CN117332698A - ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method - Google Patents
ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method Download PDFInfo
- Publication number
- CN117332698A CN117332698A CN202311464158.XA CN202311464158A CN117332698A CN 117332698 A CN117332698 A CN 117332698A CN 202311464158 A CN202311464158 A CN 202311464158A CN 117332698 A CN117332698 A CN 117332698A
- Authority
- CN
- China
- Prior art keywords
- bat
- aro
- neural network
- algorithm
- outlet temperature
- 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.)
- Pending
Links
- 238000005096 rolling process Methods 0.000 title claims abstract description 90
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000004519 manufacturing process Methods 0.000 claims abstract description 16
- 238000005457 optimization Methods 0.000 claims abstract description 13
- 238000010606 normalization Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 5
- 230000009191 jumping Effects 0.000 claims abstract description 3
- 229910000831 Steel Inorganic materials 0.000 claims description 19
- 239000010959 steel Substances 0.000 claims description 19
- 241000283973 Oryctolagus cuniculus Species 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 7
- 230000002431 foraging effect Effects 0.000 claims description 6
- 230000008034 disappearance Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000011161 development Methods 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 241000209504 Poaceae Species 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 claims description 2
- 238000004880 explosion Methods 0.000 claims description 2
- 230000006872 improvement Effects 0.000 claims description 2
- 238000012804 iterative process Methods 0.000 claims description 2
- 230000007246 mechanism Effects 0.000 claims description 2
- 244000062645 predators Species 0.000 claims description 2
- 230000008569 process Effects 0.000 claims description 2
- 238000010079 rubber tapping Methods 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 238000003860 storage Methods 0.000 claims description 2
- 210000004027 cell Anatomy 0.000 description 12
- 238000003062 neural network model Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 239000000969 carrier Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000005098 hot rolling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000009628 steelmaking Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
- B21B—ROLLING OF METAL
- B21B37/00—Control devices or methods specially adapted for metal-rolling mills or the work produced thereby
- B21B37/74—Temperature control, e.g. by cooling or heating the rolls or the product
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
Abstract
The invention provides a hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM neural network, wherein the method comprises the steps of selecting parameters affecting the hot continuous rolling slab rough rolling outlet temperature; processing the experimental data by adopting a normalization method to obtain a normalized sample set; constructing an LSTM neural network according to the normalized sample set; optimizing an ARO algorithm by using a BAT algorithm to obtain an ARO-BAT optimization algorithm, optimizing an LSTM neural network by using the ARO-BAT algorithm to obtain a model constructed by the ARO-BAT-LSTM neural network, acquiring the maximum value of the constructed model, and determining an optimal solution according to the maximum value of the constructed model to determine optimal parameters; and determining the prediction precision of the hot-rolled slab rough rolling outlet temperature prediction model according to the comparison result of the predicted value of the optimal parameter and the actual value in the actual production. The model has the advantages of high convergence speed, high prediction precision, self-adaptive search strategy, capability of jumping out of a local optimal solution and strong generalization capability. Compared with the prior art, the method can better meet the requirement on the accuracy of the prediction of the rough rolling outlet temperature of the hot continuous rolling slab, and provide theoretical guidance for actual production.
Description
Technical Field
The invention relates to the technical field of steelmaking, in particular to a hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM algorithm.
Background
Along with the progress of the steel industry and the improvement of the living standard of people, in order to meet the social requirement of high-precision and high-quality steel, the requirements of factories on the hot continuous rolling production process of strip steel are also higher and higher, so that a more accurate hot rolling strip steel temperature prediction model is needed to improve the quality of the steel, and meanwhile, the utilization rate of energy sources is improved, and the further development of the steel industry is promoted.
The hot continuous rolling production process of the strip steel is complex, the data dimension is high, the temperature of the rolled piece is used as an important parameter in the production process of the steel, the rough rolling outlet temperature of the plate blank is rapidly and accurately predicted, the internal microstructure, grain size, mechanical property and surface state of the strip steel can be effectively controlled, and the mechanical property and product quality of the finished strip steel are further improved. Therefore, the prediction of the rough rolling outlet temperature of the hot continuous rolling slab is particularly important.
In the actual hot continuous rolling production process, a machine-made model is generally adopted at present to predict the rough rolling outlet temperature of the slab. In practical application, the existing mechanism model has the defects of low prediction precision, long operation time and the like. Therefore, more advanced modeling strategies are continuously explored to improve the accuracy of the prediction model, improve the performance of strip steel products for enterprises, save the cost of raw materials and reduce waste.
In summary, in order to solve the above problems, the present invention provides a hot continuous rolling slab rough rolling outlet temperature prediction method based on ARO-BAT-LSTM neural network based on various types of prediction models.
Disclosure of Invention
In view of the problems, the invention aims to provide a hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM neural network, which can solve the problems of low accuracy and low speed of hot continuous rolling slab rough rolling outlet temperature prediction.
The invention provides a hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM neural network, which comprises the following steps: according to the production process of hot continuous rolling of the strip steel, selecting parameters influencing the rough rolling outlet temperature of a hot continuous rolling plate blank;
constructing a modeling sample set by utilizing parameter experiment data acquired from a certain steel mill;
processing the experimental data by adopting a normalization method to obtain a normalization sample set;
constructing LSTM neural network model according to normalized sample set
Optimizing the ARO algorithm by using the BAT algorithm to obtain an ARO-BAT optimization algorithm
And optimizing the LSTM neural network by using an ARO-BAT algorithm to obtain an ARO-BAT-LSTM prediction model, and determining an optimal solution according to the maximum value of the constructed model to determine optimal parameters.
And obtaining the prediction precision of the rough rolling outlet temperature of the hot continuous rolling slab according to the comparison result of the predicted value of the optimal parameter and the target value in the modeling sample set.
According to the technical scheme, the ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method provided by the invention is characterized in that production operation parameters in the hot continuous rolling production process are used as information carriers, a LSTM algorithm is firstly utilized to construct a neural network model, the ARO-BAT algorithm is utilized to mix the neural network for optimization, the optimal solution is obtained, the optimal prediction value is obtained through calculation, a scheme is provided for uncertainty in the hot continuous rolling production process, and the problems of low accuracy and low speed of hot continuous rolling slab rough rolling outlet temperature prediction are solved.
To the accomplishment of the foregoing and related ends, the invention, then, comprises the features hereinafter fully described and particularly pointed out in the claims. The following description of the drawings and the detailed description set forth certain exemplary aspects of the invention. These aspects are indicative, however, of but a few of the various ways in which the principles of the invention may be employed. Furthermore, the invention is intended to include all such aspects and their equivalents.
Drawings
Other objects and results of the present invention will become more apparent and readily appreciated by reference to the following description and claims in conjunction with the accompanying drawings and a more complete understanding of the invention. In the drawings:
FIG. 1 is a schematic flow chart of a hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM neural network according to an embodiment of the invention;
FIG. 2 is a schematic diagram of the internal architecture of an LSTM neural network constructed in accordance with an embodiment of the invention;
FIG. 3 is a schematic diagram of an ARO-BAT optimization flow of an embodiment;
fig. 4 is a schematic diagram of prediction accuracy, which is a small comparison between the predicted hot continuous rolling slab rough rolling outlet temperature and the actual hot continuous rolling slab rough rolling outlet temperature in the embodiment;
the same reference numerals will be used throughout the drawings to refer to similar or corresponding features or functions.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples.
As shown in FIG. 1, the hot continuous rolling slab rough rolling outlet temperature prediction method based on the ARO-BAT-LSTM neural network comprises the following steps:
s110: according to the production process of hot continuous rolling of the strip steel, selecting parameters influencing the rough rolling outlet temperature of a hot continuous rolling plate blank;
s120: constructing a modeling sample set by utilizing parameter experiment data acquired from a certain steel mill;
s130: processing the experimental data by adopting a normalization method to obtain a normalization sample set;
s140: constructing an LSTM neural network model according to the normalized sample set;
s150: optimizing an ARO algorithm by using a BAT algorithm to obtain an ARO-BAT optimization algorithm;
s160: optimizing the LSTM neural network by using an ARO-BAT algorithm to obtain an ARO-BAT-LSTM prediction model, and determining an optimal solution according to the maximum value of the constructed model to determine an optimal parameter;
s170: and obtaining the prediction precision of the rough rolling outlet temperature of the hot continuous rolling slab according to the comparison result of the predicted value of the optimal parameter and the target value in the modeling sample set.
In the step S110, in the actual rolling process, because there are a plurality of relevant factors that affect the rough rolling outlet temperature of the hot continuous rolling slab and the rough rolling outlet temperature of the hot continuous rolling slab, the mechanical properties and the product quality of the finished strip steel are reduced, and therefore, the relevance ranking is performed among the plurality of influencing factors, so as to obtain the parameter variable of the rough rolling outlet temperature of the hot continuous rolling slab, and verify whether the parameter variable is properly selected. Therefore, the invention adopts the inlet temperature (charging temperature), the slab thickness, the slab length, the slab width, the actual tapping temperature, the time in the furnace, the slab quality and the like as the control parameters affecting the rough rolling outlet temperature of the hot continuous rolling slab; wherein the control parameters affecting the cost are shown in table 1.
Table 1 parameters and symbol table
In step S120, a modeling sample set [ X; Y ] is constructed by using experimental data of different parameters collected from a steel mill; the data collected are shown in table 2:
table 2 data acquisition sample portion data
In step S130, data is preprocessed. Since there is generally an order of magnitude difference between the input variables, resulting in a reduction in the predictive performance of the model, experimental data is processed using a normalization method. Namely: the parameter values of the sample set are mapped to [ -1 using a linear normalization method,1]within the range, a normalized sample set is obtained]。
In step S140, an LSTM neural network model is constructed from the normalized sample set, wherein the LSTM theory is as follows:
the LSTM network internally comprises key components such as an input gate, a forgetting gate, an output gate and the like, so that the key components can selectively retain, forget and output information in different time steps, and the gradient disappearance problem in long-sequence training is effectively solved. And calculating the values of the input gate, the forget gate and the output gate according to the input of the current time step and the hidden state of the previous time step.
The value of the input gate is used to update the cell state, adding new information to the cell state.
The value of the forgetting gate is used to forget unwanted cell state information.
And screening the updated cell state through an output gate to generate the output of the current time step.
The hidden state and the cell state of the current time step are passed to the next time step as input to the next time step.
、、、Outputting a connection weight value for an input gate for the input of the data at the current moment and the LSTM unit at the previous moment;
、、、bias for each gate and memory cell;
the state value of the candidate memory cell at the time t;
the state value of the current memory cell at the time t;
outputting a value of the gate for the time t;
the internal structure of the LSTM neural network is shown in figure 2;
the LSTM neural network is a variant of the traditional RNN neural network, and can effectively capture the correlation between long sequences and effectively reduce the gradient disappearance or explosion phenomenon.
STM neural network functions can be described as:
wherein:forget the value of the gate for the time t;is a Sigmoid function;in order to input the amount of the input,the output of the memory unit at the time t,representing the presentation to beAndsplicing;、、、outputting a connection weight value for an input gate for the input of the data at the current moment and the LSTM unit at the previous moment;、、、bias for each gate and memory cell;the state value of the candidate memory cell at the time t;the state value of the current memory cell at the time t;the value of the gate is output for time t.
In step S150, the ARO algorithm is optimized using the BAT algorithm to obtain the ARO-BAT optimization algorithm, the invention embeds the BAT algorithm into the ARO search stage, takes the input weight and threshold of the ARO as the BAT position in the BAT algorithm, and iterates through the BAT optimization algorithm until the optimal solution is generated. FIG. 3 shows a network parameter optimization flow of ARO-BAT-LSTM according to an embodiment of the invention, the steps of algorithm chart 3 for optimizing ARO based on BAT algorithm are as follows:
the first step: initializing a population, and supposing that each rabbit in the population has own area, and has some grasses and d cavities, namely d problem dimensions;
and a second step of: the exploration phase, which performs random food searches based on each other's location, may be referred to as detour foraging, is expressed as follows:
wherein,is the candidate position of the ith rabbit in the t+1st iteration;is the ith rabbit at the t-th iterationN is the number of rabbit groups, d is the dimension of the problem, T is the maximum iteration number;is an upward rounding function; round represents rounding; random (d) represents a random permutation of integers returning from 1 to d; random numbers in the (0, 1) interval; l is running length and represents the movement speed when the user walks around to find food;is a random number subject to a standard normal distribution.
The BAT algorithm is embedded, and the following specific formula is as follows:
(1) Setting bat population size, maximum pulse sound intensity A and maximum pulse frequencyUpper limit of pulse frequencyLower limit ofSound intensity attenuation coefficientCoefficient of frequency increaseThe dimension of the position vector is set to n,
(2) Entering a searching stage, initializing pulse frequency of the unit bat, calculating flying speed of the unit bat, updating position of the bat, and updating the formula as follows:
wherein, rand is a random factor and is uniformly distributed in the (0, 1) interval;for the i-th bat's flight speed at t-1 and time,、representing the position.
(3) In each iteration, a random number rand1 is generated for the cell bat, if rand1>Selecting the current optimal solution to perform local disturbance, wherein the formula is as follows:
wherein,the pulse frequency for the ith bat.
(4) Calculating new fitness of the bat after disturbance, if the new fitness is better than the optimal fitness or rand2<The new position after disturbance is used for replacing the old position for storage, and the pulse frequency and the sound intensity are updated at the same time, and the specific formula is as follows:
(5) If the end condition is reached, stopping searching and outputting the position of the unit bat corresponding to the global optimal solution; otherwise, the search is continued by jumping back to (2).
And a third step of: in the development stage, cavities are generated and one is randomly selected to serve as a hiding mode, wherein a specific formula generated by the j cavity of the i rabbit is as follows;
wherein H is a hidden parameter, and in the iterative process, the hidden parameter is linearly reduced from 1 to 1 along with random disturbanced represents the number of cavities created in the vicinity of the rabbit for each dimension.
Fourth step: to avoid being caught by predators, a random concealment strategy is performed, expressed as follows:
wherein,representing randomly selected holes for hiding d holes;andis thatRandom numbers within a range.
Fifth step: after one of detouring foraging and random hiding is realized, the position of the rabbit is updated as follows:
sixth step: an energy factor is designed to simulate the conversion process from the fifth step to the first step, the energy factor being defined as follows:
wherein r is a random number in (0, 1);
seventh step: if the energy factor isThe rabbits do detouring foraging; if the energy factor isThe rabbit race is randomly hidden.
Eighth step: repeating the second to seventh steps until the termination condition is satisfied.
Ninth step: repeating the iteration to obtain the current optimal solution;
Tenth step: determining a predicted value of the rough rolling outlet temperature of the hot continuous rolling slab according to the optimal solution;
in step S160, an ARO-BAT algorithm is used to optimize the LSTM neural network, obtain an ARO-BAT-LSTM prediction model, and determine an optimal solution according to the maximum value of the constructed model to determine an optimal parameter, where:
substituting the number of main parameter hiding layers and the number of neurons of the LSTM and the number of all-connection layers and the number of neurons as inputs, and obtaining the optimal value of the LSTM parameter after the ARO-BAT algorithm meets the termination condition. Initializing main parameters of an LSTM neural network, setting the main parameters as optimal parameters after ARO-BAT optimization, and finally obtaining an ARO-BAT-LSTM prediction model through a sample set training model.
And comparing the predicted value of the parameter obtained after global optimization with the actual value to obtain a final predicted result comparison, namely, the predicted precision is shown in figure 4. Meanwhile, along with continuous training of samples, the accuracy of the model is gradually improved, and the model accords with the characteristics of dynamic modeling.
According to the embodiment, the ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method provided by the invention is characterized in that the LSTM algorithm is firstly utilized to construct a neural network model in the hot continuous rolling production process by taking production operation parameters as information carriers, then the BAT algorithm is embedded into the ARO algorithm to optimize the LSTM model, so that an optimal solution is obtained, an optimal predicted value is obtained by calculation, a scheme is provided for uncertainty existing in the hot continuous rolling production process, and the problems of low accuracy and low speed of hot continuous rolling slab rough rolling outlet temperature prediction are solved.
The ARO-BAT-LSTM neural network created by the invention solves the problems of slow and coarse processing of large-scale complex data and multidimensional data of the traditional model, can train small sample data, can quickly and accurately solve the large-scale data, and can provide guidance for improving the quality of actual production products.
The hot continuous rolling slab rough rolling outlet temperature prediction method based on the ARO-BAT-LSTM neural network according to the present invention is described above by way of example with reference to the accompanying drawings. However, it will be appreciated by those skilled in the art that various modifications may be made to the ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method set forth in the foregoing invention without departing from the teachings of the present invention. Accordingly, the scope of the invention should be determined from the following claims.
Claims (5)
1. A hot continuous rolling slab rough rolling outlet temperature prediction method based on an ARO-BAT-LSTM neural network comprises the following steps: according to the production process of hot continuous rolling of the strip steel, selecting parameters influencing the rough rolling outlet temperature of a hot continuous rolling plate blank;
inlet temperature (charging temperature), slab thickness, slab length, slab width, actual tapping temperature, time in furnace, slab mass;
constructing a modeling sample set by utilizing parameter experiment data acquired from a certain steel mill;
processing the experimental data by adopting a normalization method to obtain a normalization sample set;
constructing an LSTM neural network according to the normalized sample set;
optimizing the ARO algorithm by using the BAT algorithm to obtain an ARO-BAT optimization algorithm
Optimizing the LSTM neural network by using an ARO-BAT algorithm to obtain an ARO-BAT-LSTM prediction model, and determining an optimal solution according to the maximum value of the constructed model to determine an optimal parameter;
and obtaining the prediction precision of the rough rolling outlet temperature of the hot continuous rolling slab according to the comparison result of the predicted value of the optimal parameter and the target value in the modeling sample set.
2. The ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method of claim 1, wherein, in constructing a feed-forward neural network according to the obtained normalized sample set and an extreme learning mechanism theory improvement algorithm;
the LSTM network internally comprises key components such as an input gate, a forgetting gate, an output gate and the like, so that the key components can selectively retain, forget and output information in different time steps, and the gradient disappearance problem in long-sequence training is effectively solved.
And calculating the values of the input gate, the forget gate and the output gate according to the input of the current time step and the hidden state of the previous time step.
The value of the input gate is used to update the cell state, adding new information to the cell state.
The value of the forgetting gate is used to forget unwanted cell state information.
And screening the updated cell state through an output gate to generate the output of the current time step.
The hidden state and the cell state of the current time step are passed to the next time step as input to the next time step.
、/>、/>、/>Outputting a connection weight value for an input gate for the input of the data at the current moment and the LSTM unit at the previous moment;
、/>、/>、/>bias for each gate and memory cell;
the state value of the candidate memory cell at the time t;
for the current memory cell at time tState values of (2);
the value of the gate is output for time t.
3. The hot continuous rolling rough rolling strip thickness prediction method based on ARO-BAT-LSTM neural network according to claim 2, wherein, in constructing the LSTM neural network,
the LSTM neural network is a variant of the traditional RNN neural network, and can effectively capture the correlation between long sequences and effectively reduce the gradient disappearance or explosion phenomenon.
LSTM neural network functions can be described as:
wherein:forget the value of the gate for the time t; />Is a Sigmoid function; />For input quantity, < >>For the output of the memory cell at time t, +.>The representation will->And->Splicing; />、/>、/>、/>Outputting a connection weight value for an input gate for the input of the data at the current moment and the LSTM unit at the previous moment; />、/>、/>、/>Bias for each gate and memory cell; />The state value of the candidate memory cell at the time t; />The state value of the current memory cell at the time t; />The value of the gate is output for time t.
4. The ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method of claim 1, wherein an ARO algorithm in constructing the ARO-BAT optimization algorithm is specifically as follows;
the first step: initializing a population, and supposing that each rabbit in the population has own area, and has some grasses and d cavities, namely d problem dimensions;
and a second step of: the exploration phase, which performs random food searches based on each other's location, may be referred to as detour foraging, is expressed as follows:
wherein,is the candidate position of the ith rabbit in the t+1st iteration; />Is the current position of the ith rabbit at the t-th iteration; n is the number of rabbit groups, d is the dimension of the problem, T is the maximum iterationThe number of times; />Is an upward rounding function; round represents rounding; random (d) represents a random permutation of integers returning from 1 to d; random numbers in the (0, 1) interval; l is running length and represents the movement speed when the user walks around to find food; />Random numbers conforming to a standard normal distribution;
and a third step of: in the development stage, cavities are generated and one is randomly selected to serve as a hiding mode, wherein a specific formula generated by the j cavity of the i rabbit is as follows;
wherein H is a hidden parameter, and in the iterative process, the hidden parameter is linearly reduced from 1 to 1 along with random disturbanced represents the number of cavities generated in the vicinity of the rabbit in each dimension;
fourth step: to avoid being caught by predators, a random concealment strategy is performed, expressed as follows:
wherein,representing randomly selected holes for hiding d holes; />And->Is->Random numbers within a range.
Fifth step: after one of detouring foraging and random hiding is realized, the position of the rabbit is updated as follows:
sixth step: an energy factor is designed to simulate the conversion process from the fifth step to the first step, the energy factor being defined as follows:
wherein r is a random number in (0, 1).
Seventh step: if the energy factor isThe rabbit is woundCarrying out row foraging; if energy factor->Randomly hiding the rabbit race;
eighth step: repeating the second step to the seventh step until a termination condition is met;
ninth step: and outputting the optimal solution to finish the optimization of the parameters.
5. The ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method according to claim 1, wherein the ARO algorithm is optimized using a BAT algorithm, the BAT algorithm is embedded in an exploration phase of the ARO algorithm,
the first step: setting bat population size, maximum pulse sound intensity A and maximum pulse frequencyUpper limit of pulse frequency->Lower limit ofSound intensity attenuation coefficient->Frequency increase coefficient->Setting the dimension of the position vector as n, and setting a group of parameters of the BAT searching stage corresponding to each optimal position obtained in the claim 2, so as to finish initialization;
and a second step of: entering a searching stage, initializing pulse frequency of the unit bat, calculating flying speed of the unit bat, updating position of the bat, and updating the formula as follows:
rand
wherein, rand is a random factor and is uniformly distributed in the (0, 1) interval;for the i-th bat flying speed at t-1 and moment, +.>、/>Representing the position;
and a third step of: in each iteration, a random number rand1 is generated for the cell bat, if rand1>Selecting the current optimal solution to perform local disturbance, wherein the formula is as follows:
wherein,pulse frequency for the ith bat;
fourth step: calculating new fitness of the bat after disturbance, if the new fitness is better than the optimal fitness or rand2<The new position after disturbance is used for replacing the old position for storage, and the pulse frequency and the sound intensity are updated at the same time, and the specific formula is as follows:
fifth step: if the end condition is reached, stopping searching and outputting the position of the unit bat corresponding to the global optimal solution; otherwise, jumping back to the second step to continue searching;
sixth step: optimizing the ARO-BAT algorithm, inputting LSTM main parameters, and searching for the optimal value of the parameters;
seventh step: and determining a predicted value of the rough rolling outlet temperature of the hot continuous rolling slab according to the optimal solution.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311464158.XA CN117332698A (en) | 2023-11-06 | 2023-11-06 | ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311464158.XA CN117332698A (en) | 2023-11-06 | 2023-11-06 | ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117332698A true CN117332698A (en) | 2024-01-02 |
Family
ID=89293310
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311464158.XA Pending CN117332698A (en) | 2023-11-06 | 2023-11-06 | ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117332698A (en) |
-
2023
- 2023-11-06 CN CN202311464158.XA patent/CN117332698A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Guo et al. | A reinforcement learning decision model for online process parameters optimization from offline data in injection molding | |
CN110640982A (en) | Multi-target optimization method for injection molding process parameters of thin-wall injection molding part | |
CN111047085B (en) | Hybrid vehicle working condition prediction method based on meta-learning | |
CN112507610B (en) | Interval prediction method for hot rolling coiling temperature | |
CN109242236B (en) | PMU primary frequency modulation load prediction method based on correlation full-connection neural network and LSTM | |
CN110070145A (en) | LSTM wheel hub single-item energy consumption prediction based on increment cluster | |
CN110111848A (en) | A kind of human cyclin expressing gene recognition methods based on RNN-CNN neural network fusion algorithm | |
CN111581829A (en) | Strip steel thickness prediction method for optimizing support vector regression by improved difference wolf algorithm | |
Guo et al. | An Improved Whale Optimization Algorithm for Feature Selection. | |
CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
CN110083125A (en) | A kind of machine tool thermal error modeling method based on deep learning | |
CN116861793B (en) | Method for modeling thermal error of pelican algorithm optimized long-term and short-term memory network main shaft | |
CN114548591A (en) | Time sequence data prediction method and system based on hybrid deep learning model and Stacking | |
CN113393057A (en) | Wheat yield integrated prediction method based on deep fusion machine learning model | |
CN110110447B (en) | Method for predicting thickness of strip steel of mixed frog leaping feedback extreme learning machine | |
CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
Chen et al. | LAP: Latency-aware automated pruning with dynamic-based filter selection | |
CN113128124B (en) | Multi-grade C-Mn steel mechanical property prediction method based on improved neural network | |
CN107527110A (en) | Based on cuckoo chess game optimization MGM(1, N)Prediction of annual electricity consumption method | |
Shene et al. | Neural networks for the prediction of the state of Zymomonas mobilis CP4 batch fermentations | |
CN113762370A (en) | Depth network set generation method combined with Gaussian random field | |
CN116689502A (en) | Roll bending force parameter setting method of finishing mill group process of hot continuous rolling mill based on SA-SCSO-SCN neural network | |
CN117521511A (en) | Granary temperature prediction method based on improved wolf algorithm for optimizing LSTM | |
CN112926251A (en) | Landslide displacement high-precision prediction method based on machine learning | |
CN117332698A (en) | ARO-BAT-LSTM neural network-based hot continuous rolling slab rough rolling outlet temperature prediction method |
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 |