CN109471363B - Industrial melt index soft measurement method based on post-effect function and rhododendron search - Google Patents

Industrial melt index soft measurement method based on post-effect function and rhododendron search Download PDF

Info

Publication number
CN109471363B
CN109471363B CN201811628020.8A CN201811628020A CN109471363B CN 109471363 B CN109471363 B CN 109471363B CN 201811628020 A CN201811628020 A CN 201811628020A CN 109471363 B CN109471363 B CN 109471363B
Authority
CN
China
Prior art keywords
post
bird
function
bird nest
melt index
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.)
Expired - Fee Related
Application number
CN201811628020.8A
Other languages
Chinese (zh)
Other versions
CN109471363A (en
Inventor
张泽银
吕以豪
刘兴高
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201811628020.8A priority Critical patent/CN109471363B/en
Publication of CN109471363A publication Critical patent/CN109471363A/en
Application granted granted Critical
Publication of CN109471363B publication Critical patent/CN109471363B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
    • 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
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

The invention discloses an industrial melt index soft measurement method based on a post-effect function and rhododendron search, which is used for online prediction of an industrial melt index in a propylene polymerization production process and comprises a data preprocessing module, a post-effect function module and a rhododendron search algorithm module. The invention optimizes the parameters in the post-efficiency function by utilizing a rhododendron search algorithm, and approaches the optimal solution of the post-efficiency function by numerical values. The method gives consideration to mechanism and data modeling, and enables prediction results to be more accurate essentially. And the rhododendron search can further accelerate the optimization process of the function network parameters, and the practicability and the generalization of the system are improved. The invention has the characteristics of high prediction precision, stable operation, good popularization and strong anti-interference capability.

Description

Industrial melt index soft measurement method based on post-effect function and rhododendron search
Technical Field
The invention relates to a soft measurement method, in particular to an industrial melt index soft measurement method based on a post-effect function and rhododendron search.
Background
The melt index is a value representing the flowability of a plastic material during processing, which is determined based on the method of DuPont, U.S. for characterizing the plastic, and is also referred to as the melt flow rate. Polypropylene is a thermoplastic resin obtained by polymerizing propylene, and is mainly classified into isotactic polypropylene, atactic polypropylene and syndiotactic polypropylene. The melt index of the polypropylene is strictly controlled to be within a corresponding allowable value range, so that the good processability and quality of a polypropylene product are guaranteed. However, the online analysis and measurement of the melt index are still difficult to achieve at present, and the lack of an online analyzer of the melt index is a major problem which limits the quality of polypropylene products. The current method of obtaining MI is limited to subjecting a polymer solution to a temperature and load to measure the weight passed through a standard die in ten minutes. And generally, the analysis is carried out once every 1-3 hours, the time lag is large, and the requirement of real-time production control is difficult to meet.
Most of the recent research on online prediction of MI has focused on artificial neural networks, which has achieved good results. However, artificial neural networks have their own drawbacks, such as overfitting, the number of nodes in the hidden layer, and poor parameter determination. Secondly, noise, manual operation errors and the like of DCS data acquired in an industrial field have certain uncertain errors, so that a forecasting model using an artificial neural network with strong certainty is not strong in popularization capability generally.
The post-effect function is a product quality calculation model based on post-effect and hysteresis of the industrial process, the influence of each physical quantity on the parameter index of the final product in the industrial production is analyzed in mechanism, the melt index of the polypropylene can be more accurately predicted, the calculation process is simplified, the reaction rate of a prediction system is accelerated, and the method has high practical value.
Disclosure of Invention
In order to overcome the defects of low reaction speed, low measurement precision, weak anti-interference capability and poor popularization performance of an existing melt index online forecasting system in the propylene polymerization production process, the invention aims to provide the industrial melt index soft measurement method based on the post-effect function and the rhododendron search, which has high calculation precision, strong anti-interference capability, high calculation speed and good popularization performance.
The technical scheme adopted by the invention for solving the technical problems is as follows: a soft measurement method for industrial melt index based on a post-effect function and rhododendron search is used for online prediction of the industrial melt index in a propylene polymerization production process and comprises a data preprocessing module, a post-effect function module and a rhododendron search algorithm module. Wherein:
a data preprocessing module: the method is used for preprocessing the model training samples input from the DCS database, reducing the numerical difference among the model training samples input from the DCS database and improving the prediction precision of the system.
The processing is accomplished using the following mathematical process:
Xp＝log(X+1) (1)
Yp＝Y (2)
wherein, XpThe set of processed training sample data is a set of historical easy-to-measure data. X ═ X1,x2...,xn]To input a set of raw sample data from the DCS database, n is the number of variables in a set of training samples. Y ispAnd Y each represents XpX, for which no industrial melt index assay value is givenChanges are made.
An after-effect function module: and carrying out calculation by grouping the input variables transmitted from the data preprocessing module into a post-effect function network.
The after-effect function f (t) is expressed as follows
Figure BDA0001928392190000021
Figure BDA0001928392190000022
Wherein, a1、a2、a3The parameters of the after-effect function represent the shape of the function, and (t) is a unit step function and t represents time. The function represents the aftereffect and hysteresis of the influence of a certain physical quantity on the final product in the industrial production process. The function being at t0A (t)0> 0) is taken to a maximum value, indicating that the influence of the physical quantity on the product is at t0The moment reaches the maximum value, 0 to t0The function value at the moment is gradually increased, t0The post-function value gradually decreases and approaches 0.
Network of post-effect functions
Figure BDA0001928392190000023
Figure BDA0001928392190000024
Figure BDA0001928392190000025
Wherein, YkIs the k output vector of the after-effect function network and corresponds to the k industrial melt index test value in the training sample. N is the number of groups of training samples, M is the number of groups of easily-measured variables of each group of training samples,
Figure BDA0001928392190000026
is shown in the training sampleA group of easily measurable variables at i delta t moment before k industrial melt index test values, delta t represents sampling time of an on-site intelligent instrument, fiAnd (3) representing an after-effect function (i is 1, 2.. multidot.m) corresponding to the ith easily-measured variable in each group of training samples.
In the post-efficiency function network, data is input in a sliding window mode, namely M groups of historical easy-to-measure data are input, and a prediction vector Y is obtained. The net output is finally as follows
Figure BDA0001928392190000027
α(M)＝[1,1,...,1]1×M(9)
Wherein the content of the first and second substances,
Figure BDA0001928392190000028
representing the network prediction value obtained by the k-th training sample, α (M) representing a full 1 vector with the length of M.
And (3) a rhododendron search algorithm module: and performing parameter optimization on the post-effect function corresponding to each input quantity in the post-effect function network by using a rhododendron search algorithm, solving an adaptive value of each solution in the search process by using a plurality of groups of historical easily-measured variables and melt index discrete assay values collected by a DCS (distributed control system) database, and determining the parameters of the post-effect function network when the global error value is minimum.
The method comprises the following implementation steps: selecting the rhododendron from w bird nests, selecting the best bird nest, and putting the egg in the bird nest; host (parasitic bird) with a certain probability paWhen the eggs of the cuckoo are found in the bird nest of the bird, the bird eggs of the cuckoo are thrown away or a new bird nest is built.
Step 1 is initialized. Setting the number w of bird nests, searching the space dimension d, and initializing the positions of the bird nests to
Figure BDA0001928392190000031
Finding out the optimal bird nest, i.e. the position with the lowest error value
Figure BDA0001928392190000032
And the error value d at that timemin。
Step 2 enters a loop. Position of last generation optimal bird nest is reserved
Figure BDA0001928392190000033
And updating the positions of other bird nests using the following formula
Figure BDA0001928392190000034
Figure BDA0001928392190000035
Figure BDA0001928392190000036
Wherein
Figure BDA0001928392190000037
Respectively representing the position of the ith bird nest in the t generation and the (t + 1) th generation, L (s, lambda) represents a Levis function, s is a step length, U, V respectively represents a step length calculation intermediate parameter, α is a step length scaling factor, the selection is selected according to the range of a research problem search domain, v represents a randomly generated direction vector, the modulus of the direction vector is 1, lambda is a Levis index, (x) is a standard Gamma function, and s is a standard Gamma function0Obey a normal distribution for the minimum step sizes U and V, i.e.
U～N(0,σ2),V～N(0,1)
Figure BDA0001928392190000038
Wherein N (0, σ)2) Representing a mean of 0 and a variance of σ2N (0,1) represents a normal distribution with a mean of 0 and a variance of 1. Obtaining a new group of bird nest positions, and calculating the error of the group of bird nests and the position of the previous generation bird nest
Figure BDA0001928392190000039
Comparing, and reserving w bird nests with smaller errorsPosition to obtain a group of optimal bird nest positions
Figure BDA00019283921900000310
Step 3 uses random number r ∈ [0,1 ] obeying uniform distribution]As a possibility for the bird nest owner to find the bird egg of the rhododendron and paComparison of value less than paThe position of the bird nest is randomly adjusted to obtain the position of a new bird nest
Figure BDA0001928392190000041
Wherein
Figure BDA0001928392190000042
And
Figure BDA0001928392190000043
is randomly selected gtIs different from
Figure BDA0001928392190000044
The solution (x) is a unit step function. The position of the new nest is tested, and gtComparing every bird nest in the group, reserving w bird nest positions with smaller errors, and obtaining a group of new and better bird nest positions
Figure BDA0001928392190000045
Step 4 finding ptA position of the bird nest in the middle of the optimum
Figure BDA0001928392190000046
And an error value dmin. And if the stop condition is reached, the loop is exited, otherwise, the step 2 is returned.
The bird nest position x in the above algorithm is a parameter in the post-effect function network.
Defining a k-th set of training data errors dk
Figure BDA0001928392190000047
Wherein the content of the first and second substances,
Figure BDA0001928392190000048
representing the kth industrial melt index assay value in the training sample. And finally, the parameters in the post-efficiency function network are the clustering centers of N optimal parameters obtained by optimizing N groups of training samples.
The invention has the following beneficial effects: the method is characterized in that the melt index of important quality indexes in the propylene polymerization production process is subjected to online soft measurement, the defects of low measurement precision, low reaction speed, high model parameter setting difficulty and low popularization of the existing polypropylene melt index measuring instrument are overcome, a measuring method based on a post-effect function and rhododendron search is introduced for optimization, the post-effect function network is used for online prediction of the melt index, the rhododendron search algorithm is used for optimizing the post-effect function network parameters, the parameter set enabling the overall network error to be in a minimum value can be accurately and quickly calculated, and the online prediction precision of the network is improved. The method reduces the influence of noise and manual operation errors on the model forecasting precision, enhances the popularization performance of the model, effectively inhibits the overfitting phenomenon of the existing model, simplifies the calculation process and improves the forecasting precision of the model.
Drawings
FIG. 1 is a schematic diagram of the basic structure of an industrial melt index soft measurement method based on an after-effect function and rhododendron search.
FIG. 2 is a schematic diagram of a soft measurement model structure of an industrial melt index based on an after-effect function and rhododendron search.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings.
Referring to fig. 1, a propylene polymerization production process soft measurement method based on a post-effect function and rhododendron search comprises a propylene polymerization production process 1, an on-site intelligent instrument 2 for measuring easily-measured variables, a control station 3 for measuring operation variables, a DCS database 4 for storing data and a melt index soft measurement value display 6. The soft measuring instrument comprises an industrial melt index soft measuring algorithm based on a post-effect function and rhododendron search, the input end of an industrial melt index soft measuring model based on the post-effect function and the rhododendron search is connected with the output end of the DCS database, the output end of the soft measuring model based on the post-effect function and the rhododendron search is connected with a melt index soft measurement value display instrument, and the industrial melt index soft measuring model based on the post-effect function and the rhododendron search comprises:
referring to fig. 2, modeling variables required for the propylene polymerization production process soft measurement method based on the after-effect function and the rhododendron search are shown in table 1.
Table 1: modeling variable needed by industrial melt index soft measurement model based on post-effect function and rhododendron search
Variable sign Meaning of variables Variable sign Meaning of variables
T Temperature in the kettle f1 First propylene feed flow rate
P Pressure intensity in kettle f2 Second propylene feed flow rate
L Liquid level in the kettle f3 Third propylene feed flow rate
Xv Volume concentration of hydrogen in the autoclave f4 Main catalyst flow rate
f5 Flow rate of cocatalyst
Referring to fig. 2, the method for soft measurement of industrial melt index based on the after-effect function and rhododendron search further includes:
the data preprocessing module 7: the method is used for preprocessing the model training samples input from the DCS database, reducing the numerical difference among the model training samples input from the DCS database and improving the prediction precision of the system.
The processing is accomplished using the following mathematical process:
Xp＝log(X+1) (1)
Yp＝Y (2)
wherein, XpThe set of processed training sample data is a set of historical easy-to-measure data. X ═ X1,x2...,xn]To input a set of raw sample data from the DCS database, n is the number of variables in a set of training samples. Y ispAnd Y each represents XpAnd the industrial melt index test value corresponding to X is not changed.
The after-effect function module 8: and carrying out calculation by grouping the input variables transmitted from the data preprocessing module into a post-effect function network.
The after-effect function f (t) is expressed as follows
Figure BDA0001928392190000051
Figure BDA0001928392190000052
Wherein, a1、a2、a3The parameters of the after-effect function represent the shape of the function, and (t) is a unit step function and t represents time. The function represents the aftereffect and hysteresis of the influence of a certain physical quantity on the final product in the industrial production process. The function being at t0A (t)0> 0) is taken to a maximum value, indicating that the influence of the physical quantity on the product is at t0The moment reaches the maximum value, 0 to t0The function value at the moment is gradually increased, t0The post-function value gradually decreases and approaches 0.
Network of post-effect functions
Figure BDA0001928392190000061
Figure BDA0001928392190000062
Figure BDA0001928392190000063
Wherein, YkIs the k output vector of the after-effect function network and corresponds to the k industrial melt index test value in the training sample. N is the number of training sample groups, M is the number of groups of easily-measured variables of each training sample group, and Xi kA set of easily measurable variables representing i delta t before the k industrial melt index assay value in the training sample, delta t representing the sampling time of the on-site intelligent instrument, fiAnd (3) representing an after-effect function (i is 1, 2.. multidot.m) corresponding to the ith easily-measured variable in each group of training samples.
In the post-efficiency function network, data is input in a sliding window mode, namely M groups of historical easy-to-measure data are input, and a prediction vector Y is obtained. The net output is finally as follows
Figure BDA0001928392190000064
α(M)＝[1,1,...,1]1×M(9)
Wherein the content of the first and second substances,
Figure BDA0001928392190000065
representing the network prediction value obtained by the k-th training sample, α (M) representing a full 1 vector with the length of M.
And a rhododendron search algorithm module 9: and performing parameter optimization on the post-effect function corresponding to each input quantity in the post-effect function network by using a rhododendron search algorithm, solving an adaptive value of each solution in the search process by using a plurality of groups of historical easily-measured variables and melt index discrete assay values collected by a DCS (distributed control system) database, and determining the parameters of the post-effect function network when the global error value is minimum.
The method comprises the following implementation steps: selecting the rhododendron from w bird nests, selecting the best bird nest, and putting the egg in the bird nest; host (parasitic bird) with a certain probability paWhen the eggs of the cuckoo are found in the bird nest of the bird, the bird eggs of the cuckoo are thrown away or a new bird nest is built.
Step 1 is initialized. Setting the number w of bird nests, searching the space dimension d, and initializing the positions of the bird nests to
Figure BDA0001928392190000066
Finding out the optimal bird nest, i.e. the position with the lowest error value
Figure BDA0001928392190000067
And the error value d at that timemin。
Step 2 enters a loop. Position of last generation optimal bird nest is reserved
Figure BDA0001928392190000068
And updates other bird nests using the following equationPosition of
Figure BDA0001928392190000069
Figure BDA00019283921900000610
Figure BDA0001928392190000071
Wherein
Figure BDA0001928392190000072
Respectively representing the position of the ith bird nest in the t generation and the (t + 1) th generation, L (s, lambda) represents a Levis function, s is a step length, U, V respectively represents a step length calculation intermediate parameter, α is a step length scaling factor, the selection is selected according to the range of a research problem search domain, v represents a randomly generated direction vector, the modulus of the direction vector is 1, lambda is a Levis index, (x) is a standard Gamma function, and s is a standard Gamma function0Obey a normal distribution for the minimum step sizes U and V, i.e.
U～N(0,σ2),V～N(0,1)
Figure BDA0001928392190000073
Wherein N (0, σ)2) Representing a mean of 0 and a variance of σ2N (0,1) represents a normal distribution with a mean of 0 and a variance of 1. Obtaining a new group of bird nest positions, and calculating the error of the group of bird nests and the position of the previous generation bird nest
Figure BDA0001928392190000074
Comparing, reserving w bird nest positions with smaller errors, and obtaining a group of better bird nest positions
Figure BDA0001928392190000075
Step 3 uses random number r ∈ [0,1 ] obeying uniform distribution]As a possibility for the bird nest owner to find the bird egg of the rhododendron and paComparison of value less than paThe position of the bird nest is randomly adjusted to obtain the position of a new bird nest
Figure BDA0001928392190000076
Wherein
Figure BDA0001928392190000077
And
Figure BDA0001928392190000078
is randomly selected gtIs different from
Figure BDA0001928392190000079
The solution (x) is a unit step function. The position of the new nest is tested, and gtComparing every bird nest in the group, reserving w bird nest positions with smaller errors, and obtaining a group of new and better bird nest positions
Figure BDA00019283921900000710
Step 4
Finding ptA position of the bird nest in the middle of the optimum
Figure BDA00019283921900000711
And an error value dmin. And if the stop condition is reached, the loop is exited, otherwise, the step 2 is returned.
The bird nest position x in the above algorithm is a parameter in the post-effect function network.
Defining a k-th set of training data errors dk
Figure BDA00019283921900000712
Wherein the content of the first and second substances,
Figure BDA00019283921900000713
representing the kth industrial melt index assay value in the training sample. Parameters in the final post-efficiency function networkThe number is the clustering center of N optimal parameters obtained by optimizing N groups of training samples.
According to the analysis of reaction mechanism and process, in consideration of various factors influencing the melt index in the production process of polypropylene, nine operational variables and easily-measured variables commonly used in the actual production process are taken as modeling variables, including three propylene feeding flow rates, main catalyst flow rate, auxiliary catalyst flow rate, temperature, pressure and liquid level in the kettle, and hydrogen volume concentration in the kettle, table 1 lists 9 modeling variables input as a soft measurement model 5, namely temperature in the kettle (T), pressure in the kettle (P), liquid level in the kettle (L) and hydrogen volume concentration in the kettle (X)v) 3 propylene feed flow rates (first propylene feed flow rate f1, second propylene feed flow rate f2, third propylene feed flow rate f3), 2 catalyst feed flow rates (main catalyst flow rate f4, cocatalyst flow rate f 5). The polymerization reaction in the reaction kettle is carried out after reaction materials are repeatedly mixed, so that the process variable of the model input variable related to the materials adopts the average value of a plurality of previous moments. The data in this example were averaged over the previous hour. The melt index off-line assay value is used as an output variable of the soft measurement model 5. The test sample is obtained by manual sampling and offline assay analysis, and is analyzed and collected every 4 hours.
The on-site intelligent instrument 2 and the control station 3 are connected with the propylene polymerization production process 1 and the DCS database 4; the soft measurement model 5 is connected with the DCS database and the soft measurement value display instrument 6. The on-site intelligent instrument 2 measures the easily-measured variable of the propylene polymerization production object and transmits the easily-measured variable to the DCS database 4; the control station 3 controls manipulated variables of the propylene polymerization production target, and transmits the manipulated variables to the DCS database 4. The variable data recorded in the DCS database 4 is used as the input of an industrial melt index soft measurement model 5 of the support vector machine, and the soft measurement value display instrument 6 is used for displaying the output, namely the soft measurement value, of the industrial melt index soft measurement model 5 based on the after-effect function and the rhododendron search.
The examples are intended to illustrate the invention, but not to limit the invention, and any modifications and variations of the invention within the spirit and scope of the claims are intended to fall within the scope of the invention.

Claims (1)

1. A soft measurement method of an industrial melt index based on a post-effect function and rhododendron search is used for carrying out online prediction on the industrial melt index in a propylene polymerization production process and comprises a data preprocessing module, a post-effect function module and a rhododendron search algorithm module;
the data preprocessing module is used for preprocessing the model training samples input from the DCS database, reducing the numerical difference among the model training samples input from the DCS database and improving the prediction precision of the system;
the processing is accomplished using the following mathematical process:
Xp＝log(X+1) (1)
Yp＝Y (2)
wherein, XpA set of processed training sample data, namely a set of historical easy-to-measure data; x ═ X1,x2...,xn]Inputting a group of original sample data from a DCS database, wherein n is the number of variables in a group of training samples; y ispAnd Y each represents XpAnd the industrial melt index test value corresponding to X is not changed;
the post-effect function module carries out calculation on input variables transmitted from the data preprocessing module by groups and brings the input variables into a post-effect function network;
the after-effect function f (t) is expressed as follows
Figure FDA0002410000890000011
Figure FDA0002410000890000012
Wherein, a1、a2、a3The parameters of the after-effect function represent the shape of the function, and (t) is a unit step function and t represents time; the function embodies the maximum physical quantity in the industrial production processAftereffect and hysteresis of the end product impact; the function being at t0A (t)0> 0) is taken to a maximum value, indicating that the influence of the physical quantity on the product is at t0The moment reaches the maximum value, 0 to t0The function value at the moment is gradually increased, t0The post function value is gradually reduced and approaches to 0;
network of post-effect functions
Figure FDA0002410000890000013
Figure FDA0002410000890000014
Figure FDA0002410000890000015
Wherein, YkThe k output vector of the after-effect function network corresponds to the k industrial melt index test value in the training sample; n is the number of groups of training samples, M is the number of groups of easily-measured variables of each group of training samples,
Figure FDA0002410000890000016
a set of easily measurable variables representing i delta t before the k industrial melt index assay value in the training sample, delta t representing the sampling time of the on-site intelligent instrument, fiRepresenting an after-effect function corresponding to the ith easily-measured variable in each group of training samples (i ═ 1, 2.., M);
in the post-efficiency function network, data is input in a sliding window mode, namely M groups of historical easily-measured data are input to obtain a prediction vector Y; the net output is finally as follows
Figure FDA0002410000890000021
α(M)＝[1,1,...,1]1×M(9)
Wherein the content of the first and second substances,
Figure FDA0002410000890000022
representing the network forecast value obtained by the kth group of training samples, α (M) representing a full 1 vector with the length of M, and completing the industrial melt index real-time forecast based on historical data;
the cuckoo search algorithm module is used for carrying out parameter optimization on a post-effect function corresponding to each input quantity in a post-effect function network by using a cuckoo search algorithm, solving an adaptive value of each solution in a search process by using a plurality of groups of historical easily-measured variables and melt index discrete assay values collected by a DCS (distributed control system) database, and determining a post-effect function network parameter when a global error value is minimum;
the method comprises the following implementation steps: selecting the rhododendron from w bird nests, selecting the best bird nest, and putting the egg in the bird nest; host (parasitic bird) with a certain probability paWhen the eggs of the cuckoo are found in the bird nest of the bird, the bird eggs of the cuckoo are thrown away or a new bird nest is built;
step 1
Initializing; setting the number w of bird nests, searching the space dimension d, and initializing the positions of the bird nests to
Figure FDA0002410000890000023
Finding out the optimal bird nest, i.e. the position with the lowest error value
Figure FDA0002410000890000024
And the error value d at that timemin；
Step 2
Entering into circulation; position of last generation optimal bird nest is reserved
Figure FDA0002410000890000025
And updating the positions of other bird nests using the following formula
Figure FDA0002410000890000026
Figure FDA0002410000890000027
Figure FDA0002410000890000028
Wherein
Figure FDA0002410000890000029
Respectively representing the positions of ith bird nest in the t generation and the t +1 th generation, L (s, lambda) represents a Levis function, s is a step length, U, V respectively represents step length calculation intermediate parameters, α is a step length scaling factor, the selection is selected according to the range of a research problem search domain, v represents a randomly generated direction vector with the modulus of 1, lambda is a Levis index, x is a standard Gamma function, and s is a standard Gamma function0Obey a normal distribution for the minimum step sizes U and V, i.e.
U～N(0,σ2),V～N(0,1)
Figure FDA00024100008900000210
Wherein N (0, σ)2) Mean 0 and variance σ2N (0,1) represents a normal distribution with a mean of 0 and a variance of 1; obtaining a new group of bird nest positions, and calculating the error of the group of bird nests and the position of the previous generation bird nest
Figure FDA0002410000890000031
Comparing, reserving w bird nest positions with smaller errors, and obtaining a group of better bird nest positions
Figure FDA0002410000890000032
Step 3
With random numbers r ∈ [0,1 ] subject to uniform distribution]As a possibility for the bird nest owner to find the bird egg of the rhododendron and paComparison of value less than paThe position of the bird nest is randomly adjusted to obtain the position of a new bird nest
Figure FDA0002410000890000033
Wherein
Figure FDA0002410000890000034
And
Figure FDA0002410000890000035
is randomly selected gtIs different from
Figure FDA0002410000890000036
(x) is a unit step function; the position of the new nest is tested, and gtComparing every bird nest in the group, reserving w bird nest positions with smaller errors, and obtaining a group of new and better bird nest positions
Figure FDA0002410000890000037
Step 4
Finding ptA position of the bird nest in the middle of the optimum
Figure FDA0002410000890000038
And an error value dmin(ii) a If the stop condition is reached, the circulation is exited, otherwise, the step 2 is returned to;
the bird nest position x in the algorithm is a parameter in the post-effect function network;
defining a k-th set of training data errors dk
Figure FDA0002410000890000039
Wherein the content of the first and second substances,
Figure FDA00024100008900000310
representing the kth industrial melt index assay value in the training sample; and finally, the parameters in the post-efficiency function network are the clustering centers of N optimal parameters obtained by optimizing N groups of training samples.
CN201811628020.8A 2018-12-28 2018-12-28 Industrial melt index soft measurement method based on post-effect function and rhododendron search Expired - Fee Related CN109471363B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811628020.8A CN109471363B (en) 2018-12-28 2018-12-28 Industrial melt index soft measurement method based on post-effect function and rhododendron search

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811628020.8A CN109471363B (en) 2018-12-28 2018-12-28 Industrial melt index soft measurement method based on post-effect function and rhododendron search

Publications (2)

Publication Number Publication Date
CN109471363A CN109471363A (en) 2019-03-15
CN109471363B true CN109471363B (en) 2020-08-04

Family

ID=65677736

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811628020.8A Expired - Fee Related CN109471363B (en) 2018-12-28 2018-12-28 Industrial melt index soft measurement method based on post-effect function and rhododendron search

Country Status (1)

Country Link
CN (1) CN109471363B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101315557A (en) * 2008-06-25 2008-12-03 浙江大学 Propylene polymerization production process optimal soft survey instrument and method based on genetic algorithm optimization BP neural network
CN103675005A (en) * 2013-09-22 2014-03-26 浙江大学 Soft industrial melt index measurement instrument and method for optimal fuzzy network
CN105279346A (en) * 2015-11-20 2016-01-27 国网能源研究院 Method for evaluating distributed photovoltaic receiving capability of power distribution network
CN107292332A (en) * 2017-05-25 2017-10-24 哈尔滨工程大学 A kind of cuckoo algorithm based on Fuzzy C mean cluster
CN107782857A (en) * 2017-10-09 2018-03-09 江苏大学 Flexible measurement method based on the edible fungus fermented process key parameter for improving CS BPNN
CN108958181A (en) * 2018-06-28 2018-12-07 浙江大学 A kind of propylene polymerization production process optimal soft survey instrument of agility

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7599750B2 (en) * 2005-12-21 2009-10-06 Pegasus Technologies, Inc. Model based sequential optimization of a single or multiple power generating units

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101315557A (en) * 2008-06-25 2008-12-03 浙江大学 Propylene polymerization production process optimal soft survey instrument and method based on genetic algorithm optimization BP neural network
CN103675005A (en) * 2013-09-22 2014-03-26 浙江大学 Soft industrial melt index measurement instrument and method for optimal fuzzy network
CN105279346A (en) * 2015-11-20 2016-01-27 国网能源研究院 Method for evaluating distributed photovoltaic receiving capability of power distribution network
CN107292332A (en) * 2017-05-25 2017-10-24 哈尔滨工程大学 A kind of cuckoo algorithm based on Fuzzy C mean cluster
CN107782857A (en) * 2017-10-09 2018-03-09 江苏大学 Flexible measurement method based on the edible fungus fermented process key parameter for improving CS BPNN
CN108958181A (en) * 2018-06-28 2018-12-07 浙江大学 A kind of propylene polymerization production process optimal soft survey instrument of agility

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于布谷鸟算法的优化问题求解;张杰;《中国优秀硕士学位论文全文数据库信息科技辑》;20151215(第12期);论文第2章第7-10页 *

Also Published As

Publication number Publication date
CN109471363A (en) 2019-03-15

Similar Documents

Publication Publication Date Title
CN111369070B (en) Multimode fusion photovoltaic power prediction method based on envelope clustering
CN110889085A (en) Intelligent wastewater monitoring method and system based on complex network multiple online regression
CN111768000A (en) Industrial process data modeling method for online adaptive fine-tuning deep learning
CN109919356B (en) BP neural network-based interval water demand prediction method
CN109273096A (en) A kind of risk management grading evaluation method based on machine learning
CN111738482B (en) Method for adjusting technological parameters in polyester fiber polymerization process
Liao et al. Short term load forecasting and early warning of charging station based on PSO-SVM
CN115049024B (en) Training method and device of wind speed prediction model, electronic equipment and storage medium
CN111693667A (en) Water quality detection system and method based on gated recursive array
CN109507888B (en) Propylene polymerization production process optimal online forecasting system based on ensemble learning
CN103674778A (en) Industrial melt index soft measuring meter and method based on RBF (radial basis function) particle swarm optimization
CN112990343A (en) Water environment quality evaluation method based on artificial intelligence algorithm
CN113858566B (en) Injection molding machine energy consumption prediction method and system based on machine learning
Shang et al. Research on intelligent pest prediction of based on improved artificial neural network
CN109471363B (en) Industrial melt index soft measurement method based on post-effect function and rhododendron search
CN109445398B (en) Propylene polymerization production process optimal online forecasting system based on weighted combination learning
CN110851784A (en) Early warning method for field operation of electric energy meter
CN111932081A (en) Method and system for evaluating running state of power information system
CN116662925A (en) Industrial process soft measurement method based on weighted sparse neural network
CN113111588B (en) NO of gas turbine X Emission concentration prediction method and device
CN114842921A (en) Municipal sewage dephosphorization agent adding method based on linear regression and decision tree model
CN113836826A (en) Key parameter determination method and device, electronic device and storage medium
CN114897204A (en) Method and device for predicting short-term wind speed of offshore wind farm
CN113988311A (en) Quality variable prediction method, quality variable prediction device, terminal and storage medium
CN114357877A (en) Fishpond water quality evaluation prediction system and method based on fuzzy evaluation and improved support vector machine

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20200804

Termination date: 20211228

CF01 Termination of patent right due to non-payment of annual fee