CN108845546A - A kind of dynamic process monitoring method based on BP neural network autoregression model - Google Patents
A kind of dynamic process monitoring method based on BP neural network autoregression model Download PDFInfo
- Publication number
- CN108845546A CN108845546A CN201810658889.0A CN201810658889A CN108845546A CN 108845546 A CN108845546 A CN 108845546A CN 201810658889 A CN201810658889 A CN 201810658889A CN 108845546 A CN108845546 A CN 108845546A
- Authority
- CN
- China
- Prior art keywords
- neural network
- matrix
- output
- model
- lim
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
- G05B19/41885—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by modeling, simulation of the manufacturing system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Abstract
The present invention discloses a kind of dynamic process monitoring method based on BP neural network autoregression model, it is intended to establish nonlinear autoregression model using BP neural network, and implement dynamic process monitoring on this basis.The essential core of the method for the present invention first consists in the autocorrelation model that monitored target sampled data is identified using BP neural network, next is to monitor using the filtered error implementation process of BP neural network autoregression model.The main advantage of the method for the present invention, which is first consisted in, establishes nonlinear autoregression model using the nonlinear fitting ability of BP neural network, to achieve the purpose that eliminate the non-linear autocorrelation characteristic in measurand；Secondly, error, which is not only utilized, in the method for the present invention has an ability for being able to reflect non-linear autocorrelation characteristic anomalous variation situation, but also no longer there is autocorrelation and be similarly the foundation of consequent malfunction detection model and provide convenience in error information.It can be said that the method for the present invention is monitored more suitable for dynamic process.
Description
Technical field
The present invention relates to a kind of process monitoring methods of data-driven, more particularly to one kind to be based on BP neural network autoregression
The dynamic process monitoring method of model.
Background technique
Modern process flow industry process, which is usually at, to be continued to guarantee stable product quality with efficient production status
Property, the operations such as production security and operating status stability require to the reliability of process monitoring system and the phase of validity
Prestige is higher and higher.Under industrial big data trend, using the monitoring of mechanism model implementation process technological means more and more not
It is suitable for the detection requirement of modern process flow industry process.In addition, having embodied industrial management to the producing level of industrial big data
High-level degree.Therefore, the process monitoring method and technology of data-driven is under this overall background by favor.Due to advanced instrument
The development of table technology, sampling time interval shorten dramatically, and the timing autocorrelation between sampled data is the process of data-driven
The problem that monitoring method must be taken into consideration, because the anomalous variation of timing autocorrelation equally can reflect monitored process
Object has entered damage.Most classic dynamic process monitoring method is no more than the dynamic principal component based on augmented matrix point
Analyse (Dynamic Principal Component Analysis, DPCA) method, basic thought be exactly be each trained sample
Notebook data introduces delay measurements and constitutes augmented matrix, to allow augmented matrix simultaneously by sample data timing autocorrelation
Crossing dependency between variable takes into account.
In addition to this, it proposes to excavate using autoregression model (Auto-Regression Model, ARM) there are also scholar
Sequence self correlation between sampled data, the parameter of ARM can generally be estimated by partial least squares algorithm.Use ARM's
Advantage is, to the error of output not there are sequence self correlations in ARM model, and the situation of change of error can equally reflect
The anomalous variation situation of former sample data sequence autocorrelation out, is to kill two birds with one stone.From this with angle in terms of, utilize ARM real
The thought core for applying dynamic process monitoring is how to filter out the timing autocorrelation characteristic of former sampled data.
However, it is contemplated that the complication of modern industry course-scale, autocorrelation between sampled data uses linear
It is no longer suitable that ARM is described, and ought to be described using nonlinear ARM.As a kind of Nonlinear Modeling algorithm of classics,
Nerual network technique obtained it is extensive research and application, the fields such as pattern-recognition, Nonlinear Systems Identification, market analysis all
Figure can be found.Among these, error back propagation (Back Propagation, BP) learning algorithm is the most common feedforward mind
Through Learning Algorithms, corresponding neural network model is normally referred to as BP neural network model.It is noted that nerve net
Network model needs to provide input and output data in training.And the training data of process monitoring is usually to establish singly to classify
Which sampled data generally can not be simply divided into as which inputs as exporting, because of any measurand by model
Anomalous variation is all a kind of external embodiment of failure.This be also for where the process monitoring field of data-driven, rarely have using mind
Process monitoring method through network occurs.
