CN109238760B - Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis - Google Patents
Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis Download PDFInfo
- Publication number
- CN109238760B CN109238760B CN201811056765.1A CN201811056765A CN109238760B CN 109238760 B CN109238760 B CN 109238760B CN 201811056765 A CN201811056765 A CN 201811056765A CN 109238760 B CN109238760 B CN 109238760B
- Authority
- CN
- China
- Prior art keywords
- monitoring
- slow
- matrix
- residual
- variable
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
- G01M99/005—Testing of complete machines, e.g. washing-machines or mobile phones
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Disintegrating Or Milling (AREA)
Abstract
The invention discloses an online monitoring method of a coal mill of a coal-fired power plant generator set based on typical correlation analysis and slow characteristic analysis. Aiming at the problems of numerous process variables and complex regulation effect of a controller of the coal mill, the invention utilizes typical variables to analyze and extract information of the time sequence correlation relationship of the process, and simultaneously combines a slow characteristic analysis algorithm to extract the dynamic and static characteristics of the process, synthesizes the time sequence correlation relationship and the dynamic and static characteristics of the regulation effect of a closed loop system and the correlation of the running state, and establishes dynamic and static online monitoring indexes in different subspaces to monitor the process of the coal mill. The method has the advantages that the regulation effect of the closed-loop system can be fully reflected, the complex process characteristics can be known, the reliability and the credibility of the online process monitoring of the coal mill are enhanced, the accurate judgment of the running state of the coal mill by a thermal power plant is facilitated, the fault is found in time, and the safe and reliable running of the coal mill is ensured.
Description
Technical Field
The invention belongs to the field of process monitoring of a thermal power process, and particularly relates to an online process monitoring method for dynamic and static coordination of a coal mill by considering the regulation effect of a closed loop system, comprehensively analyzing the time sequence correlation of process variables and the dynamic and static characteristics of the process.
Background
The electric power industry in China is rapidly developed, and mainly thermal power is used, and coal is the main fuel of a thermal power plant. According to data, the average power supply coal consumption per kilowatt hour in China is about 50g higher than that of standard coal in developed countries, which shows that the actual operation state and the design state of the unit are greatly different. The large-scale power failure accidents in europe and north america in 2003 remind people that the human life at the present stage cannot be powered, the electric energy becomes a part of the human life, and the loss of national economy caused by power failure far exceeds the loss of a power system, so that the safe and reliable operation of power generation equipment is particularly important. With the rapid development of the power industry in China and the continuous development of the thermal power generation technology, the capacity of a generator set is continuously increased, the complexity is continuously improved, and the operation information needing to be processed and judged is gradually increased, so that the problems of the reliability, the availability and the safety of some key equipment in a thermal power plant are increasingly highlighted. With the continuous development and application of thermal power generation high-parameter and high-capacity units, corresponding auxiliary equipment tends to be large-sized gradually, and the function of the auxiliary equipment in the production process of a thermal power plant is more important.
According to data display of unit participation reliability statistics in recent years, if five auxiliary machines in power equipment, such as a coal mill, a feed water pump, a blower, an induced draft fan and a high-pressure heater, break down, the safety and economic operation of the whole unit is affected, even safety hazards can be brought to field personnel, and the importance of the auxiliary machines is not inferior to three main machines, namely a boiler, a steam turbine and a generator. In order to discover and prevent equipment faults, ensure safe and stable continuous operation of the equipment, reduce accidents and ensure safe and economical operation. Monitoring and analyzing the running state of the equipment are needed.
The coal mill of the thermal power plant is used as a core device of a boiler combustion pulverizing system, is a mechanical device for crushing and grinding coal blocks into coal powder, and has important influence on the safety and the economical efficiency of the operation of the whole power plant system due to the working condition of the mechanical device. Such as mill pressure, ventilation, mill load, etc., which all affect production efficiency. Coal mills must generally operate in a harsh environment, the operation process is a complicated process, the failure rate of the coal mill is high, and the failure of the coal mill causes double losses of life safety and economic benefits. For example, 6 months and 19 days in 2003, the gravity center of an inverted cone (935.2kg) of a lifting separator in a coal mill of a cool power plant of Gansu thermal power engineering company is shifted, so that 1 person dies; in 2016, 2 and 25 days, the hot smoke and wind burning accident of the Datang Changshan thermal power plant occurs in a No. 1 furnace C coal mill, and 3 people die. Therefore, it is important to ensure safe and reliable operation of coal mills of thermal power plants.
Aiming at the problems of complex system and diversified process variables of the coal mill, the predecessors have made corresponding research and discussion from different angles, and provide a corresponding online process monitoring method. In general, analytical model-based methods and data-driven-based methods are mainly included. The disadvantages of the model-based approach are mainly: the method is mainly applied to linear time-invariant systems, but most systems and devices are nonlinear and time-variant. The research and application of data-driven industrial process monitoring to the last century in the 90 s have been rising, and the deepening of industrial automation and the continuous improvement of informatization level enable a great amount of process data to be stored and utilized. Accordingly, data-driven based process monitoring methods are gaining increasing attention from researchers. However, the existing data analysis methods, such as principal component analysis, partial least squares, and fisher discriminant analysis, only consider static information of the process, and ignore the influence of the regulation effect of the closed-loop system in the process on the process state. Therefore, the application of the method to a coal mill cannot achieve a good monitoring effect. The invention further considers the influence of the adjusting action of the closed-loop system on the monitoring performance in the actual process operation of the coal mill, comprehensively analyzes the time sequence correlation of process variables and the dynamic and static characteristics of the process, and provides a novel online monitoring method for the dynamic and static characteristic collaborative analysis of the coal mill.
