CN109978048B - Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower - Google Patents

Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower Download PDF

Info

Publication number
CN109978048B
CN109978048B CN201910223650.5A CN201910223650A CN109978048B CN 109978048 B CN109978048 B CN 109978048B CN 201910223650 A CN201910223650 A CN 201910223650A CN 109978048 B CN109978048 B CN 109978048B
Authority
CN
China
Prior art keywords
circulating pump
slurry circulating
neural network
data
fault
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.)
Active
Application number
CN201910223650.5A
Other languages
Chinese (zh)
Other versions
CN109978048A (en
Inventor
竺森林
杨路宽
王铁民
李光雷
吴晔
陆忠东
赵海江
李光辉
冯永超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Datang Environment Industry Group Co Ltd
Original Assignee
Datang Environment Industry Group Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Datang Environment Industry Group Co Ltd filed Critical Datang Environment Industry Group Co Ltd
Priority to CN201910223650.5A priority Critical patent/CN109978048B/en
Publication of CN109978048A publication Critical patent/CN109978048A/en
Application granted granted Critical
Publication of CN109978048B publication Critical patent/CN109978048B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Artificial Intelligence (AREA)
  • Human Resources & Organizations (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Tourism & Hospitality (AREA)
  • Probability & Statistics with Applications (AREA)
  • Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Marketing (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Development Economics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Treating Waste Gases (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention relates to a method for analyzing and diagnosing faults of a slurry circulating pump of a desulfurizing tower, which comprises the following steps: collecting historical data, and carrying out normalization processing on the processed data; constructing training and predicting samples, and learning by adopting a BP neural network method; completing the training of a BP neural network, and constructing a current analysis system of a slurry circulating pump motor; the current of the slurry circulating pump motor is clustered by combining with the actual desulfurization efficiency; constructing a probabilistic neural network model according to the classification result; utilizing a probability neural network model to diagnose the current of the slurry circulating pump motor in real time and giving the probability of the fault; and (4) combining the fault prediction and diagnosis to construct a fault analysis and diagnosis system of the slurry circulating pump. The method is suitable for solving the problems of fault analysis and diagnosis of the slurry circulating pump in the wet desulphurization system, can help operators to predict the types and trends of faults which may occur in the future by using the real-time operation data of the system, and has high practical application value.