Summary of the invention
Technical problem underlying to be solved by this invention is：How using BP neural network nonlinear autoregression mould is established
Type, and implement dynamic process monitoring on this basis.Specifically, the essential core of the method for the present invention, which first consists in, utilizes BP mind
Go out the autocorrelation model of monitored target sampled data through Network Recognition, next is to utilize BP neural network autoregression model mistake
Error implementation process monitoring after filter.
The present invention solves technical solution used by above-mentioned technical problem：One kind being based on BP neural network autoregression model
Dynamic process monitoring method, include the following steps：
(1) sample under production process normal operating condition is acquired, training dataset X ∈ R is successively formed by the sampling timen ×m, by matrix X=[x1, x1..., xn]TIn rear n-d sample data xd+1, xd+2..., xnForm the output of autoregression model
Matrix Y=[xd+1, xd+2..., xn]T, the input matrix Z building mode of autoregression model is as follows：
Wherein, n is number of training, m is monitoring object measurand number, R are set of real numbers, Rn×mIndicate n × m dimension
Real number matrix, the transposition that d is the number of delay measurement data, N=n-d, upper label T representing matrix or vector.
(2) each column vector in input matrix Z and output matrix Y is implemented to normalize respectively according to formula as follows
Processing, i.e.,：
X=(x-xmin)/(xmax-xmin) (2)
In above formula, any one column vector in x representing matrix Z or matrix Y, xmaxWith xminRespectively indicate vector x most
Big value and minimum value.
(3) one three layers of BP neural network is constructed：The number of nodes of input layer is dm, the number of nodes of output layer is m, implicit
Layer number of nodes suggestion be set as 2dm, the activation primitive of hidden layer be S type function, output layer activation primitive be linear function.
Linearly distinguish with S type activation primitive used in the method for the present invention as follows：
G (u)=u (3)
In above formula, u representative function independent variable, g (u) are linear activation primitive, f (u) is S type activation primitive.
(4) the input Z after normalization is sent to BP neural network with output Y and is trained operation, obtain BP neural network
The weight coefficient of each neuron node after optimization, specific implementation process are as follows：
1. the weight coefficient of each neuron node of random initializtion neural network.
2. calculating the output valve of hidden layer and output layer.
3. being calculated by the real output value of neural network model and the output valve of output layer in present weight coefficient item
Error size under part, and be all to meet error precision requirement according to error judgment？If satisfied, then neural network model optimization is tied
Beam；If not satisfied, then going to step 4..
4. judging whether to have reached maximum number of iterations？If so, neural network model optimization terminates；If it is not, then basis
After the weight coefficient of regulating error hidden layer and each neuron node of output layer, return step is 2..
(5) to the output error of BP neural network modelIn each column execution standardization processing, obtain mean value
The new data matrix for being 1 for 0, standard deviation
(6) Fault Model based on Principal Component Analysis Algorithm, reserving model parameter set Θ={ P, Λ, D are establishedlim,
Qlim, whereinIt is projective transformation matrix, the association side that Λ is principal component for output layer output data matrix, the P of neural network model
Poor matrix, DlimWith QlimRespectively the upper control limit of monitoring and statistics index D and Q, specific implementation process are as follows：
1. calculatingCovariance matrix
2. solving all characteristic value γ of C1≥γ2≥…≥γmCorresponding feature vector p1, p2..., pm；
3. the principal component number k that reservation is arranged is the minimum value for meeting condition as follows, and by corresponding k feature to
Amount composition loading matrix P=[p1, p2..., pk]；
4. calculating the upper control limit D of monitoring and statistics index D and Q according to formula as followslimWith Qlim：
In above formula, F (α, k, N-d-k) indicates that freedom degree is that the F of k and N-d-k is distributed under confidence alpha (generally taking 99%)
Value,Expression freedom degree is h=2a2The chi square distribution of/v is in the value under confidence alpha, weighting coefficient g=v/
(2a), a and v respectively indicate the estimation mean value and estimate variance of Q monitoring index.