Disclosure of Invention
The invention aims to provide an online monitoring method of a coal mill of an intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis aiming at the defects of the existing process monitoring method of the coal mill, the method fully considers the influence of the closed-loop system regulation action on the process characteristics, mainly reflects the change of correlation and the dynamic characteristics of the process, firstly analyzes and extracts the time sequence related information of the process by using typical variables, meanwhile, extracts the dynamic information of the process by combining with a slow characteristic analysis algorithm, establishes dynamic and static online monitoring indexes in different subspaces to carry out process detection on the coal mill, and realizes the accurate monitoring of the running state of the coal mill.
The purpose of the invention is realized by the following technical scheme: an intelligent power plant coal-fired generator set coal pulverizer online monitoring method based on typical correlation analysis and slow characteristic analysis is disclosed, the coal pulverizer is a machine for crushing and pulverizing coal blocks, and is one of important auxiliary machines in a thermal power plant, common faults of the coal pulverizer include air leakage at an inlet and an outlet, high outlet temperature, ignition of the coal pulverizer and the like, and the method comprises the following steps:
(1) and (3) acquiring data to be analyzed, namely setting J measurement variables and operation variables in the production process of a coal mill, wherein each sampling can obtain a vector of 1 × J, and the data obtained after N times of sampling is expressed as a two-dimensional matrix X (N × J), wherein the measurement variables are process variables which can be measured in the operation process and comprise temperature, rotating speed, pressure, valve opening and the like.
(2) The method comprises the following steps of extracting the time sequence related information of the coal mill operation data based on the typical variable analysis, wherein the step is realized by the following substeps:
(2.1) extracting typical variables of the coal mill operation data, and analyzing the time sequence correlation relationship: measuring the value x at each time point t(t)(1 × J) using the past measured values x(t-1),x(t-2),…,x(t-l)And h future measurements x(t),x(t+1),…,x(t+h)Deployment of (2).
Where the subscript p represents the past and f the future. x is the number ofp(t)Representing a set of past measurements, xf(t)Representing a set of future measurements;
wherein l and h are based on the current time x(t)The correlation degree of (2) is determined as follows:
measured value x of past time spaced l +1 sampling intervals from current time tp(t+l+1)And the current time x(t)The degree of correlation of (d) is expressed as:
where α is a threshold, 0<α<0.5。autocorr(XiAnd p) is the autocorrelation coefficient of the ith process variable plus p time lags. Degree of correlation with the current time, when Al+1When it is less than α, x can be consideredp(t+l+1)And x(t)No correlation exists and does not participate in constructing the matrix xp(t). The smallest value satisfying the above formula is selected and determined as l, and h is determined as l, thereby constructing xp(t)And xf(t)。
When the time interval between a certain measured value and the current time t is larger than the determined l and h, the correlation between the variables can be ignored.
(2.2) forming the vectors spread at different time instants into a past and future matrix:
Xp=[xp(t+1),xp(t+2),…,xp(t+m)](4)
Xf=[xf(t+1),xf(t+2),…,xf(t+m)](5)
where the subscript p stands for past, f stands for future, and m-N-l-h + 1.
(2.3) solving for the typical variable z at different times t(t)And a residual variable e(t):
Wherein, ∑p,fRepresenting a past matrix XpAnd future matrix XfOf (a) covariance matrix of (a), ∑p,pRepresenting a past matrix XpThe covariance matrix of (3), superscript 1/2, denotes the operation of squaring the elements within the matrix, ∑f,fRepresenting future matrix XfThe left side of the equal sign of expression (6) is subjected to SVD decomposition to obtain orthogonal matrices U and V and a diagonal matrix Λ, the coefficient on the diagonal of which is a principal component correlation coefficient gamma1≥…≥γrThe following table r shows the number of elements on the diagonal of the diagonal matrix Λ.
Wherein J represents a transformation matrix for obtaining a correlation variable, L represents a transformation matrix for obtaining a residual scalar, I represents a unit matrix, and U representskK columns comprising matrix U, the choice of k being determined according to:
wherein β is a threshold value, 0.5 ≦ β ≦ 1.
Z=JXP(10)
E=LXP(11)
Where the matrix Z and the matrix E represent the principal component space and the residual space, respectively. Each column for Z and E is the typical variable Z(t)And residual variable e(t)。
(3) The method comprises the following steps of cooperatively monitoring the operation of the coal mill based on slow feature analysis, wherein the steps are realized by the following substeps:
(3.1) SFA modeling of principal component space Z and residual space E, respectively
sc=WcZ (12)
se=WeE (13)
Wherein s iscIs a slow feature of the extracted principal component space Z, WcIs a transformation matrix of the principal component space Z; seIs the slow feature of the extracted residual space E, WeIs the transformation matrix of the residual space E.
(3.2) selecting slow characteristic s in principal component spacecThe main slow feature of (1).
