CN107169565A - Yarn quality prediction method based on fireworks algorithm improvement BP neural network - Google Patents
Yarn quality prediction method based on fireworks algorithm improvement BP neural network Download PDFInfo
- Publication number
- CN107169565A CN107169565A CN201710288559.2A CN201710288559A CN107169565A CN 107169565 A CN107169565 A CN 107169565A CN 201710288559 A CN201710288559 A CN 201710288559A CN 107169565 A CN107169565 A CN 107169565A
- Authority
- CN
- China
- Prior art keywords
- mrow
- fireworks
- msub
- network
- munderover
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of yarn quality prediction method based on fireworks algorithm improvement BP neural network, fireworks algorithm is incorporated into BP neural network, the network weight and threshold value of BP neural network model are optimized using the optimizing mechanism of fireworks algorithm, choose Input and Output Indexes, construct the predictive model for yarn quality based on FWA BP, the predictive model for yarn quality based on FWA BP set up in step 2 is learnt and trained using the data set Jing Guo standardization, the prediction to spinning quality is finally completed.The present invention is solved because influenceing yarn qualities factor numerous in spinning unit and the problem of spinning quality is difficult to accurate prediction caused by coupling each other, and the Function Mapping relation that can effectively set up between fiber index and yarn quality, the prediction to the yarn qualities in Spinning process is realized, is conducive to improving the level of spinning workshop quality management.
Description
Technical field
The invention belongs to yarn quality prediction and control technology field, it is related to a kind of based on fireworks algorithm improvement BP nerve nets
The yarn quality prediction method of network.
Background technology
Spinning unit is in the interlaced complex environment of many factors such as high temperature, high humidity and high electromagnetism, each factor
Between there is interactional coupling relation, Spinning process processing process is complicated in addition and raw material frequently undergo thing
The modifying process of Physicochemical so that the prediction of quality in process of textile production is compared with the prediction of quality of traditional purely mechanic processing
It is more challenging.Especially, fiber ATTRIBUTE INDEX increases in geometry shape, and more than 300, and spinning unit have been reached at present
Middle influence factor yarn qualities factor is numerous and there is coupled relation each other, in addition fiber attribute and yarn qualities characteristic value
Between into nonlinear correlation relation so that utilize neural network predictive model for yarn quality under Small Sample Database training
Predict the outcome, it is difficult to meet the actual requirement of spinning workshop production management.
With the raising of the Spinning process level of informatization, substantial amounts of raw material, technique, equipment are have accumulated in process of textile production
Deng yarn qualities data, this to set up under big-sample data environment the predictive model for yarn quality based on neutral net turns into can
Energy.But, under a large amount of training sample data environment, with input neuron number in neural network prediction model and training sample
Notebook data amount is increased considerably, and neural network model convergence rate is slow and is further highlighted the problem of being easily absorbed in local optimum,
Largely govern the precision of yarn quality prediction.
The content of the invention
It is an object of the invention to provide a kind of yarn quality prediction method based on fireworks algorithm improvement BP neural network, solution
The problem of precision of prediction in the training process that the existing neural network model of having determined is present is low and iterations is high.
Yarn quality prediction method of the invention based on fireworks algorithm improvement BP neural network, it is specifically real according to following steps
Apply：
Step 1, the network weight and threshold value of BP neural network model are optimized using the optimizing mechanism of fireworks algorithm,
Set up a kind of FWA-BP neural network models based on fireworks algorithm optimization；
Step 2, on the basis of the FWA-BP neural network models of step 1, Input and Output Indexes are chosen, structure is based on
FWA-BP predictive model for yarn quality；
Step 3, it is pre- to the spinning quality based on FWA-BP set up in step 2 using the data set Jing Guo standardization
Survey model to be learnt and trained, be finally completed the prediction to spinning quality.
The features of the present invention is also resided in,
The tool that the optimizing mechanism of fireworks algorithm is optimized to the network weight and threshold value of BP neural network model in step 1
Body step is：
Step 1.1, key parameter is encoded, and the coding strategy for choosing real number vector is compiled to the key parameter in model
Code, note vector X=[x1,x2,…,xD] one group of parameter to be optimized is represented, it is per one-dimensional vector by network weight and sets of threshold values
Into the dimension of fireworks population is：D=nIW(1,1)+nb(1,1)+nIW(2,1)+nb(2,1), wherein, remember nIW(1,1)For hidden layer and output layer
Between weights number, nb(1,1)For the number of hidden layer neuron threshold value, nIW(2,1)For hidden layer and the weights of output interlayer
Number, nb(2,1)The number of output layer neuron threshold value；
Step 1.2, weight system and threshold value initialization, on the basis of step 1.1, utilize fireworks individual in fireworks algorithm
xikPositional representation neutral net in neuron, by i-th of neuron in time l layers of the iterative process network of kth in neutral net
With j-th of interneuronal weight coefficientAnd threshold θiInitialization is encoded into vectorial X=[x1,x2,…,xD], and utilize
The strategy of random initializtion is initially at vector X in interval [- 1,1], then has weight coefficient wij~U [- 1,1],
Wherein, i, j refer respectively to the weight between i-th each neuron node and each neuron node of jth, l tables in network
What is shown is the network number of plies residing for this present weight, and what k was represented is current iterations；
Step 1.3, the error of fireworks individual is calculated, fitness function is introduced and calculates flat using formula (1) and formula (2)
Square error SSE, formula (1) and formula (2) are as follows：