Above-mentioned steps (1)~(6) are the off-line modeling stage of the method for the present invention, and step (7)~(11) as follows are this
The online dynamic process of inventive method monitors implementation process.
(7) the data sample x at last samples moment is collectedt∈Rm×1, and find out its delay measurement data xt-1, xt-2...,
xt-dTo form the input vector z=[x of autoregression modelt-1 T, xt-2 T..., xt-d T], wherein lower label t expression is currently newest adopts
The sample moment.
(8) to input z and output xtImplement and identical normalized in step (1).
(9) by the BP neural network that training obtains in vector z input progress rapid (4), to obtain neural network output layer
Output
(10) to errorImplementation and identical standardization in step (5), obtain data vector
(11) parameter set retained in invocation step (6) implements online fault detection, and specific implementation process includes：
1. calculating the specific value of monitoring and statistics index D and Q according to formula as follows：
2. according to D and the specific value of Q and corresponding upper control limit DlimWith QlimWhether decision breaks down, that is, judge
Whether condition is met：D≤DlimAnd Q≤Qlim？If so, current sample is nominal situation sampling, return step (7) continues to implement
Monitoring to next new samples data；If it is not, then present sample data come from fault condition.
It is compared with the traditional method, inventive process have the advantage that：
Firstly, the method for the present invention establishes nonlinear autoregression model using the nonlinear fitting ability of BP neural network,
To achieve the purpose that eliminate the non-linear autocorrelation characteristic in measurand；Secondly, the method for the present invention is using error as being supervised
Object is surveyed, the ability for being able to reflect non-linear autocorrelation characteristic anomalous variation situation that error has not only is utilized, but also accidentally
The foundation that autocorrelation in difference data no longer time of occurrence sequence is similarly the subsequent process monitoring model based on PCA algorithm mentions
Convenience is supplied.It can be said that the method for the present invention is more suitable for Dynamic Process Modeling and monitoring.
Detailed description of the invention
Fig. 1 is the implementation flow chart of the method for the present invention.
Fig. 2 is the specific implementation step schematic diagram of BP neural network.
Fig. 3 is that autocorrelation characteristic rejects schematic diagram in error.
Fig. 4 is the monitoring details comparison diagram of TE process materials C inlet temperature failure.
Specific embodiment
The method of the present invention is described in detail with specific case study on implementation with reference to the accompanying drawing.
As shown in Figure 1, the present invention discloses a kind of dynamic process monitoring method based on BP neural network autoregression model.Under
Face illustrates the specific implementation process of the method for the present invention in conjunction with the example of a specific industrial process.
Before introducing specific implementation case, the basic principle of lower BP neural network need to be briefly introduced.If BP neural network
Input pattern be α=[α1, α2..., αdm], the output of hidden layer be β=[β1, β2..., β2dm], the output of output layer be γ
=[γ1, γ2..., γm], the actual output valve of neural network is y=[y1, y2..., ym].It, can according to BP neural network principle
Obtaining hidden layer output β is：
In above formula, βjIndicate the output, w of j-th of neuron of hidden layer0j=θ, x0=-1, in the present embodiment, θ=0.