Slowly changing features can represent the general trend of process changes, while those that change more quickly can be considered noise; all slow characteristics are sorted according to the change speed, and R is selectedm(Rm<Rc) The slow characteristic with slow change is used as the main slow characteristic sc,dThe remaining slow features can be considered as noise, where RcAll slow feature numbers;
dominant slow characteristic number RmFrom the point of view of reconstruction, the selection of (c) is specifically as follows:
wherein the content of the first and second substances,is thatThe (c) th column of (a),is to beIs replaced by 0, the number of the elements is Rm,RmThe determination method of (2) is as follows: noise reducing reconstructed process variablesDegree of slowness ofHow much important information can be retained on behalf of the reconstructed process variable;should contain as little rapidly changing noise as possible, and soΔ(xj) Representing a process variable xjDegree of slowness of, noise-reduced reconstructed process variableRatio of change xjIs slow; and Δ (x)j) And Δ(s)i) In a linear relationship, Δ(s) in slow characteristic si)>Δ(xj) Part of xjThe change is accelerated, and the part is removed, so thatCan satisfyConsidering the reconstruction effect of all variables comprehensively, the feature to be removed is as follows:
number of divided main and slow characteristics Rm=Rc-cnt (F) is the total slow feature number minus the number of elements in set F; definition of WcFront R ofmBehavior Wc,d(Rm×J)。
sc,d=Wc,dZ (18)
Wherein s isc,dIs extracted slow characteristics which can represent the main trend of the change in the process in the principal component space; wc,dRepresenting a transformation matrix;
(3.3) calculating control limits for static monitoring in principal component space:
wherein the content of the first and second substances,and is sc,dT of2Monitoring statistics; determination using kernel density estimationControl limit of (Ctr)c,Td;
(3.4) calculating dynamically monitored control limits in principal component space:
wherein the content of the first and second substances,is sc,dS of2Monitoring statistics; whereinIs sc,dA first order difference of; omegac,dIs thatAn empirical covariance matrix of (2); solving using kernel density estimationControl limit of (Ctr)c,Sd;
(3.5) according to the method in (3.2), selecting slow features s in residual spaceeMain slow characteristic number R ofme: according to the degree of slowness of the features, carryTaking out seMain slow feature in (1):
(3.6) calculating the control limit of static monitoring in the residual space:
wherein the content of the first and second substances,is se,dT of2Monitoring statistics; determining T using kernel density estimation2Control limit of (Ctr)e,Td;
(3.7) calculating the control limit of dynamic monitoring in the residual space:
wherein the content of the first and second substances,is se,dS of2Monitoring statistics; whereinIs se,dA first order difference of; omegae,dIs thatAn empirical covariance matrix of (2); solving for S using kernel density estimation2Control limit of (Ctr)e,Sd;
(4) On-line monitoring of the coal mill: based on the steps, the variable correlation of the coal mill operation process data and a dynamic and static cooperative monitoring model are obtained, and the main component space and the residual space are usedThe monitoring statistics can be atMonitoring the running state of the process on line; this step is realized by the following substeps:
(4.1) acquiring and expanding newly measured data in the operation process of the coal mill: during on-line monitoring, new process measurement data x are collectednew,t(J × 1) where the index new represents the new sample, the index t represents the current time, J is the measured variable, the same as in step 1, extended with the data of the past time/times:
(4.2) solving a typical variable and a residual variable at the current moment:
znew=Jxnew(25)
enew=Lxnew(26)
(4.3) extracting respective slow feature vectors from the typical variable and the residual variable at the current moment respectively:
(4.4) calculating online static monitoring statistics in principal component space and residual space respectively:
(4.5) calculating online dynamic monitoring statistics in principal component space and residual space respectively:
(5) judging the running state of the coal mill: comparing two monitoring indexes in different subspaces with respective statistical control limits in real time:
(a) in the principal component space, if the two monitored quantities are both within the control limit range, the process has a constant time sequence correlation relationship, and the variation of the typical variable is within the normal range.
(b) In the principal component space, if the static monitoring quantity is not exceeded and the dynamic monitoring quantity is exceeded, it indicates that an abnormality of process dynamics is monitored, which may cause a steady state deviation of the variable autocorrelation relation in the principal component space.
(c) In the principal component space, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored, but the dynamic characteristic of the process is not influenced, and the process is subjected to state switching under good control.
(d) In the principal component space, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the process shows that both the steady state deviation and the dynamic abnormity occur, the process is detected to have faults and exceed the regulation capability of the control system, and the faults are processed in time.
(e) In the residual space, if both monitored quantities are within the control limits, the variation of the residual variable is within the normal range.
(f) In the residual space, if the static monitoring amount is not exceeded, the dynamic monitoring amount is exceeded, and the abnormality of the process dynamic behavior is detected, which may cause the state deviation in the residual subspace.
(g) In the residual subspace, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored in the residual space, but the dynamic characteristic of the process is not influenced, and the process is under good control.
(h) In the residual subspace, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the steady state deviation and the dynamic abnormity are detected in the residual subspace, the process has faults, the control performance is poor, and the processing is required to be carried out in time.
Compared with the prior art, the invention has the beneficial effects that: the invention provides an online monitoring method for dynamic and static characteristic collaborative analysis, which aims at a device which is a coal mill and has a complex system and is easy to break down. The method fully considers the influence of the regulation effect of the closed-loop system on the process characteristics, firstly, the typical variable is used for analyzing and extracting the time sequence related information of the process, then, the slow characteristic analysis algorithm is combined for extracting the dynamic information of the process, the coal mill is large-scale mechanical equipment, and the slow characteristic s is used for extracting the dynamic information of the processc,d,se,dCan show the main trend of the change in the process, and the monitoring can be realized by focusing on the part in the application. The method comprehensively utilizes time sequence related information and dynamic and static information related to the adjustment effect of the controller and the process running state, fully reflects the influence of the adjustment effect of a closed loop system on the process characteristics, enhances the reliability and credibility of the online process monitoring of the coal mill, is beneficial to a thermal power plant to accurately judge the running state of the coal mill, finds faults in time and ensures the safe and reliable running of the coal mill.
