CN103309347B - A kind of multiple operating modes process method for supervising based on rarefaction representation - Google Patents

Abstract

This discloses a kind of multiple operating modes process method for supervising based on rarefaction representation, belongs to industrial process monitoring and diagnostic techniques field.The method does not require process data Normal Distribution, and under only supposing process operating mode, normal service data is identical with this regime history Data distribution8.First, structure dictionary is set up according to each regime history data; Then, calculate the rarefaction representation of online data on this dictionary, then according to representing whether the intensity deterministic process of coefficient exception occurs.In addition, for normal data can also identification process be current is in certain single operating mode or transient process, to ensure that product meets production requirement.The thought of rarefaction representation is used for multiple operating modes process monitoring by the present invention, and the method does not require process data Normal Distribution, and its scope of application is wider and interpretation is stronger.

Description

A kind of multiple operating modes process method for supervising based on rarefaction representation

Technical field

The invention belongs to process flow industry process monitoring and fault diagnosis field, particularly a kind of multiple operating modes process method for supervising based on rarefaction representation.

Background technology

For process monitoring and troubleshooting issue, traditional method adopts multivariatestatistical process control technology (MultivariableStatisticalProcessControl mostly, MSPC), wherein with pivot analysis (PrincipalComponentAnalysis, PCA) and offset minimum binary (PartialLeastSquares, PLS) be successfully applied in industrial process monitoring for methods such as representatives.Traditional MSPC method all supposes that process operation is under single operation operating mode, but is in fact everlasting in multiple operating mode due to reason processes such as product change, production capacity adjustment and switches frequently.

For multi-state problem, classic method or adopt single MSPC model to cover all operation operating modes, or adopt the method for multi-model to set up sub-MSPC model to operating mode respectively, or the change of the method adaptation condition utilizing model iteration to upgrade.The most hypothetical process variable of above method meets normal distribution hypothesis, and such hypothesis might not tally with the actual situation, and method applicability can be caused weak.

Summary of the invention

Object of the present invention, for the deficiencies in the prior art, provides a kind of multiple operating modes process method for supervising based on rarefaction representation.

The multiple operating modes process method for supervising based on rarefaction representation that the present invention proposes, comprises following steps:

1) data of each nominal situation of multi-sensor data collection systematic collection process are utilized to form dictionary wherein, k represents the number of process nominal situation, represent the data matrix (sub-dictionary) of corresponding process operating mode i, m is process variable number.

2) to dictionary be normalized, make in the l of each column data ₂norm is equal to 1, obtains new dictionary matrix to be

3) gatherer process on-line operation data

4) to process on-line operation data calculate its rarefaction representation on dictionary A, monitor according to expression sparse set Exponential SCI.

5) industry and mining city.For being judged to be normal service data, industry and mining city can be carried out according to it in the rarefaction representation residual error of dictionary A further and being in certain steady working condition or operating mode transition stage so that deterministic process is current.

The invention has the beneficial effects as follows: the thought of rarefaction representation is used for multiple operating modes process monitoring by the present invention, and the method does not require process data Normal Distribution, its scope of application is wider and interpretation is stronger.In addition, for normal processes data, also can operating mode residing for the current operation of identification process to guarantee that production meets the requirements.

Accompanying drawing explanation

Fig. 1 is the FB(flow block) of the inventive method.

Embodiment

A kind of multiple operating modes process method for supervising based on rarefaction representation that the present invention proposes, its FB(flow block) as shown in Figure 1, comprises following steps:

1) data of each nominal situation of multi-sensor data collection systematic collection process are utilized to form dictionary (representing database here) wherein, k represents the number of process nominal situation, represent the data matrix (sub-dictionary) of corresponding process operating mode i, m is process variable number.

2) to dictionary be normalized, make in the l of each column data ₂the norm i.e. length of this column vector length is equal to 1, obtains new normalized dictionary matrix to be

3) process on-line operation, utilizes multi-sensor data collection system to gather m process variable data equally, and the process on-line operation data at every turn collected are t represents sampling instant.Through type (1) solves and obtains $\hat{x}={[x_1,...,x_j,...,x_n]}^T$

$\hat{x}=\underset{x}{\arg \min }{\left|\right|x\left|\right|}_1=\underset{x}{\arg \min }\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\left|x_j\right|---\left(1\right)$

Constraint condition is

Ax=y _tor || Ax-y _t|| ₂≤ ε (2)

Wherein, || || ₂represent the l of vector in this symbol ₂the norm i.e. length of vector, represent the error upper limit.

4) whether deterministic process is normally run.First, according to the coefficient that step (3) obtains calculate sparse intensity (SparseConcentrationIndex, SCI)

$\mathrm{SCI}\left(\hat{x}\right)=\frac{k\·\frac{\underset{i,j}{\max }\left[{\left|\right|{\mathrm{\δ}}_j\left(\hat{x}\right)\left|\right|}_1{\left|\right|{\mathrm{\δ}}_i\left(\hat{x}\right)\left|\right|}_1\right]}{{\left|\right|\hat{x}\left|\right|}_1}-1}{k-1}\∈\left[\mathrm{0,1}\right]---\left(3\right)$

Wherein, it is fundamental function. effect be by in at indexed set element on correspondence position is constant is set to 0 by the element on other position, indexed set I simultaneously _irepresent the column index of i-th floor data in dictionary matrix A.

If then decision process occurs abnormal.Otherwise, need further deterministic process to be in the transient process between certain steady working condition or operating mode.

5) operating mode residing for deterministic process.If according to step (4) decision process run no exceptions, so need further deterministic process current be in certain stablize single operating mode under or the transient process between two operating modes.First, online data y is calculated _tresidual error under dictionary A rarefaction representation

${\mathrm{\γ}}_i\left(y_t\right)=\left|\right|{y_t-A\·{\mathrm{\δ}}_i\left(\hat{x}\right)\left|\right|}_2,i=1,...,k---\left(4\right)$

${\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right)=\left|\right|{y_t-A\·{\mathrm{\σ}}_{p,q}\left(\hat{x}\right)\left|\right|}_2,p=1,...k,q=1,...,k,q\≠p---\left(5\right)$

Wherein, it is fundamental function. effect be by in at indexed set element on correspondence position is constant is set to 0 by the element on other position, indexed set I simultaneously _p,qrepresent the column index of p, q two floor datas in dictionary matrix A.

Then, ask for least residual and residing for this value identification online monitoring data operating mode

$a=\min [\underset{i}{\min }{\mathrm{\γ}}_i\left(y_t\right),\underset{p,q}{\min }{\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right)]---\left(6\right)$

When time, decision process is run and is in the operating mode; When $a=\underset{p,q}{\min }{\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right)$ Time, decision process is run and is in operating mode $\stackrel{\‾}{p},\stackrel{\‾}{q}(\left[\begin{array}{c}\stackrel{\‾}{p}\\ \stackrel{\‾}{q}\end{array}\right]=\underset{p,q}{\arg \min }{\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right))$ Transition period.

Claims (1)

1., based on a multiple operating modes process method for supervising for rarefaction representation, it is characterized in that the method comprises following steps:

1) data of each nominal situation of multi-sensor data collection systematic collection process are utilized to form dictionary wherein, k represents the number of process nominal situation, represent the sub-dictionary of corresponding process operating mode i, m is process variable number, n _ifor each floor data number, n is the total numbers of data;

2) to dictionary be normalized, make in the l of each column data ₂norm is equal to 1, obtains the dictionary matrix after normalization to be

3) gatherer process on-line operation data t represents sampling instant; Through type (1) solves and obtains $\hat{x}={[x_1,\·\·\·,x_j,\·\·\·,x_n]}^T$

$\hat{x}=\underset{x}{\arg \min }{\left|\right|x\left|\right|}_1=\underset{x}{\arg \min }\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\left|x_j\right|---\left(1\right)$

Constraint condition is

Ax=y _tor || Ax-y _t|| ₂≤ ε (2)

Wherein, || || ₂represent the l of vector in this symbol ₂norm, represent the error upper limit;

4) whether deterministic process is normally run; First, design factor sparse set Exponential SCI

$\mathrm{SCI}\left(\hat{x}\right)=\frac{k\·\frac{\underset{i,j}{\max }[{\left|\right|{\mathrm{\δ}}_j\left(\hat{x}\right)\left|\right|}_1+{\left|\right|{\mathrm{\δ}}_i\left(\hat{x}\right)\left|\right|}_1]}{{\left|\right|\hat{x}\left|\right|}_1}-1}{k-1}\∈\left[\mathrm{0,1}\right]---\left(3\right)$

Wherein, it is fundamental function; effect be by in at indexed set element on correspondence position is constant is set to 0 by the element on other position, indexed set I simultaneously _irepresent the column index in i-th floor data dictionary matrix A after normalization;

If then decision process occurs abnormal; Otherwise, need further deterministic process to be in the transient process between certain steady working condition or operating mode;

5) operating mode residing for deterministic process; First, online data y is calculated _tdictionary matrix A is after normalization rare

Dredge the residual error under representing

${\mathrm{\γ}}_i\left(y_t\right)={\left|\right|y_t-A\·{\mathrm{\δ}}_i\left(\hat{x}\right)\left|\right|}_2,i=1,\·\·\·,k---\left(4\right)$

${\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right)={\left|\right|y_t-A\·{\mathrm{\σ}}_{p,q}\left(\hat{x}\right)\left|\right|}_2,p=1,\·\·\·,k;q=1,\·\·\·,k;q\≠p---\left(5\right)$

Wherein, it is fundamental function; effect be by in at indexed set element on correspondence position is constant is set to 0 by the element on other position, indexed set I simultaneously _p,qrepresent the column index in p, q two floor datas dictionary matrix A after normalization;

Then, ask for least residual and residing for this least residual identification online monitoring data operating mode

$a=\min [\underset{i}{\min }{\mathrm{\γ}}_i\left(y_t\right),\underset{p,q}{\min }{\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right)]---\left(6\right)$

So, when time, decision process is run and is in the operating mode; When time, decision process is run and is in operating mode transition period, wherein $\left[\begin{array}{c}\stackrel{\‾}{p}\\ \stackrel{\‾}{q}\end{array}\right]=\underset{p,q}{\arg \min }{\stackrel{\‾}{\mathrm{\γ}}}_{p,q}\left(y_t\right).$