Description

Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower
Technical Field
The invention relates to a failure analysis and diagnosis method of a desulfurizing tower slurry circulating pump based on Bayesian classification rules and a neural network, and belongs to the field of equipment credibility analysis, equipment failure analysis and equipment failure diagnosis.
Background
The pollution of sulfur to the environment is large, the pollution of sulfur oxide and hydrogen sulfide to the atmosphere, and the pollution of sulfate and hydrogen sulfide to the water body are the key points of the current environment protection work. In the production process of a thermal power plant, a large amount of coal is combusted, and the coal often contains a certain amount of sulfur elements which release a large amount of SO after combustion2If the desulfurization agent is not treated, the desulfurization agent can cause great harm to the environment, so that the desulfurization equipment is generally used in the power industry at present.
Limestone/gypsum wet desulphurization is to add water into limestone powder to prepare slurry which is used as an absorbent and pumped into an absorption tower to be fully contacted and mixed with flue gas, and sulfur dioxide in the flue gas is oxidized with calcium carbonate in the slurry and air blown from the lower part of the tower to generate calcium sulfate. And (4) removing fog drops from the desulfurized flue gas through a demister, heating the flue gas through a heat exchanger, and exhausting the flue gas into the atmosphere through a chimney.
The slurry circulating pump can achieve the desulfurization effect of the absorbent slurry in the absorption tower in a mode of repeatedly contacting with the flue gas. The conventional method for analyzing and diagnosing the faults of the slurry circulating pump wastes time and labor, depends on the operation and operation experience of field operators, and has poor timeliness, rapidity and effectiveness of fault analysis and diagnosis.
Disclosure of Invention
In order to solve the above problems, the present invention provides a method for analyzing and diagnosing a fault of a slurry circulation pump of a desulfurization tower.
The purpose of the invention is realized by the following technical scheme:
a fault analysis and diagnosis method for a desulfurization tower slurry circulating pump comprises the following steps:
step 1, collecting historical operation data and historical monitoring data of a desulfurizing tower, and selecting motor current of a slurry circulating pump set, slurry spraying amount of a slurry circulating pump and a flue outlet SO2Concentration as the main physical quantity;
step 2, removing historical operating data and historical monitoring data of the current of a motor of a backup pump of the slurry circulating pump group, and performing normalization processing on the screened data;
step 3, constructing a BP neural network training and predicting sample;
step 4, motor current of the slurry circulating pump set, slurry spraying amount of the slurry circulating pump and flue outlet SO2Historical data of concentration is used as input, the motor current of the diagnosed slurry circulating pump is used as output, a BP neural network is adopted for learning and predicting, the training process of the BP neural network is completed, and the prediction result does not exceed a preset error;
step 5, constructing a slurry circulating pump motor current analysis system based on the BP neural network according to the steps 1 to 4;
step 6, clustering the motor current of the slurry circulating pump by combining the concentration of the SO2 at the flue outlet and the current analysis system of the slurry circulating pump motor in the step 5, and providing basic information for neurons of a pattern layer in the probabilistic neural network;
step 7, constructing a failure prediction algorithm of the probabilistic neural network based on the Bayesian classification rule, and determining the structure of the probabilistic neural network and the configuration of input parameters;
step 8, constructing a slurry circulating pump fault diagnosis system based on a Bayesian classification rule probability neural network according to the steps 5 to 7;
and 9, diagnosing the current of the motor of the slurry circulating pump in real time by using the fault diagnosis system of the slurry circulating pump in the step 8, and giving the probability of the fault of the slurry circulating pump.
Further, the step 1 of collecting the historical operation data and the historical monitoring data of the desulfurizing tower is as follows: the current of a motor of a slurry circulating pump set, the slurry spraying amount of a slurry circulating pump and the SO at the outlet of a flue2The historical data of the concentration is recorded into the database at time intervals of once every 5 seconds.
Further, the method for eliminating historical operation data and historical monitoring data of the current of the motor of the backup pump of the slurry circulating pump group in the step 2 is as follows: and taking the range of the motor current of the slurry circulating pump under the normal working condition as a reference, regarding the current signal which deviates from the range for a long time as the current signal of the motor of the slurry circulating pump which is not put into operation, and rejecting the current signal in a data screening stage.
Further, the method for constructing the training and prediction sample by the BP neural network in step 3 is as follows: dividing the data normalized in the step 2 into two parts: training data and validation data; the training data is used for training the BP neural network, and the verification data does not participate in the training of the BP neural network and is used for verifying the training effect of the BP neural network.
Further, the step 4, the step of predicting that the result of the prediction does not exceed the preset error specifically includes: and applying the trained BP neural network to verification data, and controlling the error between the output predicted by applying the BP neural network and the actual output in the collected data of the power plant to be within 5 percent.
Further, in step 7, the method for determining the structure of the probabilistic neural network and the configuration of the input parameters is as follows: dividing the probabilistic neural network into 4 layers including an input layer, a mode layer, a summation layer and an output layer according to a Bayes minimum risk criterion; the input layer receives historical data values of the motor current of the slurry circulating pump and data values predicted based on the BP neural network, the mode layer calculates the historical data values and the predicted data values of the motor current of the slurry circulating pump of the input layer and concentration values of SO2 at an outlet of a flue in a training set, the historical data values and the predicted data values and the concentration values are matched with fault characteristics obtained through clustering, the summing layer obtains estimated probability density functions under various matching relations, each neuron in the output layer corresponds to one data type, namely the matching relation, and finally a fault probability result is output.