Similar, the output γ of k-th of neuron of output layerkCalculation be：
So, the error under present weight coefficient condition is：
The thinking of BP neural network adjustment weight coefficient is that the most fast method of the reduction of the error according to shown in formula (7) changes
Generation adjustment.Therefore, a step-length η can be set, adjusts η unit along negative gradient direction every time, i.e., the adjustment amount of each weight is：
BP neural network adjustment sequence be：
Firstly, weight of the adjustment hidden layer to output layer, if vkFor the input of k-th of neuron of output layer, then have：
Then, the weight iteration adjustment formula of hidden layer to output layer is：
wjk=wjk-η(yk-γk)g′(vk)βj (11)
It is similar, can reasoning obtain the weighed value adjusting iterative formula from input layer to hidden layer and be：
In above formula, ukFor the input of k-th of neuron of hidden layer.
Be iterated adjustment according to above-mentioned several formula, error can be gradually reduced, until reach realize setting precision or
The specific implementation step of maximum the number of iterations, BP neural network is as shown in Figure 2.
Application comes from the experiment of U.S.'s Tennessee-Yi Siman (TE) chemical process, and prototype is that Yi Siman chemical industry is raw
Produce an actual process process in workshop.Currently, complexity of the TE process because of its process, has been used as a standard test platform quilt
It is widely used in fault detection research.Entire TE process includes that 22 measurands, 12 performance variables and 19 composition measurements become
Amount.The TE process object can be with a variety of different fault types of analog simulation, such as the variation of material inlet temperature jump, cooling water event
Barrier variation etc..In order to be monitored to the process, 33 process variables as shown in Table 1 are chosen.Due to sampling interval duration
Shorter, inevitably there is sequence self correlation in TE process sampling data.Moreover, because the complex characteristics of TE process, sampling
Nonlinear characteristic between data is more apparent, ought to implement Nonlinear Modeling.Next combine the TE process to of the invention specific real
Step is applied to be explained in detail.
Table 1：TE process monitoring variable.
Firstly, establishing dynamic process monitoring model using 960 sampled datas under TE process nominal situation, including following
Step：
Step (1)：The sample under production process normal operating condition is acquired, successively forms training dataset by the sampling time
X∈R960×33, by matrix X=[x1, x1..., xn]TIn rear n-d=958 sample data x3, x4..., x960Composition returns mould
Type output matrix Y=[x3, x4..., x960], the input matrix Z of regression model is as follows：
Step (2)：Normalized is implemented respectively to each column vector in input matrix Z and output matrix Y.
Step (3)：The BP neural network of one three layers of building.
Step (4)：Input Z after normalization is sent to BP neural network with output Y and is trained operation, obtains BP nerve
The weight coefficient of each neuron node after the network optimization.
Step (5)：To the output error of BP neural network modelIn each column execution standardization processing, obtain
The new data matrix that mean value is 0, standard deviation is 1
Step (6)：Fault Model of the foundation based on Principal Component Analysis Algorithm, reserving model parameter set Θ=P, Λ,
Dlim, Qlim}。
It is special by error matrix no longer to include non-linear autocorrelation characteristic in validation errorIn preceding 18 column vector pair
The auto-correlation schematic diagram answered is shown in Fig. 3.It can be found that the autocorrelation characteristic of former training data has been rejected, accidentally from Fig. 3
Autocorrelation is not present in difference.
Secondly, the test data set that acquisition TE process materials C inlet temperature breaks down under change condition, implements online mistake
Journey monitoring.It is worth noting that 160 sample data acquisitions are from nominal situation before the test data set, fault condition is from 161
It is introduced from moment.
Step (7)：Collect the data sample x at last samples momentt∈R33×1, and find out its delay measurement data xt-1,
xt-2To form the input vector z of autoregression model.
Step (8)：To input z and output xtImplement and identical normalized in step (1).
Step (9)：By the BP neural network that training obtains in vector z input progress rapid (4), so that it is defeated to obtain neural network
The output of layer out
Step (10)：To errorImplementation and identical standardization in step (5), obtain data vector
Step (11)：The parameter set retained in invocation step (6) implements online fault detection.
Finally, the process monitoring details that the method for the present invention changes material C inlet temperature are shown in Fig. 4.From Fig. 4
It can be found that the method for the present invention is after failure generation, it can continual triggering fault warning.