Description of the drawings:
FIG. 1 is a flow chart of an online monitoring method for dynamic and static characteristic collaborative analysis of a coal pulverizer, wherein (a) is a flow chart of an offline modeling process, and (b) is a flow chart of online process monitoring;
FIG. 2 is a graph of the monitoring results in principal component space of the method of the present invention in an embodiment of the present invention, (a) the monitoring results for static monitoring statistics, and (b) the monitoring results for dynamic monitoring statistics.
FIG. 3 shows the results of monitoring in residual space of the method of the present invention in an embodiment of the present invention, (a) is the result of monitoring static monitoring statistics, and (b) is the result of monitoring dynamic monitoring statistics.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific examples.
The invention takes a coal mill in No. 8 unit of Jiahua power plant belonging to Zhe energy group as an example, and comprises 14 process variables, wherein the variables relate to temperature, rotating speed, pressure, valve opening degree and the like. In the example, 21 normal process operation data in the coal mill operation process are selected.
As shown in FIG. 1, the invention relates to an online monitoring method of a coal pulverizer of a coal-fired power generating set of an intelligent power plant based on typical correlation analysis and slow characteristic analysis, which comprises the following steps:
(1) the method comprises the steps of obtaining data to be analyzed, wherein J measurable process variables are arranged in the running process of a coal mill, a vector of 1 × J can be obtained through each sampling, and the data obtained after N times of sampling are expressed as a two-dimensional matrix X (N × J). in the example, the sampling period is 10 minutes, 400 samples are collected, 14 process variables are collected, and the measured variables are temperature, rotating speed, pressure, valve opening degree and the like in the running process;
(2) the method comprises the following steps of extracting the time sequence related information of the coal mill operation data based on the typical variable analysis, wherein the step is realized by the following substeps:
(2.1) extracting typical variables of the coal mill operation data, and analyzing the time sequence correlation relationship: measuring the value x at each time point t(t)(1 × J) using the past measured values x(t-1),x(t-2),…,x(t-l)And h future measurements x(t),x(t+1),…,x(t+h)Deployment of (2).
Where the subscript p represents the past and f the future. x is the number ofp(t)Representing a set of past measurements, xf(t)Representing a set of future measurements;
wherein l and h are based on the current time x(t)The correlation degree of (2) is determined as follows:
measured value x of past time spaced l +1 sampling intervals from current time tp(t+l+1)And the current time x(t)The degree of correlation of (d) is expressed as:
where α is a threshold, 0<α<0.5。autocorr(XiAnd p) is the autocorrelation coefficient of the ith process variable plus p time lags. Degree of correlation with the current time, when Al+1When it is less than α, x can be consideredp(t+l+1)And x(t)No correlation exists and does not participate in constructing the matrix xp(t). The smallest value satisfying the above formula is selected and determined as l, and h is determined as l, thereby constructing xp(t)And xf(t)。
When the time interval between a certain measured value and the current time t is larger than the determined l and h, the correlation between the variables can be ignored.
(2.2) forming the vectors spread at different time instants into a past and future matrix:
Xp=[xp(t+1),xp(t+2),…,xp(t+m)](4)
Xf=[xf(t+1),xf(t+2),…,xf(t+m)](5)
where the subscript p stands for past, f stands for future, and m-N-l-h + 1.
(2.3) solving for the typical variable z at different times t(t)And a residual variable e(t):
Wherein, ∑p,fRepresenting a past matrix XpAnd future matrix XfOf (a) covariance matrix of (a), ∑p,pRepresenting a past matrix XpThe covariance matrix of (3), superscript 1/2, denotes the operation of squaring the elements within the matrix, ∑f,fRepresenting future matrix XfThe superscript 1/2 denotes the mapping of the elements in the matrixPerforming an extraction operation, performing SVD on the left side of the equal sign of expression (6) to obtain orthogonal matrices U and V and a diagonal matrix Λ whose diagonal coefficients are principal component correlation coefficients gamma1≥…≥γrThe following table r shows the number of elements on the diagonal of the diagonal matrix Λ.
Wherein J represents a transformation matrix for obtaining a correlation variable, L represents a transformation matrix for obtaining a residual scalar, I represents a unit matrix, and U representskK columns comprising matrix U, the choice of k being determined according to:
wherein β is a threshold value, 0.5 ≦ β ≦ 1.
Z=JXP(10)
E=LXP(11)
Where the matrix Z and the matrix E represent the principal component space and the residual space, respectively. Each column for Z and E is the typical variable Z(t)And residual variable e(t)。
(3) The method comprises the following steps of cooperatively monitoring the operation of the coal mill based on slow feature analysis, wherein the steps are realized by the following substeps:
(3.1) SFA modeling of principal component space Z and residual space E, respectively
sc=WcZ (12)
se=WeE (13)
Wherein s iscIs a slow feature of the extracted principal component space Z, WcIs a transformation matrix of the principal component space Z; seIs the slow feature of the extracted residual space E, WeIs the transformation matrix of the residual space E.