Further, the BP neural network learning process includes the steps of:
1) initializing;
2) selecting a mode pair Ak and Yk to provide for a network;
3) using input pattern Ak and connection weight { WijCalculating the input S of each unit of the middle layerjThen using { SjCalculating output of each unit of the middle layer by an S-shaped function { b }j}；
Figure 39746DEST_PATH_IMAGE001
Figure 691307DEST_PATH_IMAGE002
4) With output b of the intermediate layerjAnd connection weights WijCalculating the inputs of the units of the output layer (L)tThen with { L }tCalculating the response of each unit of the output layer through an S-shaped function
Figure 143148DEST_PATH_IMAGE003
；
Figure 522045DEST_PATH_IMAGE004
Figure 805259DEST_PATH_IMAGE005
5) With desired output pattern YkAnd network real output
Figure 299826DEST_PATH_IMAGE006
Calculating generalized errors for cells of an output layer
Figure 43223DEST_PATH_IMAGE007
；
Figure 710965DEST_PATH_IMAGE008
6) Using the connection weight { U }, the generalized error of the output layer { d }jAnd output of the middle layer bjCalculating the generalized error of each unit of the middle layer
Figure 51947DEST_PATH_IMAGE009
；
Figure 28999DEST_PATH_IMAGE010
7) By generalising errors of cells of the output layer
Figure 455433DEST_PATH_IMAGE011
And output of each unit of the middle layer bjRevise the connection weight
Figure 661286DEST_PATH_IMAGE012
；
Figure 669825DEST_PATH_IMAGE013
8) Generalized error of each unit by using intermediate layer
Figure 568510DEST_PATH_IMAGE014
And input A of each unit of the input layerkModified connection weights wij}；
Figure 482240DEST_PATH_IMAGE015
9) Selecting the next learning mode pair to provide for the network, and returning to the step 3) until all m mode pairs are trained;
10) randomly selecting one mode pair from the m learning mode pairs again, and returning to the step 3) until the network global error function E is smaller than a preset minimum value;
11) and finishing the learning.
The invention has the beneficial effects that:
the invention constructs a desulfurization tower slurry circulating pump fault analysis and diagnosis system based on Bayesian classification rules and a neural network, combines the Bayesian classification rules with a BP neural network prediction technology, provides a fault analysis and diagnosis method combining a probabilistic neural network and the BP neural network, and can achieve the effects of accuracy and rapidity. The system for analyzing and diagnosing the faults of the slurry circulating pump of the desulfurization tower is suitable for solving the problems of analyzing and diagnosing the faults of the slurry circulating pump in a wet desulfurization system, can help operators to predict the types and trends of the faults which may occur in the future by utilizing real-time operation data of the system without other preconditions, so that the system can be conveniently responded before the faults occur, reduces the loss and has high practical application value.
When the probabilistic neural network is adopted to carry out dynamic fault analysis and diagnosis on the slurry circulating pump, historical data and data predicted by the BP neural network are used as the input of the probabilistic neural network, so that the overall accuracy of the model is improved.
The method for analyzing and diagnosing the faults of the slurry circulating pump of the desulfurization tower based on the Bayesian classification rules and the neural network has the characteristics of no need of establishing an actual physical model, high judgment speed and the like, can provide a set of simple methods for fault prediction, and is favorable for improving the desulfurization operation level.
Drawings
FIG. 1 is a neural network combinatorial model employed by the present invention;
FIG. 2 is a predicted output of a BP neural network model designed based on the method of the present invention, trained in a desulfurization system of a three gorges power plant;
FIG. 3 shows a prediction error obtained by training a BP neural network model designed based on the method of the present invention in a desulfurization system of a three gorges power plant;
FIG. 4 is a graph of the effect and error of fault analysis obtained by training a probabilistic neural network model designed based on the method in a three gorges power plant desulfurization system;
FIG. 5 shows the predicted effect based on the Bayesian classification rules and the neural network model designed based on the method of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a failure analysis and diagnosis method of a desulfurizing tower slurry circulating pump based on Bayesian classification rules and a neural network, which comprises the following steps: collecting historical data, and carrying out normalization processing on the processed data; constructing training and predicting samples, and learning by adopting a BP neural network method; completing the training of a BP neural network, and constructing a current analysis system of a slurry circulating pump motor; the current of the slurry circulating pump motor is clustered by combining with the actual desulfurization efficiency; giving out clustering results, and classifying according to the probability of faults; constructing a probabilistic neural network model according to the classification result; utilizing a probability neural network model to diagnose the current of the slurry circulating pump motor in real time and giving the probability of the fault; and (4) combining the fault prediction and diagnosis to construct a fault analysis and diagnosis system of the slurry circulating pump.
The method comprises the following specific steps:
a fault analysis and diagnosis method for a desulfurization tower slurry circulating pump comprises the following steps:
step 1, collecting historical operation data and historical monitoring data of a desulfurizing tower, and selecting motor current of a slurry circulating pump set, slurry spraying amount of a slurry circulating pump and a flue outlet SO2Concentration as the main physical quantity.
The method for collecting the historical operation data and the historical monitoring data of the desulfurizing tower comprises the following steps: the current of a motor of a slurry circulating pump set, the slurry spraying amount of a slurry circulating pump and the SO at the outlet of a flue2The historical data of the concentration is recorded into the database at time intervals of once every 5 seconds.
And 2, removing historical operating data and historical monitoring data of the motor current of the backup pump of the slurry circulating pump group, and performing normalization processing on the screened data.
The method for eliminating historical operating data and historical monitoring data of the current of the motor of the backup pump of the slurry circulating pump group comprises the following steps: and taking the range of the motor current of the slurry circulating pump under the normal working condition as a reference, regarding the current signal which deviates from the range for a long time as the current signal of the motor of the slurry circulating pump which is not put into operation, and rejecting the current signal in a data screening stage. The normalization treatment refers to the motor current of the slurry circulating pump set, the slurry spraying amount of the slurry circulating pump and the SO at the flue outlet2The three physical quantities of concentration are different in dimension and unit, and in order to prevent the phenomenon that a large number eats a small number in the training process by adopting the neural network, the number of the large number needs to be logarithmized in advanceAnd carrying out normalization processing.
And 3, constructing a BP neural network training and predicting sample.
The method for constructing the training and prediction sample by the BP neural network comprises the following steps: dividing the data normalized in the step 2 into two parts: training data and validation data; the training data is used for training the BP neural network, and the verification data does not participate in the training of the BP neural network and is used for verifying the training effect of the BP neural network.
Step 4, motor current of the slurry circulating pump set, slurry spraying amount of the slurry circulating pump and flue outlet SO2And (3) taking historical data of the concentration as input, taking the motor current of the slurry circulating pump to be diagnosed as output, learning and predicting by adopting a BP (back propagation) neural network, and finishing the training process of the BP neural network, wherein the prediction result does not exceed a preset error.
The BP neural network learning process comprises the following steps:
1) initializing;
2) selecting a mode pair Ak and Yk to provide for a network;
3) using input pattern Ak and connection weight { WijCalculating the input S of each unit of the middle layerjThen using { SjCalculating output of each unit of the middle layer by an S-shaped function { b }j}；
Figure 209893DEST_PATH_IMAGE016
Figure 525468DEST_PATH_IMAGE017
4) With output b of the intermediate layerjAnd connection weights WijCalculating the inputs of the units of the output layer (L)tThen with { L }tCalculating the response of each unit of the output layer through an S-shaped function
Figure 77279DEST_PATH_IMAGE018
；
Figure 540621DEST_PATH_IMAGE019
Figure 291539DEST_PATH_IMAGE020
5) With desired output pattern YkAnd network real output
Figure 507626DEST_PATH_IMAGE021
Calculating generalized errors for cells of an output layer
Figure 13694DEST_PATH_IMAGE022
；
Figure 636436DEST_PATH_IMAGE023
6) Using the connection weight { U }, the generalized error of the output layer { d }jAnd output of the middle layer bjCalculating the generalized error of each unit of the middle layer
Figure 738515DEST_PATH_IMAGE024
；
Figure 825420DEST_PATH_IMAGE025
7) By generalising errors of cells of the output layer
Figure 236810DEST_PATH_IMAGE026
And output of each unit of the middle layer bjRevise the connection weight
Figure 596116DEST_PATH_IMAGE027
；
Figure 485574DEST_PATH_IMAGE028
8) Using the generalized error { ek j } of each cell in the middle layer and the input A of each cell in the input layerkModified connection weights wij}；
Figure 692565DEST_PATH_IMAGE029
9) Selecting the next learning mode pair to provide for the network, and returning to the step 3) until all m mode pairs are trained;
10) randomly selecting one mode pair from the m learning mode pairs again, and returning to the step 3) until the network global error function E is smaller than a preset minimum value;
11) and finishing the learning.
The specific error that the prediction result does not exceed the preset error is as follows: and applying the trained BP neural network to verification data, and controlling the error between the output predicted by applying the BP neural network and the actual output in the collected data of the power plant to be within 5 percent.
And 5, constructing a slurry circulating pump motor current analysis system based on the BP neural network according to the steps 1 to 4.
Step 6, combining the flue outlet SO2And 5, clustering the motor current of the slurry circulating pump by the concentration and current analysis system of the slurry circulating pump in the step 5.
And 7, constructing a fault prediction algorithm of the probabilistic neural network based on the Bayesian classification rule, and determining the structure of the probabilistic neural network and the configuration of input parameters.
The method for determining the structure of the probabilistic neural network and the configuration of the input parameters comprises the following steps: dividing the probabilistic neural network into 4 layers including an input layer, a mode layer, a summation layer and an output layer according to a Bayes minimum risk criterion; the input layer receives historical data values and predicted data values of the current of the motor of the slurry circulating pump, and the mode layer calculates input data of the input layer and trains the concentrated flue outlet SO2Matching the concentration, obtaining the estimated probability density function under various matching relations by the summation layer, wherein each neuron in the output layer corresponds to a data type, namely, matchingAnd matching relation.
And 8, constructing a slurry circulating pump fault diagnosis system based on the Bayesian classification rule probability neural network according to the steps 5 to 7.
And 9, diagnosing the current of the motor of the slurry circulating pump in real time by using the fault diagnosis system of the slurry circulating pump in the step 8, and giving the probability of the fault of the slurry circulating pump.
As shown in fig. 1, the present invention comprises the steps of:
firstly, a non-linear mathematical model is established for physical quantities strongly related to the faults of the slurry circulating pump, and the motor current of the slurry circulating pump set, the slurry circulating pump slurry spraying quantity and the flue outlet SO are processed by adopting a BP neural network model2The historical data of the concentration is recorded into the database at time intervals of once every 5 seconds.
Historical operating data and historical monitoring data of the motor current of the backup pump of the slurry circulating pump group are removed, and the screened data are normalized.
After preprocessing a plurality of groups of historical data, 1900 groups of data are selected as the input of a BP neural network for training so as to improve the precision of the model, 100 groups of data are selected for the verification of the neural network, and the training is completed when the total error is less than 5%.
The current of the motor of the slurry circulating pump reflects the working state of the slurry circulating pump to a great extent, and further reflects the desulfurization efficiency. And clustering the current of the slurry circulating pump motor by means of the concentration of sulfur dioxide at the outlet of the flue in historical operating data.
And constructing a probabilistic neural network model according to the clustering result. And dividing the clustering result into four modes of normal operation, fault precursor, slight abnormity and obvious abnormity.
The test set is sent to the constructed probabilistic neural network model for testing, and the accuracy of the prediction result is high as can be seen from FIG. 4.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (6)

1. A fault analysis and diagnosis method for a slurry circulating pump of a desulfurization tower is characterized by comprising the following steps:
step 1, collecting historical operation data and historical monitoring data of a desulfurizing tower, and selecting motor current of a slurry circulating pump set, slurry spraying amount of a slurry circulating pump and a flue outlet SO2Concentration as the main physical quantity;
step 2, removing historical operating data and historical monitoring data of the current of a motor of a backup pump of the slurry circulating pump group, and performing normalization processing on the screened data;
step 3, constructing a BP neural network training and predicting sample;
step 4, motor current of the slurry circulating pump set, slurry spraying amount of the slurry circulating pump and flue outlet SO2Historical data of concentration is used as input, the motor current of the diagnosed slurry circulating pump is used as output, a BP neural network is adopted for learning and predicting, the training process of the BP neural network is completed, and the prediction result does not exceed a preset error;
step 5, constructing a slurry circulating pump motor current analysis system based on the BP neural network according to the steps 1 to 4;
step 6, clustering the motor current of the slurry circulating pump by combining the concentration of the SO2 at the flue outlet and the current analysis system of the slurry circulating pump motor in the step 5, and providing basic information for neurons of a pattern layer in the probabilistic neural network;
step 7, constructing a failure prediction algorithm of the probabilistic neural network based on the Bayesian classification rule, and determining the structure of the probabilistic neural network and the configuration of input parameters;
step 8, constructing a slurry circulating pump fault diagnosis system based on a Bayesian classification rule probability neural network according to the steps 5 to 7;
and 9, diagnosing the current of the motor of the slurry circulating pump in real time by using the fault diagnosis system of the slurry circulating pump in the step 8, and giving the probability of the fault of the slurry circulating pump.
2. The method for analyzing and diagnosing faults of a slurry circulating pump of a desulfurization tower as recited in claim 1, wherein the step 1 of collecting historical operation data and historical monitoring data of the desulfurization tower comprises the following steps: the current of a motor of a slurry circulating pump set, the slurry spraying amount of a slurry circulating pump and the SO at the outlet of a flue2The historical data of the concentration is recorded into the database at time intervals of once every 5 seconds.
3. The method for analyzing and diagnosing the fault of the slurry circulating pump of the desulfurization tower as recited in claim 1, wherein the historical operation data and the historical monitoring data of the current of the motor of the backup pump of the slurry circulating pump set in the step 2 are removed as follows: and taking the range of the motor current of the slurry circulating pump under the normal working condition as a reference, regarding the current signal which deviates from the range for a long time as the current signal of the motor of the slurry circulating pump which is not put into operation, and rejecting the current signal in a data screening stage.
4. The method for analyzing and diagnosing the fault of the slurry circulating pump of the desulfurization tower as recited in claim 1, wherein the training and prediction samples are constructed by the BP neural network in the step 3 as follows: dividing the data normalized in the step 2 into two parts: training data and validation data; the training data is used for training the BP neural network, and the verification data does not participate in the training of the BP neural network and is used for verifying the training effect of the BP neural network.
5. The method for analyzing and diagnosing the fault of the slurry circulating pump of the desulfurization tower as recited in claim 4, wherein the error of the prediction result not exceeding the preset error in the step 4 is specifically as follows: and applying the trained BP neural network to verification data, and controlling the error between the output predicted by applying the BP neural network and the actual output in the collected data of the power plant to be within 5 percent.
6. The method for analyzing and diagnosing the fault of the slurry circulating pump of the desulfurization tower as recited in claim 1, wherein the determining of the configuration of the structure and the input parameters of the probabilistic neural network in step 7 is performed as follows: dividing the probabilistic neural network into 4 layers including an input layer, a mode layer, a summation layer and an output layer according to a Bayes minimum risk criterion; the input layer receives historical data values of the motor current of the slurry circulating pump and data values predicted based on the BP neural network, the mode layer calculates the historical data values and the predicted data values of the motor current of the slurry circulating pump of the input layer and concentration values of SO2 at an outlet of a flue in a training set, the historical data values and the predicted data values and the concentration values are matched with fault characteristics obtained through clustering, the summing layer obtains estimated probability density functions under various matching relations, each neuron in the output layer corresponds to one data type, namely the matching relation, and finally a fault probability result is output.
CN201910223650.5A 2019-03-22 2019-03-22 Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower Active CN109978048B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910223650.5A CN109978048B (en) 2019-03-22 2019-03-22 Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910223650.5A CN109978048B (en) 2019-03-22 2019-03-22 Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower

Publications (2)

Publication Number Publication Date
CN109978048A CN109978048A (en) 2019-07-05
CN109978048B true CN109978048B (en) 2021-05-07

Family

ID=67080140

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910223650.5A Active CN109978048B (en) 2019-03-22 2019-03-22 Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower

Country Status (1)

Country Link
CN (1) CN109978048B (en)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110426996B (en) * 2019-07-12 2023-06-23 张鹏 Environmental pollution monitoring method based on big data and artificial intelligence
CN111013370A (en) * 2019-11-08 2020-04-17 中国大唐集团科学技术研究院有限公司火力发电技术研究院 Wet desulphurization slurry supply amount prediction method based on deep neural network
CN111736574B (en) * 2020-06-28 2023-12-05 江苏方天电力技术有限公司 Universal thermal power plant fault diagnosis system and diagnosis method thereof
CN113864208B (en) * 2020-06-30 2023-11-17 宝武装备智能科技有限公司 Intelligent monitoring and diagnosing method for operation state of desulfurization slurry circulating pump
CN112308228A (en) * 2020-11-12 2021-02-02 安徽江机重型数控机床股份有限公司 Fault detection early warning method for cutting fluid pumping system of numerical control machine tool
CN112365065B (en) * 2020-11-16 2024-07-02 大唐环境产业集团股份有限公司 WFGD self-adaptive online optimization scheduling method
CN112747011B (en) * 2020-12-29 2023-07-07 广东精铟海洋工程股份有限公司 Fault prediction method based on pile gripper hydraulic system and pile gripper hydraulic system
CN112906967B (en) * 2021-02-24 2024-06-14 大唐环境产业集团股份有限公司 Method and device for predicting performance of slurry circulating pump of desulfurization system
CN113762345A (en) * 2021-08-06 2021-12-07 国网冀北电力有限公司经济技术研究院 Oil-immersed transformer fault diagnosis method and device
CN115326147B (en) * 2022-10-17 2023-11-24 江苏煵笙重工有限公司 Real-time monitoring method and system for desulfurizing tower for ship
CN116236892A (en) * 2023-03-16 2023-06-09 福建龙净环保股份有限公司 Method and device for controlling desulfurization equipment

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101561442A (en) * 2009-05-11 2009-10-21 江南大学 Restructured Pichia pastoris in expression period two-phase on line fault diagnostic method based on artificial neural network
US20140039817A1 (en) * 2012-08-06 2014-02-06 Phase3 Technologies Ltd. System and method for monitoring an electrically-connected system having a periodic behavior
CN102788014B (en) * 2012-08-08 2015-04-22 徐州重型机械有限公司 Gear oil pump tooth surface wearing fault diagnosis device and method
US20150122037A1 (en) * 2013-10-30 2015-05-07 Syncrude Canada Ltd. In Trust For The Owners Of The Syncrude Project Method for diagnosing faults in slurry pump impellers
CN105782071B (en) * 2016-03-04 2018-07-13 长沙有色冶金设计研究院有限公司 A kind of water-isolation slurry pump method for diagnosing faults based on probabilistic neural network
CN108152612A (en) * 2017-11-23 2018-06-12 全球能源互联网研究院有限公司 A kind of transformer fault Forecasting Methodology, device, terminal and readable storage medium storing program for executing
CN208207843U (en) * 2018-04-10 2018-12-07 西南石油大学 A kind of slush pump trouble-shooter
CN109061341A (en) * 2018-07-10 2018-12-21 杭州安脉盛智能技术有限公司 Kalman filtering transformer fault prediction technique and system neural network based

Also Published As

Publication number Publication date
CN109978048A (en) 2019-07-05

Similar Documents

Publication Publication Date Title
CN109978048B (en) Fault analysis and diagnosis method for slurry circulating pump of desulfurizing tower
Liu et al. Optimized approximation algorithm in neural networks without overfitting
CN114191953A (en) Flue gas desulfurization and denitrification control method based on convolutional neural network and XGboost
CN112733876A (en) Soft measurement method for nitrogen oxides in urban solid waste incineration process based on modular neural network
CN112364562B (en) Flue gas environment-friendly island cooperative control method and system
CN113094988A (en) Data-driven slurry circulating pump operation optimization method and system
CN111190349A (en) Method, system and medium for monitoring state and diagnosing fault of ship engine room equipment
CN116688754A (en) Ship flue gas desulfurization automatic control system and method thereof
Wang et al. A remaining useful life prediction model based on hybrid long-short sequences for engines
Zafari et al. Modeling and optimization of CO2 capture into mixed MEA-PZ amine solutions using machine learning based on ANN and RSM models
CN117150677A (en) Method and device for determining type selection parameters of electric dust collector
CN116227689A (en) Photovoltaic panel pollution degree state evaluation and cleaning decision system
CN113112072A (en) NOx emission content prediction method based on deep bidirectional LSTM
CN115964503B (en) Safety risk prediction method and system based on community equipment facilities
CN117113166A (en) Industrial boiler fault detection method based on improved integrated learning
CN116862081A (en) Operation and maintenance method and system for pollution treatment equipment
CN115983048A (en) Condenser quality detection method and system based on digital twinning
Ren et al. A short-term wind speed prediction method based on interval type 2 fuzzy model considering the selection of important input variables
Abate Formal verification of complex systems: Model-based and data-driven methods
Duer Qualitative evaluation of the regeneration process of a technical object in a maintenance system with an artificial neural network
Shojaee et al. Applying SVSSI sampling scheme to np-chart to decrease the time of detecting shifts using Markov chain approach and Monte Carlo simulation
CN112016248A (en) Elman neural network based SCR denitration system bad data identification method
Utkin et al. A fuzzy software reliability model with multiple-error introduction and removal
CN116304663B (en) Train control vehicle-mounted equipment health state management device based on unbalanced sample enhancement
CN111598220B (en) Gas turbine fault prediction method based on correlation analysis

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