Above-mentioned case study on implementation is only used to illustrate specific implementation of the invention, rather than limits the invention.?
In the protection scope of spirit and claims of the present invention, to any modification that the present invention makes, protection of the invention is both fallen within
Range.
Claims (3)
1. a kind of dynamic process monitoring method based on BP neural network autoregression model, which is characterized in that include the following steps：
The implementation process in off-line modeling stage is as follows：
Step (1)：The sample under production process normal operating condition is acquired, training dataset X ∈ R is successively formed by the sampling timen ×m, by matrix X=[x1, x1..., xn]TIn rear n-d sample data xd+1, xd+2..., xnForm the output of autoregression model
Matrix Y=[xd+1, xd+2..., xn]T, the input matrix Z building mode of autoregression model is as follows：
Wherein, n is number of training, m is monitoring object measurand number, R are set of real numbers, Rn×mIndicate the real number of n × m dimension
Matrix, the transposition that d is the number of delay measurement data, N=n-d, upper label T representing matrix or vector；
Step (2)：Normalizing is implemented respectively to each column vector in input matrix Z and output matrix Y according to formula as follows
Change processing
X=(x-xmin)/(xmax-xmin) (2)
In above formula, any one column vector in x representing matrix Z or matrix Y, xmaxWith xminRespectively indicate the maximum value of vector x
With minimum value；
Step (3)：The BP neural network of one three layers of building：The number of nodes of input layer is dm, the number of nodes of output layer is m, hidden
Number of nodes containing layer is set as 2dm, the activation primitive of hidden layer be S type function, output layer activation primitive be linear function, point
It is not as follows：
G (u)=u (3)
In above formula, u representative function independent variable, g (u) are linear activation primitive, f (u) is S type activation primitive；
Step (4)：Input Z after normalization is sent to BP neural network with output Y and is trained operation, obtains BP neural network
The weight coefficient of each neuron node after optimization；
Step (5)：To the output error of BP neural network modelIn each column execution standardization processing, obtain mean value
The new data matrix for being 1 for 0, standard deviationWhereinFor the output layer output data matrix of neural network model
Step (6)：Establish the Fault Model based on Principal Component Analysis Algorithm, reserving model parameter set Θ={ P, Λ, Dlim,
Qlim, wherein P is projective transformation matrix, Λ is the covariance matrix of principal component, DlimWith QlimRespectively monitoring and statistics index D with
The upper control limit of Q；
The implementing procedure of online process monitoring is as follows：
Step (7)：Collect the data sample x at last samples momentt∈Rm×1, and by the measurement data x at d moment of the frontt-1,
xt-2..., xt-dForm the input vector z=[x of autoregression modelt-1 T, xt-2 T..., xt-d T], wherein lower label t is indicated currently most
New sampling instant；
Step (8)：To input z and output xtImplement and identical normalized in step (1)；
Step (9)：By the BP neural network that training obtains in vector z input progress rapid (4), to obtain neural network output layer
Output
Step (10)：To errorImplementation and identical standardization in step (5), obtain data vector
Step (11)：The parameter set retained in invocation step (6) implements online fault detection, and specific implementation process includes：
1. calculating the specific value of monitoring and statistics index D and Q according to formula as follows：
2. according to D and the specific value of Q and corresponding upper control limit DlimWith QlimWhether decision breaks down, that is, judge whether
Meet condition：D≤DlimAnd Q≤Qlim？If so, current sample is nominal situation sampling, return step (7) continues to implement under
The monitoring of one new samples data；If it is not, then present sample data come from fault condition.
2. a kind of dynamic process monitoring method based on BP neural network autoregression model according to claim 1, special
Sign is that the specific implementation process of training BP neural network is as follows in the step (4)：
1. the weight coefficient of each neuron node of random initializtion neural network；
2. calculating the output valve of hidden layer and output layer；
3. being calculated under present weight coefficient condition by the real output value of neural network model and the output valve of output layer
Error size, and be all to meet error precision requirement according to error judgment？If satisfied, then neural network model optimization terminates；
If not satisfied, then going to step 4.；
4. judging whether to have reached maximum number of iterations？If so, neural network model optimization terminates；If it is not, then according to error
After the weight coefficient for adjusting hidden layer and each neuron node of output layer, return step is 2..
3. a kind of dynamic process monitoring method based on BP neural network autoregression model according to claim 1, special
Sign is that the specific implementation process that Fault Model is established in the step (6) is as follows：
1. calculatingCovariance matrix
2. solving all characteristic value γ of C1≥γ2≥…≥γmCorresponding feature vector p1, p2..., pm；
3. the principal component number k that reservation is arranged is the minimum value for meeting condition as follows, and by corresponding k feature vector group
At loading matrix P=[p1, p2..., pk]；
4. calculating the upper control limit D of monitoring and statistics index D and Q according to formula as followslimWith Qlim：
In above formula, F (α, k, N-d-k) indicates that freedom degree is that the F of k and N-d-k is distributed in taking under confidence alpha (generally taking 99%)
Value,Expression freedom degree is h=2a2The chi square distribution of/v under confidence alpha value, weighting coefficient g=v/ (2a), a with
V respectively indicates the estimation mean value and estimate variance of Q monitoring index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810658889.0A CN108845546B (en) | 2018-06-11 | 2018-06-11 | Dynamic process monitoring method based on BP neural network autoregressive model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810658889.0A CN108845546B (en) | 2018-06-11 | 2018-06-11 | Dynamic process monitoring method based on BP neural network autoregressive model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108845546A true CN108845546A (en) | 2018-11-20 |
| CN108845546B CN108845546B (en) | 2020-10-27 |
Family
ID=64202055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810658889.0A Active CN108845546B (en) | 2018-06-11 | 2018-06-11 | Dynamic process monitoring method based on BP neural network autoregressive model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108845546B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109669414A (en) * | 2018-12-13 | 2019-04-23 | 宁波大学 | A kind of dynamic process monitoring method decomposed based on autocorrelation characteristic |
| CN111694329A (en) * | 2019-03-12 | 2020-09-22 | 宁波大学 | Dynamic process monitoring method based on distributed extreme learning machine |
| CN111912875A (en) * | 2020-06-23 | 2020-11-10 | 宁波大学 | Fractionating tower benzene content soft measurement method based on stack type Elman neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070288409A1 (en) * | 2005-05-31 | 2007-12-13 | Honeywell International, Inc. | Nonlinear neural network fault detection system and method |
| CN104714537A (en) * | 2015-01-10 | 2015-06-17 | 浙江大学 | Fault prediction method based on joint relative change analysis and autoregression model |
| CN107045283A (en) * | 2011-09-19 | 2017-08-15 | 费希尔-罗斯蒙特系统公司 | Use reasoning process modeling, prediction of quality and the fault detect of multistage data separating |
| CN107092242A (en) * | 2017-06-02 | 2017-08-25 | 宁波大学 | A kind of Industrial Process Monitoring method based on missing variable pca model |
| CN107967489A (en) * | 2017-11-29 | 2018-04-27 | 中国科学院空间应用工程与技术中心 | A kind of method for detecting abnormality and system |
-
2018
- 2018-06-11 CN CN201810658889.0A patent/CN108845546B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070288409A1 (en) * | 2005-05-31 | 2007-12-13 | Honeywell International, Inc. | Nonlinear neural network fault detection system and method |
| CN107045283A (en) * | 2011-09-19 | 2017-08-15 | 费希尔-罗斯蒙特系统公司 | Use reasoning process modeling, prediction of quality and the fault detect of multistage data separating |
| CN104714537A (en) * | 2015-01-10 | 2015-06-17 | 浙江大学 | Fault prediction method based on joint relative change analysis and autoregression model |
| CN107092242A (en) * | 2017-06-02 | 2017-08-25 | 宁波大学 | A kind of Industrial Process Monitoring method based on missing variable pca model |
| CN107967489A (en) * | 2017-11-29 | 2018-04-27 | 中国科学院空间应用工程与技术中心 | A kind of method for detecting abnormality and system |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109669414A (en) * | 2018-12-13 | 2019-04-23 | 宁波大学 | A kind of dynamic process monitoring method decomposed based on autocorrelation characteristic |
| CN111694329A (en) * | 2019-03-12 | 2020-09-22 | 宁波大学 | Dynamic process monitoring method based on distributed extreme learning machine |
| CN111694329B (en) * | 2019-03-12 | 2022-03-18 | 宁波大学 | Dynamic process monitoring method based on distributed extreme learning machine |
| CN111912875A (en) * | 2020-06-23 | 2020-11-10 | 宁波大学 | Fractionating tower benzene content soft measurement method based on stack type Elman neural network |
| CN111912875B (en) * | 2020-06-23 | 2024-02-13 | 江苏淮河化工有限公司 | Fractionation tower benzene content soft measurement method based on stacked Elman neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108845546B (en) | 2020-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abiyev | Fuzzy wavelet neural network based on fuzzy clustering and gradient techniques for time series prediction | |
| CN108897286A (en) | A kind of fault detection method based on distributing nonlinear dynamical relations model | |
| CN101334893B (en) | Fused image quality integrated evaluating method based on fuzzy neural network | |
| Chen et al. | Artificial neural networks to classify mean shifts from multivariate χ2 chart signals | |
| CN109002845A (en) | Fine granularity image classification method based on depth convolutional neural networks | |
| CN108803520A (en) | A kind of dynamic process monitoring method rejected based on the non-linear autocorrelation of variable | |
| CN108845546A (en) | A kind of dynamic process monitoring method based on BP neural network autoregression model | |
| Kultur et al. | Ensemble of neural networks with associative memory (ENNA) for estimating software development costs | |
| CN108960309A (en) | A kind of dynamic process monitoring method rejected based on RBF neural autocorrelation | |
| CN110708318A (en) | Network abnormal flow prediction method based on improved radial basis function neural network algorithm | |
| CN109298633A (en) | Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization | |
| CN111122811A (en) | Sewage treatment process fault monitoring method of OICA and RNN fusion model | |
| CN110032799A (en) | A kind of the angle similarity divided stages and monitoring method of microbiological pharmacy process | |
| Ruiz-Aguilar et al. | A novel three-step procedure to forecast the inspection volume | |
| Xu et al. | Global attention mechanism based deep learning for remaining useful life prediction of aero-engine | |
| Stepchenko | NDVI index forecasting using a layer recurrent neural network coupled with stepwise regression and the PCA | |
| Panella et al. | A study on crude oil prices modeled by neurofuzzy networks | |
| WO2014188659A1 (en) | Latent feature models estimation device, method, and program | |
| Harrington | Temperature-constrained backpropagation neural networks | |
| CN114529063A (en) | Financial field data prediction method, device and medium based on machine learning | |
| CN111695581B (en) | Nonlinear dynamic process monitoring method based on autoregressive extreme learning machine | |
| Mateus et al. | Comparing LSTM and GRU Models to Predict the Condition of a Pulp Paper Press. Energies 2021, 14, 6958 | |
| Gorbachev et al. | A synergistic effect in the measurement of neuro-fuzzy system | |
| Ramos et al. | Forecasting financial time series: a comparative study | |
| Becnel et al. | A Deep Learning Approach to Sensor Fusion Inference at the Edge |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221020 Address after: 7/F, Building 8, No. 303, Yushan Road, Jingkou District, Zhenjiang City, Jiangsu Province, 212000 Patentee after: Zhenjiang Yunyou Information Technology Co.,Ltd. Address before: Room 521, Information Institute, 818 Fenghua Road, Jiangbei District, Ningbo City, Zhejiang Province Patentee before: Ningbo University |
|
| TR01 | Transfer of patent right |