(3.2) selecting principal component spaceMedium and slow characteristic scThe main slow feature of (1).
Slowly changing features can represent the general trend of process changes, while those that change more quickly can be considered noise; all slow characteristics are sorted according to the change speed, and R is selectedm(Rm<Rc) The slow characteristic with slow change is used as the main slow characteristic sc,dThe remaining slow features can be considered as noise, where RcAll slow feature numbers;
dominant slow characteristic number RmFrom the point of view of reconstruction, the selection of (c) is specifically as follows:
wherein the content of the first and second substances,is thatThe (c) th column of (a),is to beIs replaced by 0, the number of the elements is Rm,RmThe determination method of (2) is as follows: noise reducing reconstructed process variablesDegree of slowness ofCan substitute forHow much important information is retained by the table reconstructed process variables;should contain as little rapidly changing noise as possible, and soΔ(xj) Representing a process variable xjDegree of slowness of, noise-reduced reconstructed process variableRatio of change xjIs slow; and Δ (x)j) And Δ(s)i) In a linear relationship, Δ(s) in slow characteristic si)>Δ(xj) Part of xjThe change is accelerated, and the part is removed, so thatCan satisfyConsidering the reconstruction effect of all variables comprehensively, the feature to be removed is as follows:
number of divided main and slow characteristics Rm=Rc-cnt (F) is the total slow feature number minus the number of elements in set F; definition of WcFront R ofmBehavior Wc,d(Rm×J)。
sc,d=Wc,dZ (18)
Wherein s isc,dIs extracted slow characteristics which can represent the main trend of the change in the process in the principal component space; wc,dRepresenting a transformation matrix;
(3.3) calculating control limits for static monitoring in principal component space:
wherein the content of the first and second substances,and is sc,dT of2Monitoring statistics; determination using kernel density estimationControl limit of (Ctr)c,Td;
(3.4) calculating dynamically monitored control limits in principal component space:
wherein the content of the first and second substances,is sc,dS of2Monitoring statistics; whereinIs sc,dA first order difference of; omegac,dIs thatAn empirical covariance matrix of (2); solving using kernel density estimationControl limit of (Ctr)c,sd;
(3.5) according to the method in (3.2), selecting slow features s in residual spaceeMain slow characteristic number R ofme: according to the slow degree of the characteristics, s is extractedeMain slow feature in (1):
(3.6) calculating the control limit of static monitoring in the residual space:
wherein the content of the first and second substances,is se,dT of2Monitoring statistics; determining T using kernel density estimation2Control limit of (Ctr)e,Td;
(3.7) calculating the control limit of dynamic monitoring in the residual space:
wherein the content of the first and second substances,is se,dS of2Monitoring statistics; whereinIs se,dA first order difference of; omegae,dIs thatAn empirical covariance matrix of (2); solving for S using kernel density estimation2Control limit of (Ctr)e,sd;
(4) On-line monitoring of the coal mill: based on the steps, the variable correlation of the coal mill operation process data and a dynamic and static cooperative monitoring model are obtained, and the main component space and the residual space are usedThe monitoring statistics can monitor the running state of the process on line; this step is realized by the following substeps:
(4.1) acquiring and expanding newly measured data in the operation process of the coal mill: during on-line monitoring, new process measurement data x are collectednew,t(J × 1) wherein the index new represents the new sample, the index t represents the current time, J is the measured variable, and step 1 is performedThe variables are the same; it is extended with the data of the past i time instants:
(4.2) solving a typical variable and a residual variable at the current moment:
znew=Jxnew(25)
enew=Lxnew(26)
(4.3) extracting respective slow feature vectors from the typical variable and the residual variable at the current moment respectively:
(4.4) calculating online static monitoring statistics in principal component space and residual space respectively:
(4.5) calculating online dynamic monitoring statistics in principal component space and residual space respectively:
(5) judging the running state of the coal mill: comparing two monitoring indexes in different subspaces with respective statistical control limits in real time:
(a) in the principal component space, if the two monitored quantities are both within the control limit range, the process has a constant time sequence correlation relationship, and the variation of the typical variable is within the normal range.
(b) In the principal component space, if the static monitoring quantity is not exceeded and the dynamic monitoring quantity is exceeded, it indicates that an abnormality of process dynamics is monitored, which may cause a steady state deviation of the variable autocorrelation relation in the principal component space.
(c) In the principal component space, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored, but the dynamic characteristic of the process is not influenced, and the process is subjected to state switching under good control.
(d) In the principal component space, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the process shows that both the steady state deviation and the dynamic abnormity occur, the process is detected to have faults and exceed the regulation capability of the control system, and the faults are processed in time.
(e) In the residual space, if both monitored quantities are within the control limits, the variation of the residual variable is within the normal range.
(f) In the residual space, if the static monitoring amount is not exceeded, the dynamic monitoring amount is exceeded, and the abnormality of the process dynamic behavior is detected, which may cause the state deviation in the residual subspace.
(g) In the residual subspace, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored in the residual space, but the dynamic characteristic of the process is not influenced, and the process is under good control.
(h) In the residual subspace, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the steady state deviation and the dynamic abnormity are detected in the residual subspace, the process has faults, the control performance is poor, and the processing is required to be carried out in time.
The results of on-line process monitoring of the coal mill by the monitoring method of the invention are shown in fig. 2 and fig. 3. Fig. 2 shows the monitoring results in the principal component subspace, and as can be seen from fig. 2(a), in the principal component subspace, the static monitoring statistics are always within the control limit, the dynamic monitoring statistics of the first 123 samples are all within the control limit, and from the 124 th sample, the dynamic monitoring amount starts to exceed the control limit and always keeps exceeding the control limit. This means that the change of the system dynamic characteristics is detected in the principal component subspace, but no state deviation is generated, which means that the process dynamic characteristics are changed but the state of the stable operation of the process is not affected, and at this time, the process may be disturbed due to external small disturbance, and the closed-loop control system adjusts. But the disturbance is in the adjusting range of the closed-loop system and does not affect the running state of the system. Also, analyzing the monitoring results of FIG. 3 in the residual subspace, it can be seen that the process is operating in a steady state, subject to small disturbances, but within the closed loop system regulation range, the system is under good control. Generally speaking, the online monitoring method based on the dynamic and static characteristic collaborative analysis provided by the invention can fully analyze the influence of the adjustment effect of the closed-loop system on the process characteristics in the process operation, thereby improving the accuracy of process monitoring in the online monitoring, which cannot be realized by the common monitoring method. The method can provide a high-precision online process monitoring result for a technical management department in an actual industrial field of a thermal power plant, provides a reliable basis for judging the process running state in real time and identifying whether a fault occurs, and finally lays a foundation for the safe and reliable running of the coal mill.
Claims (1)
1. An intelligent power plant coal-fired generator set coal mill online monitoring method based on typical correlation analysis and slow characteristic analysis is characterized by comprising the following steps:
(1) setting a coal mill production process to have J measurement variables and operation variables, wherein each sampling can obtain a vector of 1 × J, the data obtained after N times of sampling is expressed as a two-dimensional matrix X (N × J), the measurement variables are process variables which can be measured in the running process, and the measurement variables and the operation variables comprise temperature, rotating speed, pressure and valve opening;
(2) extracting the time sequence related information of the coal mill operation data based on the typical variable analysis, wherein the step is realized by the following substeps:
(2.1) extracting typical variables of the coal mill operation data, and analyzing the time sequence correlation relationship: measuring the value x at each time point t(t)(1 × J) using the past measured values x(t-1),x(t-2),…,x(t-l)And h future measurements x(t),x(t+1),…,x(t+h)Deployment of (2);
wherein the subscript p represents the past and f represents the future; x is the number ofp(t)Representing a set of past measurements, xf(t)Representing a set of future measurements;
wherein l and h are determined according to the degree of correlation with the current time t, and the specific steps are as follows:
measured value x of past time spaced 1+1 sample intervals from current time tp(t+l+1)And the measured value x at the current moment(t)The degree of correlation of (d) is expressed as:
wherein α is a threshold, 0 < α < 0.5, autocorr (X)iP) is the autocorrelation coefficient of the ith process variable plus p time lags; degree of correlation with the current time, when Al+1When it is less than α, x is considered to bep(t+l+1)And x(t)No correlation exists and does not participate in constructing the matrix xp(t)(ii) a The smallest value satisfying the above formula is selected and determined as l, and h is determined as l, thereby constructing xp(t)And xf(t);
When the time interval between a certain measured value and the current time t is larger than the determined l and h, ignoring the correlation among the variables;
(2.2) forming the vectors spread at different time instants into a past and future matrix:
Xp=[xp(t+1),xp(t+2),…,xp(t+m)](4)
Xf=[xf(t+1),xf(t+2),…,xf(t+m)](5)
where the subscript p represents past, f represents future, m-N-l-h + 1;
(2.3) solving for the typical variable z at different times t(t)And a residual variable e(t):
Wherein, ∑pfRepresenting a past matrix XpAnd future matrix XfOf (a) covariance matrix of (a), ∑ppRepresenting a past matrix XpThe covariance matrix of (3), superscript 1/2, denotes the operation of squaring the elements within the matrix, ∑ffRepresenting future matrix XfThe upper mark 1/2 represents the operation of square root on the elements in the matrix, the formula (6) represents the left side of equal sign to carry out SVD to obtain orthogonal matrixes U and V and a diagonal matrix Λ, and the coefficient on the diagonal is the main component correlation coefficient gamma1≥…≥γrThe following table r represents the number of elements on the diagonal of the diagonal matrix Λ;
wherein J represents a transformation matrix for obtaining a correlation variable, L represents a transformation matrix for obtaining a residual scalar, I represents a unit matrix, and U representskK columns comprising matrix U, the choice of k being determined according to:
wherein β is a threshold value, 0.5 ≦ β ≦ 1;
Z=JXP(10)
E=LXP(11)
wherein the matrix Z and the matrix E respectively represent a principal component space and a residual space; each column for Z and E is the typical variable Z(t)And residual variable e(t);
(3) The method comprises the following steps of cooperatively monitoring the operation of the coal mill based on slow characteristic analysis, wherein the steps are realized by the following substeps:
(3.1) SFA modeling of principal component space Z and residual space E, respectively
sc=WcZ (12)
se=WeE (13)
Wherein s iscIs a slow feature of the extracted principal component space Z, WcIs a transformation matrix of the principal component space Z; seIs the slow feature of the extracted residual space E, WeIs the transformation matrix of the residual space E;
(3.2) selecting slow characteristic s in principal component spacecThe dominant slow feature of (1);
slowly changing features can represent the general trend of process changes, while those that change more quickly are considered noise; all slow characteristics are sorted according to the change speed, and R is selectedmThe slow characteristic with slow change is used as the main slow characteristic sc,dThe remaining slow features can be considered as noise, where Rm<Rc,RcAll slow feature numbers;
dominant slow characteristic number RmFrom the point of view of reconstruction, the selection of (c) is specifically as follows:
process variable xjCan be reconstructed by the slow feature s:
wherein the content of the first and second substances,is thatThe (c) th column of (a),is to beIs replaced by 0, the number of the elements is Rm,RmThe determination method of (2) is as follows: noise reducing reconstructed process variablesDegree of slowness ofHow much important information can be retained on behalf of the reconstructed process variable;should contain as little rapidly changing noise as possible, and soΔ(xj) Representing a process variable xjDegree of slowness of, noise-reduced reconstructed process variableRatio of change xjIs slow; and Δ (x)j) And Δ(s)i) Is in a linear relationIs Δ(s) in slow feature si)>Δ(xj) Part of xjThe change is accelerated, and the part is removed, so thatCan satisfyConsidering the reconstruction effect of all variables comprehensively, the feature to be removed is as follows:
number of divided main and slow characteristics Rm=Rc-cnt (F) is the total slow feature number minus the number of elements in set F; definition of WcFront R ofmBehavior Wc,d(Rm×J);
sc,d=Wc,dZ (18)
Wherein s isc,dIs extracted slow characteristics which can represent the main trend of the change in the process in the principal component space; wc,dRepresenting a transformation matrix;
(3.3) calculating control limits for static monitoring in principal component space:
wherein the content of the first and second substances,is sc,dT of2Monitoring statistics; determination using kernel density estimationControl limit of (Ctr)c,Td;
(3.4) calculating dynamically monitored control limits in principal component space:
wherein the content of the first and second substances,is sc,dS of2Monitoring statistics; whereinIs sc,dA first order difference of; omegac,dIs thatAn empirical covariance matrix of (2); solving using kernel density estimationControl limit of (Ctr)c,Sd;
(3.5) according to the method in (3.2), selecting slow features s in residual spaceeMain slow characteristic number R ofme: according to the slow degree of the characteristics, s is extractedeMain slow feature in (1):
(3.6) calculating the control limit of static monitoring in the residual space:
wherein the content of the first and second substances,is se,dT of2Monitoring statistics; determination using kernel density estimationControl limit of Ctre,Td;
(3.7) calculating the control limit of dynamic monitoring in the residual space:
wherein the content of the first and second substances,is se,dS of2Monitoring statistics; whereinIs se,dA first order difference of; omegae,dIs thatAn empirical covariance matrix of (2); solving using kernel density estimationControl limit of (Ctr)e,sd;
(4) On-line monitoring of the coal mill: based on the steps, the variable correlation of the coal mill operation process data and the dynamic and static cooperative monitoring model are obtained, and the monitoring statistics in the principal component space are usedAnd monitoring statistics in residual spaceMonitoring the running state of the process on line; this step is realized by the following substeps:
(4.1) acquiring and expanding newly measured data in the operation process of the coal mill: during on-line monitoring, new process measurement data x are collectednew,t(J × 1) wherein the index new represents the new sample, the index t represents the current time, J is the same measured variable as in step (1), the past l are usedThe data of the moment extends it:
(4.2) solving a typical variable and a residual variable at the current moment:
znew=Jxnew(25)
enew=Lxnew(26)
(4.3) extracting respective slow feature vectors from the typical variable and the residual variable at the current moment respectively:
(4.4) calculating online static monitoring statistics in principal component space and residual space respectively:
(4.5) calculating online dynamic monitoring statistics in principal component space and residual space respectively:
(5) judging the running state of the coal mill: comparing two monitoring indexes in different subspaces with respective statistical control limits in real time:
(a) in the principal component space, if the two monitored quantities are both in the control limit range, the process is indicated to have a constant time sequence correlation relationship, and the change of the typical variable is in the normal range;
(b) in the principal component space, if the static monitoring quantity is not over-limited and the dynamic monitoring quantity is over-limited, the fact that the abnormality of the process dynamic state is monitored can cause the steady state deviation of the variable autocorrelation relation in the principal component space;
(c) in the principal component space, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored, but the dynamic characteristic of the process is not influenced, and the process is subjected to state switching under good control;
(d) in the principal component space, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the process shows that both steady state deviation and dynamic abnormity occur, the process is detected to have faults and exceed the regulation capability of a control system, and the faults are processed in time;
(e) in the residual error space, if the two monitoring quantities are both in the control limit range, the change of the residual error variable is in the normal range;
(f) in a residual error space, if the static monitoring quantity is not over-limited, the dynamic monitoring quantity is over-limited, and the abnormity of the dynamic behavior of the process is detected, which may cause the state deviation in a residual error subspace;
(g) in the residual error subspace, if the static monitoring quantity is out of limit and the dynamic monitoring quantity is not out of limit, the steady state deviation is monitored in the residual error space, but the dynamic characteristic of the process is not influenced, and the process is under good control;
(h) in the residual subspace, if both the static monitoring quantity and the dynamic monitoring quantity exceed the limits, the steady state deviation and the dynamic abnormity are detected in the residual subspace, the process has faults, the control performance is poor, and the processing is required to be carried out in time.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811056765.1A CN109238760B (en) | 2018-09-11 | 2018-09-11 | Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811056765.1A CN109238760B (en) | 2018-09-11 | 2018-09-11 | Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109238760A CN109238760A (en) | 2019-01-18 |
CN109238760B true CN109238760B (en) | 2020-06-30 |
Family
ID=65067356
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811056765.1A Active CN109238760B (en) | 2018-09-11 | 2018-09-11 | Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109238760B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110187242A (en) * | 2019-07-02 | 2019-08-30 | 上海交通大学 | A kind of power distribution network abnormality detection and localization method based on spatio-temporal correlation analysis |
CN110879580B (en) * | 2019-12-10 | 2020-12-18 | 浙江大学 | Analysis and monitoring method for large-range non-steady transient continuous process |
CN111474911B (en) * | 2020-04-28 | 2021-03-16 | 浙江浙能技术研究院有限公司 | Gaussian non-Gaussian characteristic collaborative analysis and monitoring method for non-steady operation of high-end coal-fired power generation equipment |
CN112101733B (en) * | 2020-08-18 | 2023-11-10 | 浙江大学 | Multi-indicating variable decoupling intelligent monitoring method for coal mill of high-end power generation equipment |
CN112541017A (en) * | 2020-12-02 | 2021-03-23 | 中海油信息科技有限公司 | Industrial production process state monitoring method, device, equipment and storage medium |
CN112686477B (en) * | 2021-01-28 | 2022-04-01 | 北京工业大数据创新中心有限公司 | Coal mill blockage early warning method and system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104598681A (en) * | 2015-01-14 | 2015-05-06 | 清华大学 | Method and system for monitoring process based on slow feature analysis |
CN105091944A (en) * | 2015-08-20 | 2015-11-25 | 国家电网公司 | Thermal power plant set coal-fired calorific value and coal consumption rate index online monitoring method |
EP3056088A2 (en) * | 2016-04-13 | 2016-08-17 | DSM IP Assets B.V. | Cheese ripening |
-
2018
- 2018-09-11 CN CN201811056765.1A patent/CN109238760B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104598681A (en) * | 2015-01-14 | 2015-05-06 | 清华大学 | Method and system for monitoring process based on slow feature analysis |
CN105091944A (en) * | 2015-08-20 | 2015-11-25 | 国家电网公司 | Thermal power plant set coal-fired calorific value and coal consumption rate index online monitoring method |
EP3056088A2 (en) * | 2016-04-13 | 2016-08-17 | DSM IP Assets B.V. | Cheese ripening |
Non-Patent Citations (1)
Title |
---|
机组一次调频参数指标在线估计方法;高林 等;《中国电机工程学报》;20120605;第32卷(第16期);第62-68页 * |
Also Published As
Publication number | Publication date |
---|---|
CN109238760A (en) | 2019-01-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109238760B (en) | Online monitoring method of coal mill of intelligent power plant coal-fired generator set based on typical correlation analysis and slow characteristic analysis | |
CN110262450B (en) | Fault prediction method for cooperative analysis of multiple fault characteristics of steam turbine | |
Hsu et al. | An adaptive forecast-based chart for non-Gaussian processes monitoring: with application to equipment malfunctions detection in a thermal power plant | |
CN111159844B (en) | Abnormity detection method for exhaust temperature of gas turbine of power station | |
CN111582392A (en) | Multi-working-condition health state online monitoring method for key components of wind turbine generator | |
CN107728059B (en) | Pitch system state evaluation method | |
CN109185917B (en) | Boiler combustion state online diagnosis method and system based on flame intensity signal | |
CN109538311B (en) | Real-time monitoring method for control performance of steam turbine in high-end power generation equipment | |
CN109471420B (en) | CVA-SFA-based method for monitoring control performance of air preheater of large coal-fired power generator set of intelligent power plant | |
CN109184821B (en) | Online monitoring method for closed-loop information analysis of steam turbine of large-scale generator set | |
CN109188905B (en) | A kind of on-line monitoring method of the sound feature Cooperative Analysis towards gigawatt extra-supercritical unit | |
CN109491358A (en) | A kind of control performance monitoring method towards gigawatt Boiler of Ultra-supercritical Unit multidate information | |
Wu et al. | Fault detection and assessment for solid oxide fuel cell system gas supply unit based on novel principal component analysis | |
CN107516174A (en) | The generation method of power plant's technical supervision policy library | |
CN112465412A (en) | Power plant production information graphical method based on IMS system | |
CN110222393A (en) | The fan blade icing method for monitoring abnormality divided based on fine granularity wind-powered electricity generation generating state | |
CN114326486A (en) | Process monitoring method based on probability slow feature analysis and elastic weight consolidation | |
CN109270917B (en) | Intelligent power plant steam turbine bearing-oriented closed-loop control system fault degradation state prediction method | |
CN109283912B (en) | Distributed dynamic and static cooperative comprehensive monitoring method for pulverizing system of large coal-fired generator set of intelligent power plant | |
Agrawal et al. | Application of K-NN regression for predicting coal mill related variables | |
CN111624979B (en) | Industrial closed-loop control loop multi-oscillation detection and tracing method based on slow characteristic analysis | |
Kler et al. | The optimization problems of CP operation | |
CN114139638A (en) | Fan blade icing fault diagnosis method considering multivariable correlation | |
Weiming et al. | Fault diagnosis of coal mills based on a dynamic model and deep belief network | |
Qiao et al. | Research on SCADA data preprocessing method of Wind Turbine |
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 |