Wherein, t is the desired output of network, and p is the number of plies of network, and s is the number of network output unit, and y is that network is defeated
Go out value, its formula specific as follows：
Wherein, xjFor the input of network, wijFor the weight of network node, θiFor the threshold value and θ of i-th of neuron in networki
=-wi(n+1)；
Step 1.4, obtained each fireworks individual x is calculated in step 1.3iOn the basis of error, f is introducedi(x) function is made
For fitness function, pass through each fireworks of vector X individual x in fitness function calculation procedure 1.2iFitness value, adapt to
Spend function such as formula (3) as follows,
Step 1.5, fireworks population optimizing, on the basis of step 1.4, for each fireworks individual xiCarry out quick-fried
Fried, displacement and mutation operation, wherein blast mutation operation and Gaussian mutation mapping ruler are formula (4)~formula (6),
H=Ai×rand(1,-1) (4)
exik=xik+h (5)
mxik=xik×e (6)
Wherein, AiFor the burst radius of i-th of fireworks, h is position offset, xikRepresent the kth of i-th of fireworks in population
Dimension, exikFor spark of i-th of fireworks after blast, mxikFor xikGaussian mutation spark after Gaussian mutation, e~N
(1,1) Gaussian Profile；
Step 1.6, fireworks population of future generation is selected, for the cigarette in step 1.5 after blast, displacement and mutation operation
Spend individual xi, each fireworks individual x is calculated using the formula in step 1.4iFitness value, and use formula (7) and formula
(8) selection strategy, selects optimal fireworks individual composition fireworks population of future generation, and specific selection strategy is：
Min (f (the x for selecting fitness value minimumi)) individual xkIt is directly fireworks population at individual, remaining N-1 cigarette
Individual is spent to take roulette mode, for candidate individual xiIts selected probability such as following formula：
Wherein, R (xi) represent fireworks individual xiWith other individuals apart from sum, formula specific as follows；
Step 1.7, judge end condition, the adaptation of fireworks individual in fireworks population is calculated according to formula (3) and formula (8)
Angle value f (xi) fireworks individual between Euclidean distance R (xi), and judge whether the maximum iteration that is reached in end condition, if
Meet the minimum fitness value min (f (x for then calculating the fireworks individual obtained in current fireworks populationi)) and fireworks population in
It is maximum apart from max (R (x between fireworks individuali)), and take current fireworks population to be optimal fireworks population Xbest, otherwise after
It is continuous to perform step 1.3；
Step 1.8, optimization network weight and threshold value, using obtaining optimal fireworks population X in step 1.7bestTo step 1.2
In vectorial X in weight in corresponding neutral net and threshold value initialized.
The specific method that the predictive model for yarn quality based on FWA-BP is built in step 2 is：
Step 2.1, the selection of Input and Output Indexes：Choose original related to yarn qualities in Spinning process process
The data such as material, technique are as input variable, and the CV values for choosing yarn are output-index, then the whole yarn qualities based on FWA-BP
The input and output of forecast model are：
Input quantity is：X1=sliver percentage of impuritys, x2=roving twist factors, x3=regains, x4=fibre diameters, x5=is fine
Dimension length, x6=diameter coefficient of dispersion, x7=fiber quality irregularities, x8=drawing of fiber multiples, x9=spun yarn wire loops number,
X10=roller rotating speeds；Output quantity is：Y=yarn CV values；
Step 2.2, the inputoutput data obtained according to step 2.1 sets up the data set of model, and uses Min-Max side
The data that method is concentrated to data are standardized；
Step 2.3, the strategy of network structure is determined, according to the input, output-index chosen in step 2.1, it is determined that input,
Output and the number of plies of hidden layer, the nodes m=10 of the input layer of FWA-BP predictive model for yarn quality, output layer nodes n
=1, the number of wherein hidden neuron is determined by following formula
Calculating obtains s=7；
Step 2.4, the selection of activation primitive, input layer uses tansig activation primitives, and output layer is activated using purelin
Function, chooses trainlm functions as the training function of network model.
Using the data set Jing Guo standardization to the spinning quality based on FWA-BP set up in step 2 in step 3
Forecast model is learnt and concretely comprising the following steps of predicting：
Step 3.1, the selection strategy of training dataset, using the data set in step 2.2 Jing Guo standardization, therefrom
The data set of selection 80% is used as test data set as training dataset, the data set of residue 20%；
Step 3.2, in fireworks algorithm key parameter setting, the size N=70 of fireworks population, fireworks burst radius regulation
Dividing value lm=0.8 in constant d=5, fireworks explosive spark number regulating constant m=40, fireworks explosive spark number, fireworks blast fire
Flower number floor value bm=0.04, Gaussian mutation spark number g=5, wherein maximum iteration T=100, variable dimension D=
85, it is the sum that neuron weight and threshold value in network model are taken on the basis of step 1.1, is specifically in step 2.3
Calculated on basis by equation below
D=m × s+s × n+s+n=10 × 7+7 × 1+7+1=85
Wherein, m, s, n are respectively of input layer, hidden layer neuron and the output layer neuron of network
Number；
Step 3.3, on the basis that fireworks algorithm parameter is set in step 3.2, the training selected in step 3.1 is used
Data set is trained to the predictive model for yarn quality based on FWA-BP, the related parameter setting in network training process
For learning rate is 0.01, and factor of momentum is 0.9, and maximum iteration is 20000, and training minimal error is 0.05；
Step 3.4, trained by step 3.1~3.3 and obtained the predictive model for yarn quality based on FWA-BP, use step
The test data set selected in rapid 3.1, test statisticses analysis and experiment simulation are carried out to the prediction effect of model.
The present invention, which is compared with the prior art, to be had the following effects that：The present invention improves the precision of yarn quality prediction, reduction
The iterations of network.Fireworks algorithm is mainly incorporated into neural network model by the present invention, using in fireworks blast process
The mechanism that multiple spot spreads simultaneously, weight and threshold value to neural network model are optimized, so as to reduce forecast model
Iterations, improve model prediction accuracy rate.
Brief description of the drawings
Fig. 1 is the flow chart of the yarn quality prediction method of the invention based on fireworks algorithm improvement BP neural network；
Fig. 2 is the spinning of embodiment in the yarn quality prediction method of the invention based on fireworks algorithm improvement BP neural network
The simulation result figure of quality predictions and actual value；
Fig. 3 be in the yarn quality prediction method based on fireworks algorithm improvement BP neural network of the invention embodiment and other
The yarn quality prediction Comparative result simulation result figure of BP neural network, GA-BP neutral nets and PSO-BP neutral nets；
Fig. 4 is to train in embodiments of the invention what is obtained to become based on the input that BP neural network is set up by identical parameters
The correlation analysis figure of mapping relations between amount and output variable；
Fig. 5 be trained by identical parameters in embodiments of the invention obtain based on the defeated of GA-BP neural networks
Enter the correlation analysis figure of mapping relations between variable and output variable；
Fig. 6 be trained by identical parameters in embodiments of the invention obtain based on the defeated of PSO-BP neural networks
Enter the correlation analysis figure of mapping relations between variable and output variable；
Fig. 7 is the input variable and output variable based on FWA-BP neural networks proposed in embodiments of the invention
Between mapping relations correlation analysis figure.
Embodiment
The present invention is described in detail with reference to the accompanying drawings and detailed description.
The flow chart of yarn quality prediction method based on fireworks algorithm improvement BP neural network in the embodiment of the present invention, such as
Shown in Fig. 1, the yarn quality prediction method of the invention based on fireworks algorithm improvement BP neural network is specifically real according to following steps
Apply：
Step 1, the network weight and threshold value of BP neural network model are optimized using the optimizing mechanism of fireworks algorithm,
A kind of FWA-BP neural network models based on fireworks algorithm optimization are set up, the optimizing mechanism of fireworks algorithm is to BP neural network mould
What the network weight and threshold value of type were optimized concretely comprises the following steps：
Step 1.1, key parameter is encoded, and the coding strategy for choosing real number vector is compiled to the key parameter in model
Code, note vector X=[x1,x2,…,xD] one group of parameter to be optimized is represented, it is per one-dimensional vector by network weight and sets of threshold values
Into the dimension of fireworks population is：D=nIW(1,1)+nb(1,1)+nIW(2,1)+nb(2,1), wherein, remember nIW(1,1)For hidden layer and output layer
Between weights number, nb(1,1)For the number of hidden layer neuron threshold value, nIW(2,1)For hidden layer and the weights of output interlayer
Number, nb(2,1)The number of output layer neuron threshold value；
Step 1.2, weight system and threshold value initialization, on the basis of step 1.1, utilize fireworks individual in fireworks algorithm
xikPositional representation neutral net in neuron, by i-th of neuron in time l layers of the iterative process network of kth in neutral net
With j-th of interneuronal weight coefficientAnd threshold θiInitialization is encoded into vectorial X=[x1,x2,…,xD], and utilize
The strategy of random initializtion is initially at vector X in interval [- 1,1], then has weight coefficient wij~U [- 1,1],
Wherein, i, j refer respectively to the weight between i-th each neuron node and each neuron node of jth, l tables in network
What is shown is the network number of plies residing for this present weight, and what k was represented is current iterations；
Step 1.3, the error of fireworks individual is calculated, fitness function is introduced and calculates flat using formula (1) and formula (2)
Square error SSE, formula (1) and formula (2) are as follows：
Wherein, t is the desired output of network, and p is the number of plies of network, and s is the number of network output unit, and y is that network is defeated
Go out value, its formula specific as follows：
Wherein, xjFor the input of network, wijFor the weight of network node, θiFor the threshold value and θ of i-th of neuron in networki
=-wi(n+1)；
Step 1.4, obtained each fireworks individual x is calculated in step 1.3iOn the basis of error, f is introducedi(x) function is made
For fitness function, pass through each fireworks of vector X individual x in fitness function calculation procedure 1.2iFitness value, adapt to
Spend function such as formula (3) as follows,
Step 1.5, fireworks population optimizing, on the basis of step 1.4, for each fireworks individual xiCarry out quick-fried
Fried, displacement and mutation operation, wherein blast mutation operation and Gaussian mutation mapping ruler are formula (4)~formula (6),
H=Ai×rand(1,-1) (4)
exik=xik+h (5)
mxik=xik×e (6)
Wherein, AiFor the burst radius of i-th of fireworks, h is position offset, xikRepresent the kth of i-th of fireworks in population
Dimension, exikFor spark of i-th of fireworks after blast, mxikFor xikGaussian mutation spark after Gaussian mutation, e~N
(1,1) Gaussian Profile；
Step 1.6, fireworks population of future generation is selected, for the cigarette in step 1.5 after blast, displacement and mutation operation
Spend individual xi, each fireworks individual x is calculated using the formula in step 1.4iFitness value, and use formula (7) and formula
(8) selection strategy, selects optimal fireworks individual composition fireworks population of future generation, and specific selection strategy is：
Min (f (the x for selecting fitness value minimumi)) individual xkIt is directly fireworks population at individual, remaining N-1 cigarette
Individual is spent to take roulette mode, for candidate individual xiIts selected probability such as following formula：
Wherein, R (xi) represent fireworks individual xiWith other individuals apart from sum, formula specific as follows；
Step 1.7, judge end condition, the adaptation of fireworks individual in fireworks population is calculated according to formula (3) and formula (8)
Angle value f (xi) fireworks individual between Euclidean distance R (xi), and judge whether the maximum iteration that is reached in end condition, if
Meet the minimum fitness value min (f (x for then calculating the fireworks individual obtained in current fireworks populationi)) and fireworks population in
It is maximum apart from max (R (x between fireworks individuali)), and take current fireworks population to be optimal fireworks population Xbest, otherwise after
It is continuous to perform step 1.3；
Step 1.8, optimization network weight and threshold value, using obtaining optimal fireworks population X in step 1.7bestTo step 1.2
In vectorial X in weight in corresponding neutral net and threshold value initialized.
Step 2, on the basis of the FWA-BP neural network models of step 1, Input and Output Indexes are chosen, structure is based on
FWA-BP predictive model for yarn quality, builds concretely comprising the following steps for the predictive model for yarn quality based on FWA-BP：
Step 2.1, the selection of Input and Output Indexes：
In view of Spinning process process be in it is multifactor intercouple under effect, the parameter such as raw material, technique, equipment is all
Yarn qualities can be produced with influence, thus choose in terms of material performance, technological parameter, device parameter as shown in the table 10
Individual index chooses yarn CV values as the output-index of predictive model for yarn quality as the input of predictive model for yarn quality.
Step 2.2, the inputoutput data obtained according to step 2.1 sets up the data set of model, and uses Min-Max side
The data that method is concentrated to data are standardized；
The standardization of data is completed by below equation：
Wherein, max (X) is the maximum that training data is concentrated, and min (X) is the maximum that training data is concentrated, by right
After data are standardized, training data is mapped to interval [0,1], overall merit contrast is convenient for.
Step 2.3, the strategy of network structure is determined, according to the input, output-index chosen in step 2.1, it is determined that input,
Output and the number of plies of hidden layer, the nodes m=10 of the input layer of FWA-BP predictive model for yarn quality, output layer nodes n
=1, the number of wherein hidden neuron is determined by following formula
Wherein, m=10 and n=1 are respectively the input layer number and output layer node number of network, and calculating obtains s
=7；
Step 2.4, the selection of activation primitive, input layer uses tansig activation primitives, and output layer is activated using purelin
Function, chooses trainlm functions as the training function of network model.
Step 3, it is pre- to the spinning quality based on FWA-BP set up in step 2 using the data set Jing Guo standardization
Survey model to be learnt and trained, be finally completed the prediction to spinning quality, concretely comprise the following steps：
Step 3.1, the selection strategy of training dataset, takes the cotton spun yarn qualitative data of certain company, matter of being spinned to FWA-BP
The validity for measuring predictive model algorithm carries out experimental verification.In the training process of algorithm model, preceding the 80% of data set is taken
Data are training dataset, for being trained to FWA-BP models, and rear 20% data for taking data set are test set, are used for
Test to model prediction performance；
Step 3.2, in fireworks algorithm key parameter setting, according to the weighted value of network to be optimizedAnd threshold value
θiSpecific optimization aim, and combine pertinent literature in experimental result, the key parameter in fireworks algorithm is set to, fireworks
The size N=70 of population, fireworks burst radius regulating constant d=5, fireworks explosive spark number regulating constant m=40, fireworks blast
Dividing value lm=0.8 in spark number, fireworks explosive spark number floor value bm=0.04, Gaussian mutation spark number g=5, it is maximum
Iterations T=100, the dimension D=85 of variable, be taken on the basis of step 1.1 in network model neuron weight and
The sum of threshold value, is specifically to be calculated on the basis of step 2.3 by equation below
D=m × s+s × n+s+n=10 × 7+7 × 1+7+1=85
Wherein, m, s, n are respectively of input layer, hidden layer neuron and the output layer neuron of network
Number；
Step 3.3, on the basis that fireworks algorithm parameter is set in step 3.2, the training selected in step 3.1 is used
Data set is trained to the predictive model for yarn quality based on FWA-BP, and related parameter setting is such as in network training process
Under, learning rate is 0.01, and factor of momentum is 0.9, and maximum iteration is 20000, and training minimal error is 0.05；
Step 3.4, trained by step 3.1~3.3 and obtained the predictive model for yarn quality based on FWA-BP, use step
The test data set selected in rapid 3.1, test statisticses analysis and experiment simulation are carried out to the prediction effect of model.
In addition, using the identical training set data in step 2 Jing Guo standardization to traditional BP neural network,
The predictive model for yarn quality such as GA-BP and PSO-BP are trained, and calculate error rate that algorithms of different predicts the outcome and repeatedly
Generation number.
In order to reduce the accidental sexual factor in experimentation, same algorithm model is trained using same data and tested
10 times, the error of 10 predictions and the average value of iterations are taken respectively as the prediction error value and convergence speed for evaluating the algorithm
Degree, the training result of the yarn qualities forecast model based on FWA-BP is as shown in table 1, as can be seen from Table 1：This specific implementation is carried
The yarn quality prediction method based on FWA-BP neutral nets gone out, relative to particle group optimizing neutral net (PSO-BP) its
The error rate of spinning quality characteristic value fluctuation forecast have dropped 49.52%, and the precision of forecast reaches 97.88%, and the algorithm
Iterations reduce 31.11%.
The training result of yarn qualities forecast model of the table 1 based on FWA-BP
The simulation result figure of the yarn quality prediction value of embodiment and actual value in the present invention, as shown in Fig. 2 can from figure
The predictive model for yarn quality based on GA-BP neutral nets proposed with finding out in the embodiment of the present invention, can preferably be realized
Prediction to spinning quality；
The spinning of embodiment and other BP neural networks, GA-BP neutral nets and PSO-BP neutral nets in the present invention
Prediction of quality Comparative result simulation result figure, as shown in figure 3, the as can be seen from the figure neural network model pair based on FWA-BP
In predicting the outcome closer to actual value for spinning quality；
In the present invention in embodiment by identical parameters train obtain based on BP neural network set up input variable and
The correlation analysis figure of mapping relations between output variable, as shown in figure 4, its coefficient R is 0.85176；
In the present invention the obtained input variable based on GA-BP neural networks is trained in embodiment by identical parameters
The correlation analysis figure of mapping relations between output variable, as shown in figure 5, its coefficient R is 0.91472；
The input variable based on PSO-BP neural networks is obtained by identical parameters training in embodiment in the present invention
The correlation analysis figure of mapping relations between output variable, as shown in fig. 6, its coefficient R is 0.92182；
Reflected between the input variable based on FWA-BP neural networks and output variable that are proposed in the present invention in embodiment
The correlation analysis figure of relation is penetrated, as shown in fig. 7, its coefficient R is 0.9479.
Fireworks algorithm is incorporated into the optimization of neural network weight and threshold value by the present invention, it is proposed that one kind is based on FWA-BP
The forecast model of network, experiment and simulation result show that method proposed by the present invention has relatively low predicated error rate and less
Iterations, the precision of prediction that can effectively solve the presence of traditional neural network forecast model is low and iterations is high asks
Topic, a kind of new method is provided fast and effeciently to solve forecasting problem under big-sample data.
Claims (4)
1. the yarn quality prediction method based on fireworks algorithm improvement BP neural network, it is characterised in that specifically according to following step
It is rapid to implement：
Step 1, the network weight and threshold value of BP neural network model are optimized using the optimizing mechanism of fireworks algorithm, set up
A kind of FWA-BP neural network models based on fireworks algorithm optimization；
Step 2, on the basis of the FWA-BP neural network models of step 1, Input and Output Indexes are chosen, builds and is based on FWA-
BP predictive model for yarn quality；
Step 3, using the data set Jing Guo standardization to the yarn quality prediction mould based on FWA-BP set up in step 2
Type is learnt and trained, and is finally completed the prediction to spinning quality.
2. the yarn quality prediction method according to claim 1 based on fireworks algorithm improvement BP neural network, its feature
It is, the optimizing mechanism of fireworks algorithm is optimized to the network weight and threshold value of BP neural network model in the step 1
Concretely comprise the following steps：
Step 1.1, key parameter is encoded, and the coding strategy for choosing real number vector is encoded to the key parameter in model, is remembered
Vectorial X=[x1,x2,…,xD] one group of parameter to be optimized is represented, it is made up of per one-dimensional vector network weight and threshold value, fireworks
The dimension of population is：D=nIW(1,1)+nb(1,1)+nIW(2,1)+nb(2,1), wherein, remember nIW(1,1)For the power of hidden layer and output interlayer
The number of value, nb(1,1)For the number of hidden layer neuron threshold value, nIW(2,1)For hidden layer with output interlayer weights number,
nb(2,1)The number of output layer neuron threshold value；
Step 1.2, weight system and threshold value initialization, on the basis of step 1.1, utilize fireworks individual x in fireworks algorithmik's
Neuron in positional representation neutral net, by i-th of neuron and jth in l layers of iterative process network of kth in neutral net time
Individual interneuronal weight coefficientAnd threshold θiInitialization is encoded into vectorial X=[x1,x2,…,xD], and utilize random first
The strategy of beginningization is initially at vector X in interval [- 1,1], then has weight coefficient wij~U [- 1,1],
Wherein, i, j refer respectively to the weight between i-th each neuron node and each neuron node of jth in network, what l was represented
It is the network number of plies residing for this present weight, what k was represented is current iterations；
Step 1.3, the error of fireworks individual is calculated, fitness function is introduced and calculates square mistake using formula (1) and formula (2)
Poor SSE, formula (1) and formula (2) are as follows：
<mrow>
<mi>S</mi>
<mi>S</mi>
<mi>E</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>p</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>t</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>s</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>-</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, t is the desired output of network, and p is the number of plies of network, and s is the number of network output unit, and y is network output valve,
Its formula specific as follows：
<mrow>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<msub>
<mi>f</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>w</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, xjFor the input of network, wijFor the weight of network node, θiFor the threshold value and θ of i-th of neuron in networki=-
wi(n+1)；
Step 1.4, obtained each fireworks individual x is calculated in step 1.3iOn the basis of error, f is introducedi(x) function is as suitable
Response function, passes through each fireworks of vector X individual x in fitness function calculation procedure 1.2iFitness value, fitness letter
Number such as formula (3) is as follows,
<mrow>
<msub>
<mi>f</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>p</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>t</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>s</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>-</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>p</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>t</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>s</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>-</mo>
<msub>
<mi>f</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mrow>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>w</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
1
Step 1.5, fireworks population optimizing, on the basis of step 1.4, for each fireworks individual xiExploded, displacement
And mutation operation, wherein blast mutation operation and Gaussian mutation mapping ruler are formula (4)~formula (6),
H=Ai×rand(1,-1) (4)
exik=xik+h (5)
mxik=xik×e (6)
Wherein, AiFor the burst radius of i-th of fireworks, h is position offset, xikThe kth dimension of i-th of fireworks in population is represented,
exikFor spark of i-th of fireworks after blast, mxikFor xikGaussian mutation spark after Gaussian mutation, e~N (1,1)
Gaussian Profile；
Step 1.6, fireworks population of future generation is selected, for the fireworks in step 1.5 after blast, displacement and mutation operation
Body xi, each fireworks individual x is calculated using the formula in step 1.4iFitness value, and use formula (7) and formula (8)
Selection strategy, selects optimal fireworks individual composition fireworks population of future generation, and specific selection strategy is：
Min (f (the x for selecting fitness value minimumi)) individual xkIt is directly fireworks population at individual, remaining N-1 fireworks
Body takes roulette mode, for candidate individual xiIts selected probability such as following formula：
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>(</mo>
<mfrac>
<mrow>
<mi>R</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>&Element;</mo>
<mi>K</mi>
</mrow>
</munder>
<mi>R</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>)</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, R (xi) represent fireworks individual xiWith other individuals apart from sum, formula specific as follows；
<mrow>
<mi>R</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</munderover>
<mi>d</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</munderover>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>|</mo>
<mo>|</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Step 1.7, judge end condition, the fitness value of fireworks individual in fireworks population is calculated according to formula (3) and formula (8)
f(xi) fireworks individual between Euclidean distance R (xi), and judge whether the maximum iteration that is reached in end condition, if meeting
Then calculate the minimum fitness value min (f (x of the fireworks individual obtained in current fireworks populationi)) and fireworks population in fireworks
It is maximum apart from max (R (x between individuali)), and take current fireworks population to be optimal fireworks population Xbest, otherwise continue to hold
Row step 1.3；
Step 1.8, optimization network weight and threshold value, using obtaining optimal fireworks population X in step 1.7bestTo in step 1.2
Weight and threshold value in vectorial X in corresponding neutral net are initialized.
3. the yarn quality prediction method according to claim 1 based on fireworks algorithm improvement BP neural network, its feature
It is, concretely comprising the following steps for the predictive model for yarn quality based on FWA-BP is built in the step 2：
Step 2.1, the selection of Input and Output Indexes：Choose raw material related to yarn qualities in Spinning process process, work
The data such as skill are as input variable, and the CV values for choosing yarn are output-index, then the whole yarn qualities prediction based on FWA-BP
The input and output of model are：
Input quantity is：X1=sliver percentage of impuritys, x2=roving twist factors, x3=regains, x4=fibre diameters, x5=fibers are long
Degree, x6=diameter coefficient of dispersion, x7=fiber quality irregularities, x8=drawing of fiber multiples, x9=spun yarn wire loops number, x10
=roller rotating speed；Output quantity is：Y=yarn CV values；
Step 2.2, the inputoutput data obtained according to step 2.1 sets up the data set of model, and uses Min-Max methods pair
Data in data set are standardized；
Step 2.3, the strategy of network structure is determined, according to the input, output-index chosen in step 2.1, it is determined that input, output
And the number of plies of hidden layer, the nodes m=10 of the input layer of FWA-BP predictive model for yarn quality, output layer nodes n=1,
The number of wherein hidden neuron is determined by following formula
<mrow>
<mi>s</mi>
<mo>=</mo>
<msqrt>
<mrow>
<mn>0.43</mn>
<mi>m</mi>
<mi>n</mi>
<mo>+</mo>
<mn>0.12</mn>
<msup>
<mi>n</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<mn>2.54</mn>
<mi>m</mi>
<mo>+</mo>
<mn>0.77</mn>
<mi>n</mi>
<mo>+</mo>
<mn>0.35</mn>
</mrow>
</msqrt>
<mo>+</mo>
<mn>0.51</mn>
<mo>&ap;</mo>
<mn>7</mn>
</mrow>
Calculating obtains s=7；
Step 2.4, the selection of activation primitive, input layer uses tansig activation primitives, and output layer activates letter using purelin
Number, chooses trainlm functions as the training function of network model.
4. the yarn quality prediction method according to claim 3 based on fireworks algorithm improvement BP neural network, its feature
It is, using the data set Jing Guo standardization to the spinning quality based on FWA-BP set up in step 2 in the step 3
Forecast model is learnt and concretely comprising the following steps of predicting：
Step 3.1, the selection strategy of training dataset, using the data set in step 2.2 Jing Guo standardization, is therefrom selected
80% data set is used as test data set as training dataset, the data set of residue 20%；
Step 3.2, in fireworks algorithm key parameter setting, the size N=70 of fireworks population, fireworks burst radius regulating constant
Dividing value lm=0.8 in d=5, fireworks explosive spark number regulating constant m=40, fireworks explosive spark number, fireworks explosive spark
Number floor value bm=0.04, Gaussian mutation spark number g=5, wherein maximum iteration T=100, variable dimension D=85, be
The sum of neuron weight and threshold value in network model is taken on the basis of step 1.1, is specifically in the basis of step 2.3
It is upper to be calculated by equation below
D=m × s+s × n+s+n=10 × 7+7 × 1+7+1=85
Wherein, m, s, n are respectively the number of input layer, hidden layer neuron and the output layer neuron of network；
Step 3.3, on the basis that fireworks algorithm parameter is set in step 3.2, the training data selected in step 3.1 is used
Predictive model for yarn quality of the set pair based on FWA-BP is trained, and related parameter is set in network training process, is learned
It is 0.01 to practise speed, and factor of momentum is 0.9, and maximum iteration is 20000, and training minimal error is 0.05；
Step 3.4, trained by step 3.1~3.3 and obtained the predictive model for yarn quality based on FWA-BP, use step
The test data set selected in 3.1, test statisticses analysis and experiment simulation are carried out to the prediction effect of model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710288559.2A CN107169565B (en) | 2017-04-27 | 2017-04-27 | Spinning quality prediction method for improving BP neural network based on firework algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710288559.2A CN107169565B (en) | 2017-04-27 | 2017-04-27 | Spinning quality prediction method for improving BP neural network based on firework algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107169565A true CN107169565A (en) | 2017-09-15 |
| CN107169565B CN107169565B (en) | 2020-06-19 |
Family
ID=59813156
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710288559.2A Expired - Fee Related CN107169565B (en) | 2017-04-27 | 2017-04-27 | Spinning quality prediction method for improving BP neural network based on firework algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107169565B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107395433A (en) * | 2017-08-18 | 2017-11-24 | 中南大学 | A kind of wireless sensor node dispositions method based on fireworks algorithm |
| CN107994570A (en) * | 2017-12-04 | 2018-05-04 | 华北电力大学（保定） | A kind of method for estimating state and system based on neutral net |
| CN108197696A (en) * | 2018-01-31 | 2018-06-22 | 湖北工业大学 | A kind of network navy account recognition methods and system |
| CN108241298A (en) * | 2018-01-09 | 2018-07-03 | 南京航空航天大学 | A kind of aerogenerator method for diagnosing faults based on FWA-RNN models |
| CN109542103A (en) * | 2018-12-25 | 2019-03-29 | 北京理工大学 | A kind of robot welding paths planning method based on fireworks particle swarm algorithm |
| CN109636014A (en) * | 2018-11-28 | 2019-04-16 | 郑州轻工业学院 | A kind of method of distributing cotton based on yarn quality prediction |
| CN110032069A (en) * | 2019-04-02 | 2019-07-19 | 东华大学 | A kind of polyester fiber spinning process segmentation parameter configuration method based on error compensation |
| CN111277141A (en) * | 2020-02-29 | 2020-06-12 | 武汉理工大学 | Bidirectional DC/DC converter optimization control method |
| CN111460738A (en) * | 2020-04-16 | 2020-07-28 | 中南大学 | RNN-ARX modeling method and RNN-ARX model of magnetic suspension system |
| CN112529684A (en) * | 2020-11-27 | 2021-03-19 | 百维金科(上海)信息科技有限公司 | Customer credit assessment method and system based on FWA _ DBN |
| CN112733720A (en) * | 2021-01-12 | 2021-04-30 | 上海理工大学 | Face recognition method based on firework algorithm improved convolutional neural network |
| CN112765902A (en) * | 2021-02-09 | 2021-05-07 | 嘉兴学院 | RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof |
| CN113112101A (en) * | 2020-01-10 | 2021-07-13 | 北京百度网讯科技有限公司 | Method and device for monitoring production process, electronic equipment and storage medium |
| CN113408963A (en) * | 2021-07-28 | 2021-09-17 | 上海致景信息科技有限公司 | Textile yarn quality rating method and device, storage medium and processor |
| CN113919601A (en) * | 2021-12-09 | 2022-01-11 | 山东捷瑞数字科技股份有限公司 | Resin process prediction method and device based on product performance and process data model |
| CN114065631A (en) * | 2021-11-18 | 2022-02-18 | 福州大学 | Energy consumption prediction method and system for laser cutting of plate |
| CN114994289A (en) * | 2022-08-01 | 2022-09-02 | 江苏卓鹏智能机电有限公司 | Spinning product quality detection method and system |
| CN115099490A (en) * | 2022-06-24 | 2022-09-23 | 无锡物联网创新中心有限公司 | Yarn quality prediction method and related device |
| CN115221791A (en) * | 2022-07-27 | 2022-10-21 | 浙江大学 | Supercritical boiler wall temperature online prediction method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105652952A (en) * | 2016-04-18 | 2016-06-08 | 中国矿业大学 | Maximum power point tracking method for photovoltaic power generation system based on fireworks algorithm |
| CN205688082U (en) * | 2016-04-26 | 2016-11-16 | 苏州精卫智能科技有限公司 | Yarn qualities predictor |
| CN106503661A (en) * | 2016-10-25 | 2017-03-15 | 陕西师范大学 | Face gender identification method based on fireworks depth belief network |
-
2017
- 2017-04-27 CN CN201710288559.2A patent/CN107169565B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105652952A (en) * | 2016-04-18 | 2016-06-08 | 中国矿业大学 | Maximum power point tracking method for photovoltaic power generation system based on fireworks algorithm |
| CN205688082U (en) * | 2016-04-26 | 2016-11-16 | 苏州精卫智能科技有限公司 | Yarn qualities predictor |
| CN106503661A (en) * | 2016-10-25 | 2017-03-15 | 陕西师范大学 | Face gender identification method based on fireworks depth belief network |
Non-Patent Citations (3)
| Title |
|---|
| RAM KINKAR DUTTA等: "《Artificial Neural Network Training using Fireworks》", 《INTERNATIONALJOURNAL OF COMPUTER APPLICATIONS》 * |
| 刘有时等: "《基于ＢＰ神经网络组合预测的纺纱产量预测研究》", 《江苏纺织》 * |
| 苌康群: "《烟花算法优化极限学习机的研究及应用》", 《中国秀硕士论文辑信息科技辑》 * |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107395433B (en) * | 2017-08-18 | 2020-06-09 | 中南大学 | Wireless sensor node deployment method based on firework algorithm |
| CN107395433A (en) * | 2017-08-18 | 2017-11-24 | 中南大学 | A kind of wireless sensor node dispositions method based on fireworks algorithm |
| CN107994570A (en) * | 2017-12-04 | 2018-05-04 | 华北电力大学（保定） | A kind of method for estimating state and system based on neutral net |
| CN108241298A (en) * | 2018-01-09 | 2018-07-03 | 南京航空航天大学 | A kind of aerogenerator method for diagnosing faults based on FWA-RNN models |
| CN108197696A (en) * | 2018-01-31 | 2018-06-22 | 湖北工业大学 | A kind of network navy account recognition methods and system |
| CN109636014A (en) * | 2018-11-28 | 2019-04-16 | 郑州轻工业学院 | A kind of method of distributing cotton based on yarn quality prediction |
| CN109542103A (en) * | 2018-12-25 | 2019-03-29 | 北京理工大学 | A kind of robot welding paths planning method based on fireworks particle swarm algorithm |
| CN109542103B (en) * | 2018-12-25 | 2019-12-20 | 北京理工大学 | Robot welding path planning method based on firework particle swarm algorithm |
| CN110032069A (en) * | 2019-04-02 | 2019-07-19 | 东华大学 | A kind of polyester fiber spinning process segmentation parameter configuration method based on error compensation |
| CN110032069B (en) * | 2019-04-02 | 2020-09-15 | 东华大学 | Polyester fiber spinning process segmented parameter configuration method based on error compensation |
| CN113112101A (en) * | 2020-01-10 | 2021-07-13 | 北京百度网讯科技有限公司 | Method and device for monitoring production process, electronic equipment and storage medium |
| CN113112101B (en) * | 2020-01-10 | 2023-07-25 | 北京百度网讯科技有限公司 | Method and device for monitoring production process, electronic equipment and storage medium |
| CN111277141A (en) * | 2020-02-29 | 2020-06-12 | 武汉理工大学 | Bidirectional DC/DC converter optimization control method |
| CN111460738A (en) * | 2020-04-16 | 2020-07-28 | 中南大学 | RNN-ARX modeling method and RNN-ARX model of magnetic suspension system |
| CN111460738B (en) * | 2020-04-16 | 2023-06-16 | 中南大学 | RNN-ARX modeling method and RNN-ARX model of magnetic suspension system |
| CN112529684A (en) * | 2020-11-27 | 2021-03-19 | 百维金科(上海)信息科技有限公司 | Customer credit assessment method and system based on FWA _ DBN |
| CN112733720A (en) * | 2021-01-12 | 2021-04-30 | 上海理工大学 | Face recognition method based on firework algorithm improved convolutional neural network |
| CN112765902A (en) * | 2021-02-09 | 2021-05-07 | 嘉兴学院 | RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof |
| CN112765902B (en) * | 2021-02-09 | 2024-02-20 | 嘉兴学院 | Soft measurement modeling method for COD concentration in rural domestic sewage treatment process based on TentFWA-GD RBF neural network |
| CN113408963A (en) * | 2021-07-28 | 2021-09-17 | 上海致景信息科技有限公司 | Textile yarn quality rating method and device, storage medium and processor |
| CN114065631A (en) * | 2021-11-18 | 2022-02-18 | 福州大学 | Energy consumption prediction method and system for laser cutting of plate |
| CN113919601A (en) * | 2021-12-09 | 2022-01-11 | 山东捷瑞数字科技股份有限公司 | Resin process prediction method and device based on product performance and process data model |
| CN115099490A (en) * | 2022-06-24 | 2022-09-23 | 无锡物联网创新中心有限公司 | Yarn quality prediction method and related device |
| CN115221791A (en) * | 2022-07-27 | 2022-10-21 | 浙江大学 | Supercritical boiler wall temperature online prediction method and system |
| CN114994289A (en) * | 2022-08-01 | 2022-09-02 | 江苏卓鹏智能机电有限公司 | Spinning product quality detection method and system |
| CN114994289B (en) * | 2022-08-01 | 2022-10-28 | 江苏卓鹏智能机电有限公司 | Spinning product quality detection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107169565B (en) | 2020-06-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107169565A (en) | Yarn quality prediction method based on fireworks algorithm improvement BP neural network | |
| CN105930901B (en) | A kind of Diagnosis Method of Transformer Faults based on RBPNN | |
| Lu et al. | Evaluation of English interpretation teaching quality based on GA optimized RBF neural network | |
| CN109932903A (en) | The air-blower control Multipurpose Optimal Method of more parent optimization networks and genetic algorithm | |
| CN105574589B (en) | Transformer oil chromatographic method for diagnosing faults based on niche genetic algorithm | |
| CN103105246A (en) | Greenhouse environment forecasting feedback method of back propagation (BP) neural network based on improvement of genetic algorithm | |
| CN108038300A (en) | Optical fiber state evaluating method based on improved membership function combination neutral net | |
| CN106355192A (en) | Support vector machine method based on chaos and grey wolf optimization | |
| CN109558677A (en) | A kind of hot rolling strip crown prediction technique based on data-driven | |
| CN107400935B (en) | Adjusting method based on the melt-spinning technology for improving ELM | |
| CN104636801A (en) | Transmission line audible noise prediction method based on BP neural network optimization | |
| CN103699812B (en) | Plant variety authenticity identification site selection method based on genetic algorithm | |
| CN107203659B (en) | A kind of subcooled flow boiling heat exchange quality transmitting relaxation factor prediction technique | |
| CN108460461A (en) | Mars earth shear parameters prediction technique based on GA-BP neural networks | |
| CN105808426A (en) | Path coverage test data generation method used for weak mutation test | |
| CN106845012A (en) | A kind of blast furnace gas system model membership function based on multiple target Density Clustering determines method | |
| CN108224446A (en) | A kind of automatic combustion Study on Decision-making Method for Optimization of Refuse Incineration Process | |
| CN109886448A (en) | Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm | |
| CN108038538A (en) | Multi-objective Evolutionary Algorithm based on intensified learning | |
| CN107909154A (en) | A kind of air control Model Parameter Optimization method based on web search | |
| CN106096723A (en) | A kind of based on hybrid neural networks algorithm for complex industrial properties of product appraisal procedure | |
| CN105425583A (en) | Control method of penicillin production process based on cooperative training local weighted partial least squares (LWPLS) | |
| CN104656620A (en) | Comprehensive evaluation system for remanufacturing of heavy-duty machine tool | |
| CN109508498A (en) | Rubber shock absorber formula designing system and method based on BP artificial neural network | |
| CN106446315A (en) | Method for predicting performance indexes in polyester fiber spinning process |
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: 20200619 Termination date: 20210427 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |