithm model with the support vector machine, and checking the superiority of the model used in the application by calculating the recognition accuracy further includes the following steps:
s261, dividing the support vector machine into two types of modes A 1 Training set and A 2 Training set, T = { (x, y) (x) 2 ，y 2 )…(x n ，y n ) Is from mode A 1 And A 2 In a training set obtained by sampling, wherein x n ∈R M 、y n E {1, -1}, wherein R M Comprises A 1 Class A and A 2 Class;
s262, if x n Belong to A 1 Class, then correspond to y n ＝1；
S263, if x n Belong to A 2 Class, then correspond to y n ＝-1；
S264, seeking R M Transversely comparing the real function g (x) with the parameters of the GC-XGboost algorithm model, wherein the comparison result is as follows;
Wherein sgn { } denotes a sign function, and g (x) denotes a decision classification function
S265, checking the superiority of the used model by calculating the identification accuracy;
in a specific application, a learning sample X is a sample or a special case of an actual pattern, the actual pattern in work may exceed the distribution range of the learning sample, if the distribution of the actual pattern can be predicted, and a classification function is determined according to the distribution, which is called as "optimal prediction", but is difficult to achieve in practice, no matter how large-scale samples are obtained, the samples or special cases of actual problems are always obtained, any estimation made by using the data only locally speculates the global situation, the center of the maximum sideband between two types of samples is taken as the classification function by the support vector machine, which is obviously the optimal classification of the existing learning samples, and has stronger rationality, and the classification function obtained by the support vector machine is called as "optimal structure".
S3, acquiring the running state of the sector according to the quantification index and by combining the actual running state of the radiator;
in one embodiment, the actual operating conditions of the radiator include the following: the two radiators work normally, the two radiators are abnormal, one radiator is abnormal, and the other radiator works normally;
wherein, the abnormal state comprises isolation, freezing and stopping.
S4, forming a decision tree model by listing the combination of the operation states of the actuators, and judging the operation state of the sector by combining the temperature field distribution data;
in one embodiment, the forming a decision tree model by listing combinations of the operating states of the actuators and determining the operating state of the sector by combining the temperature field distribution data further includes:
s41, determining all segmentation points of the sample characteristics through the combination of the operation states of the actuators in a list, and segmenting each determined segmentation point;
s42, selecting an optimal segmentation point through a scoring function;
s44, comparing the cutting point temperature of the optimal cutting point with the temperature of each point of the sector, and judging the running state of each point of the sector according to the comparison result;
in one embodiment, the cutting criteria formula for the cut point is as follows:
in the formula, gain represents the difference between a single node obj and a tree obj of two nodes after segmentation, the segmentation points of all the characteristics are traversed, the segmentation point of the maximum Gain is found, namely the optimal segmentation point, wherein obj represents a target function, G L 、G R 、H L And H R The first-order second-order derivatives of the left and right subtrees of the current node are represented, L and R represent constants, and lambda represents a regularization coefficient;
if the gamma value is set to be too large, gain is negative, the node is not segmented, and the segmented tree structure is poor; the larger the value of γ, the more strict the requirement for the descending amplitude of obj after slicing.
In a specific application, the implementation environment in the embodiment is as follows:
the indirect cooling intelligent control system of the double-good group thermal power plant provides a hardware environment and a test system for software execution. The indirect cooling intelligent control system is provided with a grating temperature sensor, and obtains measured temperature data to calibrate parameters of the software system.
The implementation environment in the examples is as follows:
the indirect cooling intelligent control system provides a indirect cooling tower testing and testing hardware system, wherein a plurality of radiating fins which have isolation, freeze and stop using three abnormal conditions are arranged in a partial cooling triangle. An m multiplied by n fiber bragg grating temperature sensor array is arranged on the surface of a radiator of a cooling triangle to obtain the mass historical temperature field distribution measurement data of different radiators and cooling triangles of the indirect cooling tower in different environmental wind speeds and environmental temperatures and different seasons.
After acquiring the actually measured surface temperature field and inlet and outlet temperatures of the radiator: and (1) preprocessing data. And screening the measured data of the temperature, and removing abnormal data. And (2) feature selection. A feature set is created for the heat sink and corresponding temperature field under different operating conditions. (3) And dividing the data subjected to feature selection into training samples and testing samples. (4) Training the model by using training samples, and performing parameter optimization on the XGboost algorithm model through grid search and cross validation. (5) And (4) testing the model obtained in the step (4) by using the test sample. And transversely comparing model calculation results of a Support Vector Machine (SVM) and an XGboost algorithm, and checking the superiority of the used model under the application.
In summary, by means of the above technical solution of the present invention, the state judgment of the cooling surface of the present invention adopts XGBoost (eXtreme Gradient Boosting) eXtreme Gradient Boosting, which is an improvement based on GBDT (Gradient Boosting decision tree), and the XGBoost algorithm is an efficient parallel machine learning algorithm, and is adapted to the situations of multiple data types and multiple variables; compared with the traditional fault diagnosis method, the XGboost algorithm solves the problem that the traditional method is not high enough in efficiency under the condition of large data scale, the precision of fault diagnosis is improved by effectively utilizing large-scale data, the phenomenon that a radiator is easily frozen due to too low sector temperature in winter is avoided, and meanwhile, due to the fact that the environmental temperature in winter is low, the temperature field distribution state of a radiating surface which does not operate is different from the temperature distribution state of the radiating surface which operates, and the possibility of wrong judgment is avoided.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (9)
1. An intelligent method for identifying abnormal states of an indirect air cooling system is characterized by comprising the following steps:
s1, arranging an mxn fiber grating temperature sensor array on the surface of a radiator of a cooling triangle, and acquiring temperature field distribution data of the radiator and the cooling triangle;
s2, judging the accuracy of the running state of the radiating surface according to an intelligent algorithm, and providing a quantitative index for identifying the accuracy;
s3, acquiring the operation state of the sector according to the quantitative index and by combining the actual operation state of the radiator;
s4, forming a decision tree model by listing the combination of the operation states of the actuators, and judging the operation state of the sector by combining the temperature field distribution data;
wherein, the step S1 further comprises the following steps:
s11, constructing a fiber bragg grating temperature sensor array by constructing an mxn force measurement sensor array and an mxn temperature sensor;
s12, using optical fibers for the fiber grating temperature sensor array, and influencing the Bragg wavelength movement amount of the fiber grating sensor through the temperature change and the mechanical strain change;
s13, temperature compensation is carried out on the Bragg wavelength movement amount by adopting an external force method, the change of the piezoelectric sensor is used for detecting the change of the optical fiber and is reduced and transferred to the change of the piezoelectric sensor for detecting the temperature, the sum of the Bragg wavelength movement amounts is detected, and the temperature field distribution data of the radiator and the cooling triangle are obtained.
2. The intelligent method for identifying the abnormal state of the indirect air-cooling system according to claim 1, wherein the step of judging the accuracy rate of the operation state of the heat-radiating surface according to the intelligent algorithm and providing the quantitative index for identifying the accuracy rate further comprises the following steps:
s21, screening temperature field data measured by a radiating surface and temperature measurement data of an inlet and an outlet of a radiator, and eliminating abnormal data;
s22, establishing feature sets for the radiators and corresponding temperature fields in different running states;
s23, dividing the data subjected to feature selection into a training sample and a test sample;
s24, training the model by using the training samples, and performing parameter optimization on the XGboost algorithm model through grid search and cross validation to obtain a GC-XGboost algorithm model;
s25, testing the model obtained in the step S24 by using the test sample;
s26, transversely comparing the calculation results of the GC-XGboost algorithm model with a support vector machine, and checking the superiority of the used model through calculating and identifying accuracy.
3. The intelligent method for the identification of the abnormal state of the indirect air-cooling system of claim 2, wherein the step of dividing the data after the feature selection into the training sample and the testing sample further comprises the steps of:
s231, acquiring temperature field creation feature set data corresponding to the radiator, and performing data preprocessing of cleaning interpolation on the temperature field creation feature set data;
s232, modeling an XGboost model, training the XGboost model and the data, and recording a predicted evaluation index value;
and S233, enumerating all the parameters of the XGboost model to form a grid, scoring all the models of the grid based on all the parameter combinations by using cross validation, scoring the XGboost model of the parameter combination with the highest score, and obtaining the GC-XGboost model.
4. An intelligent method for identification of abnormal conditions of an indirect air-cooling system of claim 3, wherein the feature set data comprises a training set and a testing set.
5. An intelligent method for the abnormal state identification of the indirect air cooling system according to claim 3, wherein the optimized objective function modeled by the XGboost model is as follows:
in the formula, obj (t) Representing the objective function of the t-th training, y i The true value of the ith sample is represented,
is a predicted value of the model of the t-1 th round, l represents a loss function, f i (x i ) Representing the input as x i Function value of time t-th round, Ω (f) i ) Denotes a regularization term, C denotes a constant, x i Data processed in step S231 is shown.
6. An intelligent method for recognizing the abnormal state of an indirect air cooling system according to claim 3, wherein the transverse comparison between the calculation results of the GC-XGBoost algorithm model and the support vector machine is used, and the superiority of the model used in the application is verified by calculating the recognition accuracy, further comprising the following steps:
s261, dividing the support vector machine into two types of modes A 1 Training set and A 2 Training set, T = { (x) 1 ，y 1 )(x 2 ，y 2 )…(x n ，y n ) Is from mode A 1 And A 2 From a training set obtained by sampling, where x n ∈R M 、y n E {1, -1}, wherein R M Comprises A 1 Class A and A 2 Class;
s262, if x n Belong to A 1 Class, then correspond to y n ＝1；
S263, if x n Belong to A 2 Class, then correspond to y n ＝-1；
S264, seeking R M The real function g (x) is transversely compared with the parameters of the GC-XGboost algorithm model, and the comparison result is as follows;
Wherein sgn { } represents a sign function, and g (x) represents a decision classification function;
and S265, checking the superiority of the used model by calculating the identification accuracy.
7. An intelligent method for recognizing abnormal conditions of an indirect air-cooling system according to claim 1, wherein the actual operating conditions of the heat sink include the following operating conditions: the two radiators work normally, the two radiators are abnormal, one radiator is abnormal, and the other radiator works normally;
wherein the abnormal state comprises isolation, freezing and stopping.
8. An intelligent method for identification of abnormal conditions of an indirect air-cooling system according to claim 1, wherein the step of forming a decision tree model by listing the combination of the operating conditions of the actuators and determining the operating conditions of the sectors by combining the temperature field distribution data further comprises the steps of:
s41, determining all segmentation points of the sample characteristics through the combination of the operation states of the actuators in a list, and segmenting each determined segmentation point;
s42, selecting an optimal segmentation point through a scoring function;
and S44, comparing the cutting point temperature of the optimal cutting point with the point temperatures of the sectors, and judging the running state of each point position of the sectors according to the comparison result.
9. An intelligent method for the abnormal state identification of an indirect air-cooling system according to claim 8, wherein the cutting standard formula of the cutting point is as follows:
in the formula, gain represents the difference between a single node obj and a tree obj of two nodes after segmentation, the segmentation points of all the characteristics are traversed, the segmentation point of the maximum Gain is found, namely the optimal segmentation point, wherein obj represents a target function, G L 、G R 、H L And H R The first-order second-order derivatives of the left and right subtrees of the current node are represented, L and R represent constants, and lambda represents a regularization coefficient;
if the gamma value is set to be too large, gain is negative, the node is not segmented, and the segmented tree structure is poor; the larger the value of γ, the more stringent the requirement for the descending amplitude of obj after slicing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210915864.0A CN115342680B (en) | 2022-08-01 | 2022-08-01 | Intelligent method for identifying abnormal state of indirect air cooling system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210915864.0A CN115342680B (en) | 2022-08-01 | 2022-08-01 | Intelligent method for identifying abnormal state of indirect air cooling system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115342680A true CN115342680A (en) | 2022-11-15 |
| CN115342680B CN115342680B (en) | 2023-07-04 |
Family
ID=83950111
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210915864.0A Active CN115342680B (en) | 2022-08-01 | 2022-08-01 | Intelligent method for identifying abnormal state of indirect air cooling system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115342680B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557886A (en) * | 2013-10-29 | 2014-02-05 | 东南大学 | Method for simultaneously measuring distribution of high-temperature gas two-dimensional transient temperature field and concentration field based on hyper spectrum |
| CN110779379A (en) * | 2019-11-08 | 2020-02-11 | 陕西能源麟北发电有限公司 | Intelligent monitoring system for temperature field of cooling tower between smoke tower and integrated steel structure |
| CN112560316A (en) * | 2020-12-21 | 2021-03-26 | 北京航空航天大学 | Correction method for surface temperature field of space target |
| CN112905947A (en) * | 2021-02-02 | 2021-06-04 | 浙江浙能技术研究院有限公司 | Real-time monitoring method for dirt degree of finned tube heat exchanger of indirect air cooling tower |
| CN113154930A (en) * | 2021-04-28 | 2021-07-23 | 国能龙源电力技术工程有限责任公司 | Steel structure indirect cooling tower temperature field monitoring system under 5G mode |
| CN114036861A (en) * | 2021-10-25 | 2022-02-11 | 中国石油化工股份有限公司 | Three-dimensional temperature field prediction method based on infoGAN |
| US20220113420A1 (en) * | 2019-11-25 | 2022-04-14 | Goertek Inc. | Plane detection method and device based on laser sensor |
| WO2022099713A1 (en) * | 2020-11-16 | 2022-05-19 | 天华化工机械及自动化研究设计院有限公司 | Three-dimensional simulation method for tow heating process in low temperature carbonization furnace based on overset model |
-
2022
- 2022-08-01 CN CN202210915864.0A patent/CN115342680B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557886A (en) * | 2013-10-29 | 2014-02-05 | 东南大学 | Method for simultaneously measuring distribution of high-temperature gas two-dimensional transient temperature field and concentration field based on hyper spectrum |
| CN110779379A (en) * | 2019-11-08 | 2020-02-11 | 陕西能源麟北发电有限公司 | Intelligent monitoring system for temperature field of cooling tower between smoke tower and integrated steel structure |
| US20220113420A1 (en) * | 2019-11-25 | 2022-04-14 | Goertek Inc. | Plane detection method and device based on laser sensor |
| WO2022099713A1 (en) * | 2020-11-16 | 2022-05-19 | 天华化工机械及自动化研究设计院有限公司 | Three-dimensional simulation method for tow heating process in low temperature carbonization furnace based on overset model |
| CN112560316A (en) * | 2020-12-21 | 2021-03-26 | 北京航空航天大学 | Correction method for surface temperature field of space target |
| CN112905947A (en) * | 2021-02-02 | 2021-06-04 | 浙江浙能技术研究院有限公司 | Real-time monitoring method for dirt degree of finned tube heat exchanger of indirect air cooling tower |
| CN113154930A (en) * | 2021-04-28 | 2021-07-23 | 国能龙源电力技术工程有限责任公司 | Steel structure indirect cooling tower temperature field monitoring system under 5G mode |
| CN114036861A (en) * | 2021-10-25 | 2022-02-11 | 中国石油化工股份有限公司 | Three-dimensional temperature field prediction method based on infoGAN |
Non-Patent Citations (3)
| Title |
|---|
| 张涛;沈云翔;: "一种新型直接空冷凝汽器可视化温度监控系统", 电气传动自动化 * |
| 李昊;罗云;李瑞东;苏永健;陈雪林;徐义巍;郭洪远;李鹏竹;: "基于粒子群优化算法优化BP神经网络模型的间接空冷散热器性能监测", 动力工程学报 * |
| 梁涛;刘晓玲;薛广伟;张力;: "直接空冷机组冷端温度场监测及防冻措施", 应用能源技术 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115342680B (en) | 2023-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108536971B (en) | Bayesian model-based structural damage identification method | |
| CN109102101B (en) | Wind speed prediction method and system for wind power plant | |
| CN111648992B (en) | Gas turbine compressor fault identification early warning method | |
| CN112052564B (en) | Heat exchanger predictive maintenance method and system based on machine learning | |
| CN111879349A (en) | Sensor data deviation self-adaptive correction method | |
| CN115796059B (en) | Electrical equipment service life prediction method and system based on deep learning | |
| CN111695744B (en) | Maintenance equipment demand prediction analysis system based on big data | |
| Kalogirou et al. | Development of a neural network-based fault diagnostic system for solar thermal applications | |
| CN111898725A (en) | Air conditioning system sensor fault detection method and device and electronic equipment | |
| Nguyen et al. | New methodology for improving the inspection policies for degradation model selection according to prognostic measures | |
| CN110737975B (en) | Wind power plant wind speed and power prediction and abnormality correction method based on empirical mode decomposition and autoregressive model | |
| CN115342680B (en) | Intelligent method for identifying abnormal state of indirect air cooling system | |
| CN112016241B (en) | Soft measurement modeling method based on mLASSO-MLP model | |
| CN117075634A (en) | Power distribution network multi-unmanned aerial vehicle scheduling inspection method and device based on improved ant colony algorithm | |
| Zhou et al. | Quantification of value of information associated with optimal observation actions within partially observable Markov decision processes | |
| CN112800672B (en) | Evaluation method, system, medium and electronic equipment for boiler fouling coefficient | |
| Cirera et al. | Data-driven operation performance evaluation of multi-chiller system using self-organizing maps | |
| CN112988529A (en) | Method and system for predicting database system performance based on machine learning | |
| CN113836480B (en) | Heat exchanger efficiency prediction method based on Gaussian process regression | |
| CN116258467B (en) | Electric power construction management and control system | |
| CN117575568A (en) | Big data-based power equipment maintenance method, equipment and medium | |
| Belhaj Salem et al. | Prognostic and Classification of the Dynamic Degradation in a Mechanical System Using Variance Gamma Process. Mathematics 2021, 9, 254 | |
| CN116804534B (en) | Auxiliary determination method for continuous beam bridge construction process | |
| CN112365103B (en) | Method, device, equipment and medium for predicting short-term load of building in operation stage | |
| CN117708543A (en) | Accurate prediction method and system for electric load of central heating of electric heating boiler |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |