CN112489841A - Water level fault-tolerant control method for steam generator of nuclear power unit - Google Patents

Water level fault-tolerant control method for steam generator of nuclear power unit Download PDF

Info

Publication number
CN112489841A
CN112489841A CN202011320311.8A CN202011320311A CN112489841A CN 112489841 A CN112489841 A CN 112489841A CN 202011320311 A CN202011320311 A CN 202011320311A CN 112489841 A CN112489841 A CN 112489841A
Authority
CN
China
Prior art keywords
fault
tolerant
controller
training
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011320311.8A
Other languages
Chinese (zh)
Inventor
梁军
彭嘉恒
李贤民
栾振华
范家钰
刘潇
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.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
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 Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202011320311.8A priority Critical patent/CN112489841A/en
Publication of CN112489841A publication Critical patent/CN112489841A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21D—NUCLEAR POWER PLANT
    • G21D3/00—Control of nuclear power plant
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21C—NUCLEAR REACTORS
    • G21C9/00—Emergency protection arrangements structurally associated with the reactor, e.g. safety valves provided with pressure equalisation devices
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21D—NUCLEAR POWER PLANT
    • G21D3/00—Control of nuclear power plant
    • G21D3/001—Computer implemented control
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21D—NUCLEAR POWER PLANT
    • G21D3/00—Control of nuclear power plant
    • G21D3/04—Safety arrangements
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21D—NUCLEAR POWER PLANT
    • G21D3/00—Control of nuclear power plant
    • G21D3/04—Safety arrangements
    • G21D3/06—Safety arrangements responsive to faults within the plant
    • 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
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E30/00—Energy generation of nuclear origin
    • 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
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E30/00—Energy generation of nuclear origin
    • Y02E30/30—Nuclear fission reactors

Abstract

The invention discloses a water level fault-tolerant control method of a nuclear power unit steam generator, which comprises the steps of continuously acquiring data of a complete water level control loop, particularly a detection instrument and an adjusting valve, carrying out online fault diagnosis according to an acquired data set, judging the type and the degree of a fault generated by the detection instrument or the adjusting valve, carrying out fault-tolerant reconstruction decision of a controller according to the type and the degree of the fault, reconstructing a controller structure and parameters to counteract the influence of the fault, finally calculating a fault-tolerant control instruction under the condition of the fault, and issuing the fault-tolerant control instruction to a digital instrument control system of the nuclear power unit through a field bus. The invention makes up the defect that the existing digital instrument control system of the nuclear power unit can not deal with the partial failure of the instrument and the regulating valve, and ensures the operation safety of the nuclear power unit.

Description

Water level fault-tolerant control method for steam generator of nuclear power unit
Technical Field
The invention relates to the technical field of safe operation control of nuclear power plants, in particular to a method for realizing a fault-tolerant controller for controlling the water level of a steam generator of a pressurized water reactor nuclear power unit.
Background
At present, digital instrument control systems (DCS for short) are adopted in the existing global in-service nuclear power generating units and newly-built nuclear power generating units so as to realize the functions of centralized monitoring, remote operation, automatic control and protection of the whole plant process. However, the successful control of the DCS on the nuclear power plant is based on the precondition that the detecting instruments (such as temperature detection, pressure detection, water level detection, etc.) and the actuators (such as regulating valves, switch valves, various pumps, motors, etc.) in the control circuit can operate correctly. Once the detecting instruments and the executing mechanisms are in failure, the control function of the DCS system is partially or completely failed, and catastrophic risks are caused to the operation of the nuclear power generating units.
Since the accident of the nuclear power plant in fukushima, research on safe operation of nuclear power generating units has been strengthened in main nuclear power countries in the world, and improvement of safety of a control system represented by DCS is a very important aspect. However, from the current public reports and research and analysis on domestic and foreign in-service nuclear power stations, although the DCS realizes perfect control and protection tasks, the safety countermeasures for the fault occurrence of the detecting instrument, the regulating valve and the like under the DCS platform have not been successfully applied. The present invention aims to solve this serious problem of high security risk.
Disclosure of Invention
The invention aims to provide a fault-tolerant controller aiming at the defect that a DCS (distributed control system) cannot ensure safe operation when faults of a detection instrument, a regulating valve and the like occur in a water level control loop of a steam generator of a nuclear power unit in the prior art.
The purpose of the invention is realized by the following technical scheme: a fault-tolerant controller for nuclear power unit steam generator water level control includes the following steps:
step 1: collecting data such as flow of a detection instrument, opening of an adjusting valve and the like in a water level control loop, and performing denoising and normalization standardization pretreatment on the data to obtain a sample data set;
step 2: performing online fault diagnosis by using an improved fault diagnosis algorithm based on the probability principal component analysis of the instant learning model according to the sample data set obtained in the step 1, and judging the type and degree of the fault generated by the detection instrument or the regulating valve so as to obtain a fault diagnosis result;
and step 3: according to the fault diagnosis result obtained in the step 2, a fault-tolerant control algorithm is used for signal reconstruction of a fault detection instrument, fault-tolerant reconstruction of a controller structure facing to the fault of the regulating valve and optimal estimation of fault tolerance of controller parameters, an optimized fault-tolerant controller new structure and new parameters are obtained, and the fault influence is counteracted or reduced;
and 4, step 4: according to the new structure and the new parameters of the fault-tolerant controller obtained in the step 3, calculating a fault-tolerant control instruction under the condition of a fault (the calculation mode of the control instruction is divided into two conditions: 1) when the instrument fault is detected, replacing the signal reconstruction value obtained in the step 3 with the instrument signal with the fault to calculate the control instruction, and keeping the structure of the controller unchanged; 2) and (3) when the regulating valve has a fault, calculating a control instruction by adopting the new structure and the new parameters of the fault-tolerant controller obtained in the step (3), and then issuing the control instruction to a digital instrument control system of the nuclear power generating unit through a field bus to execute the control instruction so as to meet the control requirement.
Further, the step 1 comprises the following sub-steps:
and (1.1) continuously sampling all variables of the steam generator water level control loop by means of DCS to form a sample data set. The time length of the data set is determined in advance according to the dynamic time constant of the water level process, and old data exceeding the time length are removed successively to keep the data length unchanged;
(1.2) in order to ensure the quality of the sample data set, preprocessing is required, and the preprocessing mainly comprises processing of outlier data points, processing of random noise items in the data, interpolation processing of missing data points, normalization processing of the data and the like. In the invention, outliers are removed by adopting a statistical variance analysis method, the removed outliers are interpolated by adopting a power equality principle, high-frequency noise is filtered by adopting a low-pass noise filter method, missing data points are interpolated by adopting a least square fitting method, and the standard processing is carried out by adopting the statistical operation of a mean value and a variance.
Further, the step 2 comprises the following sub-steps:
(2.1) dividing the sample data set obtained in the step (1) into a training data set and a testing data set, wherein the training data set is used for training an off-line model based on the probability principal component analysis of the instant learning model, and the testing data set is used for monitoring an on-line process;
(2.2) performing off-line model training on the training data set determined in (2.1), and firstly calculating a training sample xiAnd other training samples xjSimilarity between them Si,jI is the ith training sample, j represents other training samples except i, the similarity among all samples is subjected to descending order arrangement, samples corresponding to the first L similarities are selected as related training samples, the value range of L is 10-100, and offline training of a local least square support vector regression algorithm model is carried out;
(2.3) carrying out model updating judgment on the test data set determined in (2.1), firstly, calculating a test sample x at the current momentk' and test sample x at the previous timek-1' similarity between Sk,k-1Comparing the current time k and the previous time k-1 with the minimum value of the similarity obtained in the offline training model, judging whether the training model needs to be updated, if the current time k and the previous time k-1 are lower than a threshold value, keeping (2.2) the trained local least square support vector regression model, and if not, carrying out (2.4) step;
(2.4) calculating the test sample xk' and all training samples xiSimilarity between them Sk,iI refers to all training samples, the similarity among all samples is subjected to descending order arrangement, the first L corresponding samples are selected as related training samples, and a local least square support vector regression model is updated to replace the original local least square support vector regression model (2.2);
and (2.5) judging whether the current working condition has a fault according to a detection statistic threshold value obtained by the probability principal component analysis model and the local least square support vector regression model obtained in the step (2.3) or (2.4), obtaining a fault diagnosis result, and transmitting the fault diagnosis result to the fault-tolerant control module.
Further, the step 3 comprises the following sub-steps:
(3.1) determining typical instrumentation failure SF for steam generator level control processiI 1,2, … m and regulator valve fault VFjJ ═ 1,2, … n, i.e., m typical instrumentation faults and n typical regulator valve faults are considered;
(3.2) signal reconstruction of the fault detection instrument: for each typical instrumentation failure SFiAnd i is 1,2, … m, a partial least square method PLS is adopted, and a soft measurement model MSF of the detecting instrument under the fault condition is established off lineiAnd i is 1,2 and … m, and is used for reconstructing signals of the fault detection instrument. The method comprises the following steps that m soft measurement models are provided, and each soft measurement model corresponds to a typical detection instrument fault;
(3.3) fault-tolerant reconstruction of a controller structure and optimal estimation of controller parameter fault tolerance: for each typical regulator valve failure VFiAnd i is 1,2, … n, and determining that the regulating valve can realize a base-guaranteed control fault-tolerant controller structure CVF (continuously variable frequency) under the fault condition offline by adopting a method of combining simulation and actual experimentjJ is 1,2, … n, and carrying out parameter optimization on line according to the severity of the fault of the regulating valve to obtain the optimal estimated PVF of the fault-tolerant controller parametersjJ is 1,2, … n. N fault-tolerant controllers and n sets of optimal controller parameters are total, and each fault-tolerant controller and the optimal parameters thereof correspond to a typical regulating valve fault;
and (3.4) according to the results of (3.1), (3.2) and (3.3), when the fault-tolerant control system is used on line, according to the diagnosis result of the fault type in the step two, performing two-stage fault-tolerant reconstruction decision to determine which fault-tolerant reconstruction measure is adopted to complete the fault-tolerant control task. The first-level decision judges whether the current fault belongs to the fault of the detection instrument or the fault of the regulating valve, and further sets the fault in a set (MSF)iI-1, 2, … m or { CVF }jJ ═ 1,2, … n } to achieve accurate positioning; the second decision is based on the step two on the basis of the fault of the regulating valveFurther completing the parameter optimization solution of the fault-tolerant controller according to the result of the fault degree;
and (3.5) according to the result of (3.4), completing the on-line switching application of the fault-tolerant controller through a simple one-out-of-many switching logic unit.
Further, the step 4 comprises the following sub-steps:
(4.1) according to the soft measurement model reconstructed by the fault detection instrument signal obtained in the step 3, calculating the prediction output of the fault detection instrument on line, and transmitting the prediction output to a digital instrument control system DCS of the nuclear power generation unit through a field bus, wherein the structure of the controller is not changed;
and (4.2) calculating a fault-tolerant control instruction under the condition of the fault of the regulating valve according to the new structure and the new parameters of the fault-tolerant controller obtained in the step 3, transmitting the fault-tolerant control instruction to a digital instrument control system DCS of the nuclear power generating unit through a field bus, carrying out fault-tolerant reconstruction on the controller structure, and carrying out online optimal self-setting on the PID parameters to finally achieve the control requirements and the purpose.
The invention has the advantages that the fault diagnosis function is added in the water level control loop of the steam generator of the nuclear power unit, so that operators can timely master the fault conditions of the detection instrument and the regulating valve and can take reasonable prevention and control intervention measures in advance; according to the invention, a fault-tolerant control function is added in a water level control loop of a steam generator of the nuclear power unit, so that a DCS can automatically start a fault-tolerant control means to realize bottom-preserving control when a fault occurs in a detection instrument or a regulating valve, the adverse effect of the fault is inhibited, and the active safety of the nuclear power unit is improved; in general, the invention can greatly improve the overall safety of the nuclear power station.
Drawings
Fig. 1 is a schematic diagram of the general technical architecture of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Step 1: the DCS control system monitors and collects fault information and provides the fault information for the fault diagnosis module, and the method specifically comprises the following substeps:
and (1.1) continuously sampling all variables of the steam generator water level control loop by means of DCS to form a sample data set. The time length of the data set is determined in advance according to the dynamic time constant of the water level process, and old data exceeding the time length are removed successively to keep the data length unchanged;
(1.2) in order to ensure the quality of the sample data set, preprocessing is required, and the preprocessing mainly comprises processing of outlier data points, processing of random noise items in the data, interpolation processing of missing data points, normalization processing of the data and the like. In the invention, outliers are removed by adopting a statistical variance analysis method, the removed outliers are interpolated by adopting a power equality principle, high-frequency noise is filtered by adopting a low-pass noise filter method, missing data points are interpolated by adopting a least square fitting method, and the standard processing is carried out by adopting the statistical operation of a mean value and a variance.
Step 2: the fault diagnosis module extracts fault detection information to detect the failure condition of each detection table and each actuator all the time on line, once a fault symptom occurs, a fault recognition algorithm is started to diagnose the characteristics of the fault such as the type, the position, the time and the like, and the diagnosis result is sent to the fault-tolerant control module, and the fault-tolerant control module specifically comprises the following substeps:
(2.1) dividing the data set obtained in the step one into a training data set and a testing data set, wherein data under normal working conditions are used as the training data set to train the probability principal component analysis offline model based on the just-in-time learning model, and training set samples all comprise input vectors
Figure BDA0002792696930000041
And the output vector
Figure BDA0002792696930000042
For industrial data sets, operations becomeThe quantity is generally used as an input variable, and the process variable is used as an output variable; the test data set is used for on-line process monitoring and is also divided into an input variable and an output variable;
and (2.2) performing off-line model training on the training data set determined in the step (2.1), wherein the training process is shown in the attached drawing 1. First, in each cycle, a training sample x is calculatediAnd other training samples xjSimilarity between them Si,jI is the ith training sample, j represents other training samples except i, the similarity among all samples is arranged in a descending order, the first L corresponding samples are selected as related training samples to carry out off-line training of a local least square support vector regression algorithm model, wherein the value range of L is 10-100, a local least square support vector regression model is constructed, and the output vector of the current moment is predicted
Figure BDA0002792696930000043
N1After one cycle, the output prediction values of all training samples can be known
Figure BDA0002792696930000051
Residual error calculation is carried out on the residual error and the actual value Y to obtain the residual value of each training sample
Figure BDA0002792696930000052
The extraction capability of the local least square support vector regression algorithm model on the nonlinear information and the parameters gamma and sigma in the model2Closely related, the fault diagnosis module adopts a grid search method to determine the optimal parameter combination.
Then will be
Figure BDA0002792696930000053
Inputting the training set into a PPCA (probability principal component analysis) model as a training set, and calculating T corresponding to each sample by using the PPCA model after training is finished2And SPE statistics as thresholds for comparison of the model thereafter.
(2.3) carrying out model updating judgment on the test data set determined in (2.1), and firstly calculating a test sample xk' andtest sample x at a previous timek-1' similarity between Sk,k-1Comparing the current time k and the previous time k-1 with the minimum value of the similarity obtained in the offline training model, judging whether the training model needs to be updated, if the current time k and the previous time k-1 are lower than a threshold value, keeping (2.2) the trained local least square support vector regression model, and if not, carrying out (2.4) step;
(2.4) calculating the test sample xk' and all training samples xiSimilarity between them Sk,iPerforming descending order arrangement on the similarity among all samples, selecting the first L (L value range is 10-100) corresponding samples as related training samples to update the local least square support vector regression model, replacing (2.2) the original local least square support vector regression model, and updating the original similarity threshold;
(2.5) predicting the current sample x according to the detection statistic threshold value obtained by the probability principal component analysis model and the local least square support vector regression model obtained in the step (2.3) or (2.4)iOutput of (2)
Figure BDA0002792696930000054
The corresponding T is calculated by using PPCA model2And SPE statistic is compared with a trained threshold, if the current value is smaller than the threshold, the current value is a normal working condition, if the current value is larger than the threshold, the current value is a fault working condition, and the obtained fault diagnosis result is transmitted to the fault-tolerant control module.
And step 3: the fault-tolerant control module receives the fault characteristics provided by the fault diagnosis module, determines to start a proper fault-tolerant control strategy, and further strategies the fault-tolerant control strategy into a specific fault-tolerant control instruction, and specifically comprises the following substeps:
(3.1) determining typical instrumentation failure SF for steam generator level control processiI 1,2, … m and regulator valve fault VFjJ ═ 1,2, … n, i.e., m typical instrumentation faults and n typical regulator valve faults were considered. The number and the type of the faults of the detecting instrument and the number and the type of the faults of the regulating valve in the sub-step generally depend on long-term operation experience, fault maintenance records and enterprise operation and maintenance specialtiesHome confirmation;
(3.2) signal reconstruction of the fault detection instrument: for each typical instrumentation failure SFiAnd i is 1,2, … m, a partial least square method PLS is adopted, and a soft measurement model MSF of the detecting instrument under the fault condition is established off lineiAnd i is 1,2 and … m, and is used for reconstructing signals of the fault detection instrument. The method comprises the following steps that m soft measurement models are provided, and each soft measurement model corresponds to a typical detection instrument fault. In this sub-step, the soft measurement model based on PLS is completed by the following sub-steps:
(3.2.1) for each instrumentation, selecting sufficient process variables associated with the instrumentation output as input variables for the soft measurement model to form an input vector Xm(ii) a Selecting the output of the detecting instrument as the output variable Y of the soft measuring modelm；
(3.2.2) collecting a set of soft measurement modeling sample sets from the Normal running Process
Figure BDA0002792696930000061
Modeling sample matrices XX and YY are constructed, where,
Figure BDA0002792696930000062
Figure BDA0002792696930000063
is an input vector XmThe sample value of the ith sample point of (1),
Figure BDA0002792696930000064
is an output variable YmThe sample value of the ith sample point of (1), NN is the number of sample points in the modeling sample set;
(3.2.3) carrying out standardization and normalization processing on the modeling sample matrixes XX and YY to enable the mean value of each variable to be 0 and the variance to be 1, and obtaining an input matrix X and an output matrix Y;
(3.2.4) splitting the input matrix X and the output matrix Y into a training sample matrix X1，Y1And a test sample matrix X2，Y2Adopts partial least square method based on multivariate statistical projection principleEstablishing soft measurement model in training sample
Figure BDA0002792696930000065
Wherein theta is a soft measurement model parameter matrix, X1Is an input matrix of the training samples and,
Figure BDA0002792696930000066
the soft measurement prediction value of an output matrix Y of a training sample is obtained, and model verification is carried out in a test sample;
(3.2.5) substituting the current measurement data matrix into the soft measurement model in the sub-step (3.2.4) to perform prediction calculation, and calculating the prediction result
Figure BDA0002792696930000067
And carrying out inverse standardization and inverse normalization processing to obtain a predicted value of the output signal of the detection instrument.
(3.3) fault-tolerant reconstruction of a controller structure and optimal estimation of controller parameter fault tolerance: for each typical regulator valve failure VFiAnd i is 1,2, … n, and determining that the regulating valve can realize a base-guaranteed control fault-tolerant controller structure CVF (continuously variable frequency) under the fault condition offline by adopting a method of combining simulation and actual experimentjJ is 1,2, … n (the meaning of the guaranteed-base control is that the control loop can be kept stable and the basic control performance can be kept, the basic control performance is prescribed by a control engineer in advance), and the optimal estimated PVF of the fault-tolerant controller parameters is obtained by carrying out parameter optimization on line according to the severity of the fault of the regulating valvejJ is 1,2, … n. N fault-tolerant controllers and n sets of optimal controller parameters are total, and each fault-tolerant controller and the optimal parameters thereof correspond to a typical regulating valve fault;
in the sub-step, the fault-tolerant reconstruction of the controller structure and the optimal estimation of the fault tolerance of the controller parameter are completed through the following sub-steps:
(3.3.1) the fault-tolerant reconstruction of the controller structure means that in a water level control system of a steam generator of a nuclear power unit, aiming at each typical regulating valve fault VFiI 1,2, … m, each regulator valve fault failure is artificially set one by one at the fault tolerant controller design stage (the present invention assumes each time that each regulator valve fault failure is presentNext time only one regulating valve fails, all other regulating valves are normal), and determining the fault-tolerant controller structure which can still realize the bottom-guaranteed control by the rest intact regulating valves through a verification mode combining simulation and practice. Thus, for all regulator valve faults VFiI is 1,2, … m, and there are n fault-tolerant controller structures CVFj,j＝1,2,…n；
(3.3.2) Fault-tolerant optimal estimation of controller parameters means that for n fault-tolerant controller architectures CVFjJ is 1,2, … n, and in the use stage of the fault-tolerant controller, the parameters of the fault-tolerant controller are optimized on line according to the severity of the current fault of the regulating valve to obtain the optimal parameters PVF of the fault-tolerant controllerjJ is 1,2, … n. Because n fault-tolerant controllers CVF in the inventionjJ is 1,2, … n, which adopts PID control law (same as DCS control mode of nuclear power plant, easy to use), the controller parameters are specified as PID parameters, and the fault-tolerant optimal estimation of the parameters can adopt the following method: according to the diagnosis result of the fault type and degree of the regulating valve, an intelligent optimization self-setting algorithm of PID parameters is started on line, and a setting optimization objective function can adopt an absolute deviation integral criterion according to the role of the fault regulating valve in a control loop
Figure BDA0002792696930000071
Square deviation integral criterion
Figure BDA0002792696930000072
Time by absolute deviation integration criterion
Figure BDA0002792696930000073
Time-by-square deviation integration criterion
Figure BDA0002792696930000074
Any one or a weighted combination of some (e (t)) is a control error, and a Particle Swarm Optimization (PSO) algorithm is adopted by an optimization solver;
and (3.4) performing two-stage fault-tolerant reconstruction decision to perform fault type diagnosis according to the diagnosis result of the step 2 when the online fault-tolerant reconstruction method is used online according to the results of the steps (3.1), (3.2) and (3.3)And determining which fault-tolerant reconfiguration measure is adopted to complete the fault-tolerant control task. The first-level decision judges whether the current fault belongs to the fault of the detection instrument or the fault of the regulating valve, and further sets the fault in a set (MSF)iI-1, 2, … m or { CVF }jJ ═ 1,2, … n } to achieve accurate positioning; the second-level decision is to further complete the optimization solution of the fault-tolerant controller parameters according to the result of the step 2 on the fault degree on the basis of determining the fault of the regulating valve;
and (3.5) according to the result of (3.4), completing the on-line switching application of the fault-tolerant controller through a simple one-out-of-many switching logic unit.
And 4, step 4: the obtained new structure and new parameters of the fault-tolerant controller are used for calculating a fault-tolerant control instruction under the condition of a fault and are transmitted to a digital instrument control system of a nuclear power unit through a field bus, and the method specifically comprises the following substeps:
and (4.1) calculating the prediction output of the fault detection instrument on line according to the soft measurement model reconstructed by the fault detection instrument signal obtained in the step (3). The calculation method comprises the following steps: according to the latest sample data set (composed of the detection values of related fault-free instruments and various intermediate calculation values) obtained currently, after standardization, the data passes through a soft measurement model in a sub-step (3.2.4)
Figure BDA0002792696930000075
Calculating and then using the result of the inverse standardization of the sub-step (3.2.5) as a prediction value of the fault meter; finally, a digital instrument control system DCS of the nuclear power generating unit is issued through a field bus, and the structure of a controller is not changed;
and (4.2) calculating a fault-tolerant control instruction under the condition of the fault of the regulating valve according to the new structure and the new parameters of the fault-tolerant controller obtained in the step 3. The calculation method comprises the following steps: 1) calculating a current control error e (t); 2) according to the fault-tolerant controller structure (PID mode, the same as the mode of the conventional DCS) determined in the step (3.5), calculating the current fault-tolerant control instruction
Figure BDA0002792696930000076
Figure BDA0002792696930000077
Wherein, Kp、Ti、TdIs the optimal PID parameter determined by sub-step (3.3.2); 3) and a digital instrument control system DCS of the nuclear power generating unit is issued through a field bus. Thus, the controller structure carries out fault-tolerant reconstruction, PID parameters carry out on-line optimal self-tuning, and finally the control requirement and the purpose are achieved.

Claims (3)

1. A water level fault-tolerant control method for a nuclear power unit steam generator is provided with a water level control loop, and the water level control loop is provided with a detection instrument and a regulating valve. The method is characterized by comprising the following steps:
step 1: collecting data such as flow of a detection instrument, opening of an adjusting valve and the like in a water level control loop, and performing denoising and normalization standardization pretreatment on the data to obtain a sample data set;
step 2: performing online fault diagnosis by using an improved fault diagnosis algorithm based on the probability principal component analysis of the instant learning model according to the sample data set obtained in the step 1, and judging the type and degree of the fault generated by the detection instrument or the regulating valve so as to obtain a fault diagnosis result;
and step 3: performing signal reconstruction of the fault detection instrument by using a fault-tolerant control algorithm according to the fault diagnosis result obtained in the step 2 to obtain a signal reconstruction value; the fault-tolerant reconstruction of the controller structure and the optimal estimation of the fault tolerance of the controller parameters are carried out for the faults of the regulating valve, so as to obtain the new structure and the new parameters of the optimized fault-tolerant controller and counteract or reduce the influence of the faults;
and 4, step 4: calculating a fault-tolerant control instruction under the condition of a fault according to the new structure and the new parameters of the fault-tolerant controller obtained in the step 3, wherein the calculation mode of the control instruction is divided into two conditions: (a) when the instrument is detected to be in fault, the instrument signal in fault is replaced by the signal reconstruction value obtained in the step (3) to calculate a control instruction, and the structure of the controller is not changed; (b) when the regulating valve is in fault, calculating a control instruction by adopting the new structure and the new parameters of the fault-tolerant controller obtained in the step 3; the control instruction is transmitted to a digital instrument control system of the nuclear power unit through a field bus to execute the control instruction, so that the control requirement is met.
2. The steam generator water level fault tolerant controller of claim 1, characterized in that said step 2 is implemented by the following sub-steps:
(2.1) dividing the sample data set obtained in the step (1) into a training data set and a testing data set, wherein the training data set is used for training an off-line model based on the probability principal component analysis of the instant learning model, and the testing data set is used for monitoring an on-line process;
(2.2) performing off-line model training on the training data set determined in (2.1), and firstly calculating a training sample xiAnd other training samples xjSimilarity between them Si,jI is the ith training sample, j represents other training samples except i, the similarity among all samples is subjected to descending order arrangement, samples corresponding to the first L similarities are selected as related training samples, the value range of L is 10-100, and offline training of a local least square support vector regression algorithm model is carried out;
(2.3) carrying out model updating judgment on the test data set determined in (2.1), firstly, calculating a test sample x at the current momentk' and test sample x at the previous timek-1' similarity between Sk,k-1Comparing the current time k and the previous time k-1 with the minimum value of the similarity obtained in the offline training model, judging whether the training model needs to be updated, if the current time k and the previous time k-1 are lower than a threshold value, keeping (2.2) the trained local least square support vector regression model, and if not, carrying out (2.4) step;
(2.4) calculating the test sample xk' and all training samples xiSimilarity between them Sk,iI refers to all training samples, the similarity among all samples is subjected to descending order arrangement, the first L corresponding samples are selected as related training samples, and a local least square support vector regression model is updated to replace the original local least square support vector regression model (2.2);
and (2.5) judging whether the current working condition has faults or not according to the detection statistic threshold value obtained by the probability principal component analysis model and the local least square support vector regression model obtained in the step (2.3) or (2.4), and obtaining a fault diagnosis result.
3. The steam generator water level fault tolerant controller of claim 1, characterized in that said step 3 is implemented by the following sub-steps:
(3.1) determining typical instrumentation failure SF for steam generator level control processiI 1,2, … m and regulator valve fault VFjJ ═ 1,2, … n, i.e., m typical instrumentation faults and n typical regulator valve faults are considered;
(3.2) signal reconstruction of the fault detection instrument: for each typical instrumentation failure SFiAnd i is 1,2, … m, a partial least square method PLS is adopted, and a soft measurement model MSF of the detecting instrument under the fault condition is established off lineiAnd i is 1,2 and … m, and is used for reconstructing signals of the fault detection instrument. The method comprises the following steps that m soft measurement models are provided, and each soft measurement model corresponds to a typical detection instrument fault;
(3.3) fault-tolerant reconstruction of a controller structure and optimal estimation of controller parameter fault tolerance: for each typical regulator valve failure VFiAnd i is 1,2, … n, and determining that the regulating valve can realize a base-guaranteed control fault-tolerant controller structure CVF (continuously variable frequency) under the fault condition offline by adopting a method of combining simulation and actual experimentjJ is 1,2, … n, and carrying out parameter optimization on line according to the severity of the fault of the regulating valve to obtain the optimal estimated PVF of the fault-tolerant controller parametersjJ is 1,2, … n. N fault-tolerant controllers and n sets of optimal controller parameters are total, and each fault-tolerant controller and the optimal parameters thereof correspond to a typical regulating valve fault;
and (3.4) according to the results of (3.1), (3.2) and (3.3), when the fault-tolerant control system is used on line, according to the diagnosis result of the fault type in the step 2, performing two-stage fault-tolerant reconstruction decision to determine which fault-tolerant reconstruction measure is adopted to complete the fault-tolerant control task. The first-level decision judges whether the current fault belongs to the fault of the detection instrument or the fault of the regulating valve, and further sets the fault in a set (MSF)iI-1, 2, … m or { CVF }jJ ═ 1,2, … n } to achieve accurate positioning; first, theThe secondary decision is to further complete the optimization solution of the fault-tolerant controller parameters according to the result of the step 2 on the fault degree on the basis of determining the fault of the regulating valve;
and (3.5) completing the on-line switching application of the fault-tolerant controller through a one-out-of-more switching logic unit according to the result of (3.4).
CN202011320311.8A 2020-11-23 2020-11-23 Water level fault-tolerant control method for steam generator of nuclear power unit Pending CN112489841A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011320311.8A CN112489841A (en) 2020-11-23 2020-11-23 Water level fault-tolerant control method for steam generator of nuclear power unit

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011320311.8A CN112489841A (en) 2020-11-23 2020-11-23 Water level fault-tolerant control method for steam generator of nuclear power unit

Publications (1)

Publication Number Publication Date
CN112489841A true CN112489841A (en) 2021-03-12

Family

ID=74933479

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011320311.8A Pending CN112489841A (en) 2020-11-23 2020-11-23 Water level fault-tolerant control method for steam generator of nuclear power unit

Country Status (1)

Country Link
CN (1) CN112489841A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113655714A (en) * 2021-07-02 2021-11-16 中国科学院西安光学精密机械研究所 Parameter self-tuning method for control system
CN115307670A (en) * 2022-10-12 2022-11-08 蘑菇物联技术（深圳）有限公司 Method, apparatus and medium for locating central air conditioning system anomaly sensors

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5192493A (en) * 1989-08-30 1993-03-09 Westinghouse Electric Corp. Median signal selector for feedwater control systems
CN1737423A (en) * 2005-08-10 2006-02-22 东北大学 Method and apparatus for realizing integration of fault-diagnosis and fault-tolerance for boiler sensor based on Internet
CN105137957A (en) * 2015-06-23 2015-12-09 黄红林 Control method of nuclear power steam generator
CN109917777A (en) * 2019-04-16 2019-06-21 浙江科技学院 Fault detection method based on mixing multi-sampling rate Probabilistic Principal Component Analysis model

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5192493A (en) * 1989-08-30 1993-03-09 Westinghouse Electric Corp. Median signal selector for feedwater control systems
CN1737423A (en) * 2005-08-10 2006-02-22 东北大学 Method and apparatus for realizing integration of fault-diagnosis and fault-tolerance for boiler sensor based on Internet
CN105137957A (en) * 2015-06-23 2015-12-09 黄红林 Control method of nuclear power steam generator
CN109917777A (en) * 2019-04-16 2019-06-21 浙江科技学院 Fault detection method based on mixing multi-sampling rate Probabilistic Principal Component Analysis model

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ZIYANG WANG ET AL.: "A JITL-Based Probabilistic Principal Componet Analysis for Online Monitoring of Nonlinear Processes", 《JOURNAL OF CHEMICAL ENGINEERING OF JAPAN》 *
朱少民: "蒸汽发生器水位控制系统容错控制方法研究", 《中国优秀博硕士学位论文全文数据库工程科技Ⅱ辑》 *
黄增辉: "基于数据的风车故障诊断与容错控制", 《中国优秀博硕士学位论文全文数据库工程科技Ⅱ辑》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113655714A (en) * 2021-07-02 2021-11-16 中国科学院西安光学精密机械研究所 Parameter self-tuning method for control system
CN113655714B (en) * 2021-07-02 2023-01-06 中国科学院西安光学精密机械研究所 Parameter self-tuning method for control system
CN115307670A (en) * 2022-10-12 2022-11-08 蘑菇物联技术（深圳）有限公司 Method, apparatus and medium for locating central air conditioning system anomaly sensors
CN115307670B (en) * 2022-10-12 2023-01-13 蘑菇物联技术（深圳）有限公司 Method, apparatus and medium for locating central air conditioning system anomaly sensors

Similar Documents

Publication Publication Date Title
CN108596229B (en) Method and system for monitoring and diagnosing online abnormity
US7818276B2 (en) Intelligent monitoring system and method for building predictive models and detecting anomalies
Lindemann et al. Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks
EP1982301B1 (en) Method of condition monitoring
CN109977624A (en) Photovoltaic plant soft fault monitoring method based on deep neural network
CN112489841A (en) Water level fault-tolerant control method for steam generator of nuclear power unit
CN104390657A (en) Generator set operating parameter measuring sensor fault diagnosis method and system
CN111474919B (en) Aeroengine control system sensor fault diagnosis method based on AANN network group
Guo et al. Observer-based optimal fault detection and diagnosis using conditional probability distributions
CN114239377A (en) Method and system for evaluating health state of urban rail electromechanical equipment and storage medium
Mathew et al. Regression kernel for prognostics with support vector machines
CN110513336B (en) Method for determining offline water washing time of gas turbine of power station
CN116431966A (en) Reactor core temperature anomaly detection method of incremental characteristic decoupling self-encoder
KR102156858B1 (en) Method for diagnosing and predicting robot arm's failure
CN111241629B (en) Intelligent prediction method for performance change trend of hydraulic pump of airplane based on data driving
US20230022100A1 (en) Prognostic and health management system for system management and method thereof
Rodriguez-Guerra et al. Fault-tolerant control study and classification: Case study of a hydraulic-press model simulated in real-time
CN116151312A (en) Thermal turbine unit early warning method and system based on Mann-Kendall algorithm and LSTM neural network
CN114115198A (en) Assembly production line-oriented distributed diagnosis and optimization control method and control system
CN110717630B (en) Reliability prediction method and device for turbine overspeed protection system
Zhang Comparison of data-driven and model-based methodologies of wind turbine fault detection with SCADA data
José et al. Improvements in failure detection of DAMADICS control valve using neural networks
CN112666418A (en) Power transmitter fault diagnosis method based on water head, opening and active power joint curve
Kim et al. Anomaly Detection of an Air Compressor from Time-series Measurement Data
CN112116003B (en) Self-adaptive dynamic fault diagnosis method and system for complex sensor network

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
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210312

WD01 Invention patent application deemed withdrawn after publication