CN100461044C - Melt index detection fault diagnozing system and method in propylene polymerization production - Google Patents

Abstract

A fault diagnosis system of melt index detection in propylene polymerization producing process includes the scene Intelligent Instrumentation, DCS systems, and PC connected with the propylene polymerization production, and the DCS system consists of a data interface, a control station and a database. The Intelligent Instrumentation, DCS system and PC are connected with each other in order, and the PC includes a standardized processing module, independent component analysis module, support vector machine classifier function module, signal acquisition module, diagnostic data to be identified module and fault diagnosis module. It also proposes a method of fault diagnosis.

Description

Melting index detected during a kind of propylene polymerization was produced fault diagnosis system and method

(1) technical field

The present invention relates to the industrial process fault diagnosis field, especially, relate to melting index detects in a kind of propylene polymerization production fault diagnosis system and method.

(2) background technology

Polypropylene is to be the main a kind of synthetic resin that is polymerized with the propylene monomer, is the staple product in the plastics industry.In the polyolefin resin of present China, become the third-largest plastics that are only second to tygon and Polyvinylchloride.In polypropylene production process, melting index (MI) is an important indicator of reflection product quality, is the important evidence that the quality of production control and the trade mark switch.But MI can only offline inspection, and general off-line analysis needs nearly 2 hours at least, and is expensive and consuming time, particularly can't in time understand the state of polypropylene production process during 2 of off-line analysis hours.Therefore, choose with the closely-related easy survey variable of melting index as secondary variable, therefrom analyze melting index, whether normal, most important for propylene polymerization production process if detecting production run.

(3) summary of the invention

In order to overcome finding the solution trouble, being difficult to obtain diagnosis effect, the higher deficiency of rate of false alarm preferably of fault diagnosis system that melting index in the existing propylene polymerization production detects, the invention provides a kind of find the solution convenient relatively, can access good diagnosis effect, the propylene polymerization that effectively reduces rate of false alarm produce in the melting index fault diagnosis system and the method that detect.

The technical solution adopted for the present invention to solve the technical problems is:

The fault diagnosis system that melting index detected during a kind of propylene polymerization was produced comprises the field intelligent instrument, DCS system and the host computer that are connected with propylene polymerization production process, and described DCS system is made of data-interface, control station, database; Intelligence instrument, DCS system, host computer link to each other successively, and described host computer comprises:

The standardization module, be used for to the database acquisition system just often the data of key variables carry out standardization, the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i,---\left(1\right)$

2) calculate variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}}),---\left(2\right)$

3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x},---\left(3\right)$

Wherein, TX is a training sample, and N is a number of training, and TX is the average of training sample;

The independent component analysis module is used for according to the independent component number, adopts and separates mixed matrix W based on quick independent component analysis (ICA) algorithm computation of fixed point iteration, and concrete step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂Lw _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression

Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

The support vector machine classifier functional module is used for according to support vector machine nuclear parameter and fiducial probability, calculates kernel function, adopts radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i(i=1 ..., N) be the Lagrange multiplier, x _i(i=1 ..., N) be input vector, y is an output variable, ω is the normal vector of support vector machine lineoid, the decision lineoid direction, b is the parameter of decision lineoid position, δ is a nuclear parameter;

Signal acquisition module is used to set the time slot of each sampling, the signal of collection site intelligence instrument;

The diagnostic data determination module, the data that are used for gathering are sent to the DCS real-time data base, from the real-time data base of DCS database, obtain up-to-date variable data as diagnostic data VX at each timing cycle;

Fault diagnosis module, be used for TX that data to be tested VX the time is obtained with training and

Carry out standardization, and with the input of the data after the standardization as the independent component analysis module, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to the support vector machine classifier functional module after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

As preferred a kind of scheme: described host computer also comprises: the discrimination model update module, be used for regularly adding the normal point of process status to training set VX, output to standardization module, independent component analysis module, support vector machine classifier functional module, and upgrade the disaggregated model in the support vector machine classifier functional module.

As preferred another kind of scheme: described host computer also comprises: display module as a result, be used for fault diagnosis result is passed to the DCS system, and, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows at the control station procedure for displaying state of DCS.

As preferred another scheme: described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate (f ₁, f ₂, f ₃), hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

The method for diagnosing faults that the fault diagnosis system that melting index detected during the described propylene polymerization of a kind of usefulness was produced realizes, described method for diagnosing faults may further comprise the steps:

(1), from the historical data base of DCS database acquisition system just often the data of key variables as training sample TX;

(2), in the independent component analysis module of host computer, support vector machine classifier functional module, independent component number, support vector machine nuclear parameter and fiducial probability parameter are set respectively, set the sampling period among the DCS;

(3), training sample TX in host computer, data are carried out standardization, make that the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

3.1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i---\left(1\right),$

3.2) the calculating variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}})---\left(2\right),$

3.3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x}---\left(3\right),$

Wherein, N is a number of training, and N is a number of training, and TX is the average of training sample;

(4), according to the independent component number, adopt and to separate mixed matrix W based on the quick ICA algorithm computation of fixed point iteration, concrete step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂Lw _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein

Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression

Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

(5), according to support vector machine nuclear parameter and fiducial probability, calculate kernel function, adopt radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i(i=1 ..., N) be the Lagrange multiplier, x _i(i=1 ..., N) be input vector, y is an output variable, ω is the normal vector of support vector machine lineoid, the decision lineoid direction, b is the parameter of decision lineoid position, δ is a nuclear parameter;

(6), the data of gathering are sent in the DCS real-time data base, from the real-time data base of DCS database, obtain up-to-date variable data at each timing cycle as diagnostic data VX; The TX that data to be tested VX the time is obtained with training and

Carry out standardization, and with the input of the data after the standardization as the independent component analysis module, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to the support vector machine classifier functional module after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

As preferred a kind of scheme: described method for diagnosing faults also comprises: (7), regularly the normal point of process status is added among the training set VX, repeat the training process of (3)～(5), so that the disaggregated model in the support vector machine classifier functional module that upgrades in time.

As preferred another kind of scheme: in described (6), host computer is passed to the DCS system with fault diagnosis result, and, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows at the control station procedure for displaying state of DCS.

As preferred another scheme: described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate (f ₁, f ₂, f ₃), hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

Technical conceive of the present invention is: traditional multivariate statistics monitor for faults diagnostic method adopts principal component analysis (PCA) and partial least squares analysis more, these methods satisfy the independent identically distributed while at the hypothesis variable, also require the variable Normal Distribution, and what utilize only is second-order statistic information, often is difficult to obtain fault diagnosis effect preferably.

The present invention utilizes industrial measured data, adopts the method for statistics to carry out fault diagnosis, has avoided complicated Analysis on Mechanism, and it is convenient relatively to find the solution.

Blind signal analysis (independent component analysis ICA) is a kind of signal processing method based on high-order statistic, use it for the process data analyzing and processing of process industry, can more effectively utilize the probabilistic statistical characteristics of variable, can under the statistics independent meaning, decompose observational variable, obtain the activation bit source of process inherence, thereby more constitutionally is described process feature, and is more accurate, more reliable to the monitoring and the fault diagnosis of process.

Beneficial effect of the present invention mainly shows: the decorrelation sexuality of independent component analysis and the multivariable nonlinearity mapping ability and the strong generalization ability of support vector machine are combined well, brought into play advantage separately, make fault diagnosis effectively reliable more, can better instruct production, improve productivity effect.

(4) description of drawings

Fig. 1 is the hardware structure diagram of fault diagnosis system proposed by the invention;

Fig. 2 is a fault diagnosis system functional block diagram proposed by the invention;

Fig. 3 is a polypropylene production procedure sketch;

Fig. 4 is the detection design sketch of independent component analysis and support vector machine (ICA-SVM);

Fig. 5 is the theory diagram of host computer of the present invention.

(5) embodiment

Below in conjunction with accompanying drawing the present invention is further described.The embodiment of the invention is used for the present invention that explains, rather than limits the invention, and in the protection domain of spirit of the present invention and claim, any modification and change to the present invention makes all fall into protection scope of the present invention.

Embodiment 1

With reference to Fig. 1, Fig. 2, Fig. 3, Fig. 4 and Fig. 5, the fault diagnosis system that melting index detected during a kind of propylene polymerization was produced, comprise the field intelligent instrument 2, DCS system and the host computer 6 that are connected with propylene polymerization production process, described DCS system is made of data-interface 3, control station 4, database 5; Intelligence instrument 2, DCS system, host computer 6 link to each other successively by fieldbus, and described host computer 6 comprises:

Standardization module 7, be used for to the database acquisition system just often the data of key variables carry out standardization, the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i,---\left(1\right)$

2) calculate variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}}),---\left(2\right)$

3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x},---\left(3\right)$

Wherein, TX is a training sample, and N is a number of training, and TX is the average of training sample;

Independent component analysis module 8 is used for according to the independent component number, adopts and separates mixed matrix W based on quick independent component analysis (ICA) algorithm computation of fixed point iteration, and concrete step is as follows:

7. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂Lw _I-1], i=1 ..., m;

8. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein

Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

9. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression Norm;

10. if do not restrain, return 2., otherwise always iteration to i=m;

As the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

Calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

Support vector machine classifier functional module 9 is used for according to support vector machine nuclear parameter and fiducial probability, calculates kernel function, adopts radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i(i=1 ..., N) be the Lagrange multiplier, x _i(i=1 ..., N) be input vector, y is an output variable, ω is the normal vector of support vector machine lineoid, the decision lineoid direction, b is the parameter of decision lineoid position, δ is a nuclear parameter;

Signal acquisition module 10 is used to set the time slot of each sampling, the signal of collection site intelligence instrument;

Diagnostic data determination module 11, the data that are used for gathering are sent to the DCS real-time data base, from the real-time data base of DCS database, obtain up-to-date variable data as diagnostic data VX at each timing cycle;

Fault diagnosis module 12, be used for TX that data to be tested VX the time is obtained with training and

Carry out standardization, and with the input of the data after the standardization as the independent component analysis module, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to the support vector machine classifier functional module after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

Described host computer also comprises: discrimination model update module 13, be used for regularly adding the normal point of process status to training set VX, output to standardization module, independent component analysis module, support vector machine classifier functional module, and upgrade the disaggregated model in the support vector machine classifier functional module.

Described host computer also comprises: display module 14 as a result, are used for fault diagnosis result is passed to the DCS system, and at the control station procedure for displaying state of DCS, by DCS system and fieldbus process status information are delivered to operator station simultaneously and show.

Described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate (f ₁, f ₂, f ₃), hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

The hardware structure diagram of industrial process fault diagnosis system of the present invention as shown in Figure 1, described fault diagnosis system core is made of the host computer 6 that comprises three big functional modules such as standardized module 7, independent component analysis module 8, support vector machine classifier module 9 and man-machine interface, comprise in addition: field intelligent instrument 2, DCS system and fieldbus.Described DCS system is made of data-interface 3, control station 4, database 5; Propylene polymerization production process 1, intelligence instrument 2, DCS system, host computer 6 link to each other successively by fieldbus, realize uploading and assigning of information flow.Fault diagnosis system is moved on host computer 6, can carry out message exchange with first floor system easily, in time the answering system fault.

The functional block diagram of fault diagnosis system of the present invention mainly comprises three big functional modules such as standardization module 7, independent component analysis module 8, support vector machine classifier module 9 as shown in Figure 2.

Method for diagnosing faults of the present invention is implemented according to following steps:

1, from the historical data base of DCS database 5 data of just often following nine variablees of acquisition system as training sample TX: major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate (f ₁, f ₂, f ₃) hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still;

2, in the independent component analysis module 8 and support vector machine classifier module 9 of host computer 6, parameters such as independent component number, support vector machine nuclear parameter and fiducial probability are set respectively, set the sampling period among the DCS;

3, training sample TX passes through modules such as standardization 7, independent component analysis 8, support vector machine 9 successively in host computer 6, adopts following steps to finish the training of fault diagnosis system in the host computer 6;

1) in the standardization functional module 7 of host computer 6, data are carried out standardization, make that the average of each variable is 0, variance is 1, obtains input matrix X.Adopt following process to finish:

1. computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i---\left(1\right)$

2. calculate variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}})---\left(2\right)$

3. standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x}---\left(3\right)$

Wherein N is a number of training, and N is a number of training, and TX is the average of training sample;

The standardization that the standardization functional module 7 of host computer 6 is carried out can eliminate each variable because the influence that the dimension difference causes.

2) in the independent component analysis module 8 of host computer 6, carry out independent component analysis, adopt and separate mixed matrix W based on the quick ICA algorithm computation of fixed point iteration.Concrete implementation step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂Lw _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein

Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

3) disaggregated model of the support vector machine classifier functional module 9 in the training host computer 6.

The kernel function of the support vector machine classifier functional module 9 in the described host computer 6 adopts radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to following quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right)$

Wherein, α _i(i=1 ..., N) be the Lagrange multiplier, x _i(i=1 ..., N) be input vector, y is an output variable, ω is the normal vector of support vector machine lineoid, the decision lineoid direction, b is the parameter of decision lineoid position, δ is a nuclear parameter.

Definition is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

Support vector machine adopts the structural risk minimization criterion based on Statistical Learning Theory, has solved difficult problems such as small sample, local minimum point, high dimension well, is used for classification problem and can improves nicety of grading.

4, system begins to put into operation:

1) uses timer, set the time interval of each sampling;

2) field intelligent instrument 2 testing process data and being sent in the real-time data base of DCS database 5;

3) host computer 6 from the real-time data base of DCS database 5, obtains up-to-date variable data at each timing cycle, as diagnostic data VX;

4) data to be tested VX, in the standardization functional module 7 of host computer 6, the TX that the time obtains with training and

Carry out standardization, and with the input of the data after the standardization as independent component analysis module 8;

5) the independent component analysis module 8 in the host computer 6, the mixed matrix W of separating that obtains during with training is carried out conversion to input, and matrix is input to the support vector machine classifier functional module 9 in the host computer 6 after the conversion;

6) the support vector machine classifier module 9 in the host computer 6, the discriminant function that will input data substitution training obtains, the computational discrimination functional value is differentiated and the state of procedure for displaying on the man-machine interface of host computer 6;

7) host computer 6 is passed to DCS with fault diagnosis result, and at the control station 4 procedure for displaying states of DCS, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows, makes the execute-in-place worker in time to tackle.

5, sorter model upgrades

In system puts into operation process, regularly the normal point of process status is added among the training set TX, the training process of repeating step 3 is so that the disaggregated model in the support vector machine classifier 9 of the host computer 6 that upgrades in time keeps sorter model to have classifying quality preferably.

Describe a specific embodiment of the present invention below in detail.

Producing HYPOL technology actual industrial production with polypropylene is example.Figure three has provided typical Hypol continuous stirred tank (CSTR) method and has produced polyacrylic process chart, and preceding 2 stills are that CSTR reactor, back 2 stills are fluidized-bed reactor (FBR).Choose in major catalyst flow rate, cocatalyst flow rate, three bursts of propylene feed flow rates, still inner fluid temperature, still inner fluid pressure, the still in liquid level, the still nine of hydrogen volume concentration and easily survey the input quantities of performance variables as model, from the DCS system of production run, obtain the data of these nine parameters as training sample, wherein 50 normal samples are as training set, and 22 sample points are as test set data verification diagnosis effect in addition.It is 7 that ICA extracts the independent component number, and the support vector machine nuclear parameter gets 5, fiducial probability 0.98, and the sampling period is 2 hours.Fig. 4 is the detection design sketch of independent component analysis and support vector machine (ICA-SVM), the distribution of preceding two independent components that only drawn among the figure.Table 1 has been listed physical fault point and the detected trouble spot of native system in the test set, and only fail to report No. 3 trouble spots as can be seen, and rate of false alarm is 0.Obviously, native system has higher diagnostic accuracy.

The physical fault point	1，2，3，10，12，15，16
		The detection failure point	1，2，10，12，15，16

Table 1.

Embodiment 2

With reference to Fig. 1, Fig. 2, Fig. 3, Fig. 4 and Fig. 5, the method for diagnosing faults that melting index detected during a kind of propylene polymerization was produced may further comprise the steps:

(1), from the historical data base of DCS database 5 acquisition system just often the data of key variables as training sample TX;

(2), in the independent component analysis module 8 of host computer, support vector machine classifier functional module 9, independent component number, support vector machine nuclear parameter and fiducial probability parameter are set respectively, set the sampling period among the DCS;

(3), training sample TX in host computer, data are carried out standardization, make that the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

3.1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i---\left(1\right),$

3.2) the calculating variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}})---\left(2\right),$

3.3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x}---\left(3\right),$

Wherein, N is a number of training, and N is a number of training, and TX is the average of training sample;

(4), according to the independent component number, adopt and to separate mixed matrix W based on the quick ICA algorithm computation of fixed point iteration, concrete step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂Lw _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression

Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

(5), according to support vector machine nuclear parameter and fiducial probability, calculate kernel function, adopt radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i(i=1 ..., N) be the Lagrange multiplier, x _i(i=1 ..., N) be input vector, y is an output variable, ω is the normal vector of support vector machine lineoid, the decision lineoid direction, b is the parameter of decision lineoid position, δ is a nuclear parameter;

(6), the data of gathering are sent in the DCS real-time data base 5, from the real-time data base of DCS database, obtain up-to-date variable data at each timing cycle as diagnostic data VX; The TX that data to be tested VX the time is obtained with training and

Carry out standardization, and with the input of the data after the standardization as independent component analysis module 8, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to support vector machine classifier functional module 9 after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

Described method for diagnosing faults also comprises: (7), regularly the normal point of process status is added among the training set VX, repeat the training process of (3)～(5), so that the disaggregated model in the support vector machine classifier functional module that upgrades in time.

In described (6), host computer is passed to the DCS system with fault diagnosis result, and at the control station procedure for displaying state of DCS, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows.

Described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate (f ₁, f ₂, f ₃), hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

Claims (8)

1. the fault diagnosis system that melting index detected during a propylene polymerization was produced comprises the field intelligent instrument, DCS system and the host computer that are connected with propylene polymerization production process, and described DCS system is made of data-interface, control station, database; Intelligence instrument, DCS system, host computer link to each other successively, and it is characterized in that: described host computer comprises:

The standardization module, be used for to the database acquisition system just often the data of key variables carry out standardization, the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i,---\left(1\right)$

2) calculate variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}}),---\left(2\right)$

3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x},---\left(3\right)$

Wherein, TX is a training sample, and N is a number of training, and TX is the average of training sample;

The independent component analysis module is used for according to the independent component number, adopts and separates mixed matrix W based on the quick ICA algorithm computation of fixed point iteration, and concrete step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$ W wherein _I-1=[w ₁w ₂W _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein

Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression

Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix,

X is an input matrix;

The support vector machine classifier functional module is used for according to support vector machine nuclear parameter and fiducial probability, calculates kernel function, adopts radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i, i=1 ..., N is the Lagrange multiplier, x _i, i=1 ..., N is an input vector, and y is an output variable, and ω is the normal vector of support vector machine lineoid, the direction of decision lineoid, b is the parameter of decision lineoid position, σ is the sound stage width degree;

Signal acquisition module is used to set the time slot of each sampling, the signal of collection site intelligence instrument;

The diagnostic data determination module, the data that are used for gathering are sent to the DCS real-time data base, from the real-time data base of DCS database, obtain up-to-date variable data as diagnostic data VX at each timing cycle;

Fault diagnosis module, be used for TX that diagnostic data VX the time is obtained with training and

Carry out standardization, and with the input of the data after the standardization as the independent component analysis module, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to the support vector machine classifier functional module after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

2. the fault diagnosis system that melting index detected during a kind of propylene polymerization as claimed in claim 1 was produced, it is characterized in that: described host computer also comprises:

The discrimination model update module, be used for regularly adding the normal point of process status to training set, output to standardization module, independent component analysis module, support vector machine classifier functional module, and upgrade the disaggregated model in the support vector machine classifier functional module.

3. the fault diagnosis system that melting index detected during a kind of propylene polymerization as claimed in claim 1 or 2 was produced, it is characterized in that: described host computer also comprises:

Display module is used for fault diagnosis result is passed to the DCS system as a result, and at the control station procedure for displaying state of DCS, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows.

4. the fault diagnosis system that melting index detected during a kind of propylene polymerization as claimed in claim 3 was produced, it is characterized in that: described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate f ₁, f ₂, f ₃, hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

One kind produce with propylene polymerization as claimed in claim 1 in the method for diagnosing faults realized of the fault diagnosis system that detects of melting index, it is characterized in that: described method for diagnosing faults may further comprise the steps:

(1), from the historical data base of DCS database acquisition system just often the data of key variables as training sample TX;

(2), in the independent component analysis module of host computer, support vector machine classifier functional module, independent component number, support vector machine nuclear parameter and fiducial probability parameter are set respectively, set the sampling period among the DCS;

(3), training sample TX in host computer, data are carried out standardization, make that the average of each variable is 0, variance is 1, obtains input matrix X, adopts following process to finish:

3.1) computation of mean values: $\stackrel{\‾}{\mathrm{TX}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{TX}}_i---\left(1\right),$

3.2) the calculating variance: ${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{TX}}_i-\stackrel{\‾}{\mathrm{TX}})---\left(2\right),$

3.3) standardization: $X=\frac{\mathrm{TX}-\stackrel{\‾}{\mathrm{TX}}}{{\mathrm{\σ}}_x}---\left(3\right),$

Wherein, TX is a training sample, and N is a number of training, and TX is the average of training sample;

(4), according to the independent component number, adopt and to separate mixed matrix W based on the quick ICA algorithm computation of fixed point iteration, concrete step is as follows:

1. the picked at random norm is 1 initial weight vector w _i, if i 〉=2, then $w_i=w_i-W_{i-1}{W}_{i-1}^T{w}_i,$

W wherein _I-1=[w ₁w ₂W _I-1], i=1 ..., m;

2. to w _iCarrying out iteration upgrades: $w_i^{+}=E\left\{\mathrm{xg}\left(w_i^{T}x\right)\right\}-E\{g\′\left(w_i^{T}x\right)\}{w}_i,$ Wherein

Weight vectors after expression is upgraded, E is a mathematical expectation, g representative form is g (x)=xexp (x ²/ 2) function, g ' are the inverse of g;

3. standardization $w_i=w_i^{+}/\left|\right|w_i^{+}\left|\right|,$ Wherein

Expression

Norm;

4. if do not restrain, return 2., otherwise always iteration to i=m;

5. as the w that upgrades _iWith former w _iDot product is 1 o'clock, differentiates to be convergence;

6. calculate independent component: S=WX; Wherein, S is the independent component matrix, and W separates mixed matrix, and X is an input matrix;

(5), according to support vector machine nuclear parameter and fiducial probability, calculate kernel function, adopt radial basis function K (x _i, x)=exp (‖ x-x _i‖/σ ²), training process is turned to quadratic programming finds the solution problem:

$\mathrm{\ω}\left(\mathrm{\α}\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{\α}}_j{y}_i{y}_{j}K(x_i,x_j)---\left(4\right)$

Thereby obtain classification function, promptly as the sign function f (x) of minor function:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{y}_i{\mathrm{\α}}_{i}K(x_i,x)+b---\left(5\right);$

Wherein, α _i, i=1 ..., N is the Lagrange multiplier, x _i, i=1 ..., N is an input vector, and y is an output variable, and ω is the normal vector of support vector machine lineoid, the direction of decision lineoid, b is the parameter of decision lineoid position, σ is the sound stage width degree;

(6), the data of gathering are sent in the DCS real-time data base, from the real-time data base of DCS database, obtain up-to-date variable data at each timing cycle as diagnostic data VX; The TX that diagnostic data VX the time is obtained with training and Carry out standardization, and with the input of the data after the standardization as the independent component analysis module, the mixed matrix W of separating that obtains during with training is carried out conversion to input, matrix is input to the support vector machine classifier functional module after the conversion, the discriminant function f (x) that obtains is trained in the substitution of input data, the computational discrimination functional value is as f (x) 〉=0, data sample is in normal condition; When f (x)＜0, be in abnormality.

6. method for diagnosing faults as claimed in claim 5, it is characterized in that: described method for diagnosing faults also comprises: (7), regularly the normal point of process status is added in the training set, repeat the training process of (3)～(5), so that the disaggregated model in the support vector machine classifier functional module that upgrades in time.

7. as claim 5 or 6 described method for diagnosing faults, it is characterized in that: in described (6), host computer is passed to the DCS system with fault diagnosis result, and, by DCS system and fieldbus process status information is delivered to operator station simultaneously and shows at the control station procedure for displaying state of DCS.

8. method for diagnosing faults as claimed in claim 7 is characterized in that: described key variables comprise major catalyst flow rate f ₄, cocatalyst flow rate f ₅, three gangs of propylene feed flow rate f ₁, f ₂, f ₃, hydrogen volume concentration α in liquid level l and the still in the still inner fluid temperature T, still inner fluid pressure P, still.

Images