CN109188903A - The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization - Google Patents

The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization Download PDF

Info

Publication number
CN109188903A
CN109188903A CN201810950550.8A CN201810950550A CN109188903A CN 109188903 A CN109188903 A CN 109188903A CN 201810950550 A CN201810950550 A CN 201810950550A CN 109188903 A CN109188903 A CN 109188903A
Authority
CN
China
Prior art keywords
convolutional neural
neural network
memory
layer
heating furnace
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
Application number
CN201810950550.8A
Other languages
Chinese (zh)
Inventor
王永健
李宏光
黄静雯
宿翀
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Chemical Technology
Original Assignee
Beijing University of Chemical Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing University of Chemical Technology filed Critical Beijing University of Chemical Technology
Priority to CN201810950550.8A priority Critical patent/CN109188903A/en
Publication of CN109188903A publication Critical patent/CN109188903A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention discloses a kind of flexible measurement methods of CNN furnace operation variable based on memory-enhancing effect optimization, training sample data are obtained first, the training sample data are pre-processed, the number of plies of convolutional neural networks is preset according to convolutional neural networks algorithm, every layer of number of nodes, initialize every layer of weight and biasing, then it is optimized using weight of the memory-enhancing effect optimization algorithm to convolutional neural networks, obtain optimal weighting parameter, to improve the modeling accuracy of model, the measured value of furnace operation variable is finally obtained according to the convolutional neural networks model that training is formed, the thermal efficiency of heating furnace is made to remain at optimum condition by the measured value of performance variable, to improve industrial production benefit.Technical solution provided by the invention can obtain stable network structure, superior Generalization Capability and higher precision, so that the quality of product is improved, so that heating furnace is remained at compared with even running under high thermal efficiency level.

Description

Memory enhancement optimization-based soft measurement method for CNN heating furnace operating variables
Technical Field
The invention relates to the technical field of heating furnace operation, in particular to a soft measurement method of CNN heating furnace operation variables based on memory enhancement optimization.
Background
In the chemical industry, the establishment of an accurate mathematical model of a controlled object plays a crucial role in the production of the process industry. A good mathematical model of the controlled object is established, and the cause-effect relationship of the object can be effectively described, so that the real-time monitoring of the working condition of the whole process is realized, and the performance and the efficiency of the process industry are improved. The following three modeling methods are commonly used: mechanism modeling, data-driven modeling, and hybrid modeling in which mechanisms are combined with data-driving. The mechanism modeling is to establish an accurate physical model, and usually requires sufficiently reliable prior knowledge and a large amount of practical experience. However, as the amount of data and complexity in industrial processes increases, accurate mechanistic models become increasingly difficult to obtain, which presents challenges to the modeling and optimization of the process industry. Meanwhile, with the use of a modern Distributed Control System (DCS), it is possible to acquire a large amount of data in the process industry. Therefore, many scholars have begun to adopt a data-driven modeling-based approach. The data-driven model is built by fitting the mapping relation between the input data and the output data based on historical sample data, so that the modeling process is greatly simplified.
Various data-driven modeling methods exist, such as signal processing, machine learning, artificial neural networks, and the like. Among the data-driven modeling methods, the modeling method based on the artificial neural network has the advantages of strong nonlinear mapping capability, self-adaptive capability, good generalization capability and the like. Any continuous non-linear function can be approximately fitted using a neural network. Therefore, the artificial neural network can be applied to complex flow system modeling with strong nonlinearity.
In building neural network models, accuracy and stability are often chosen as two important indicators for judging model performance. Although artificial neural networks are a common data-driven algorithm in fitting complex Back Propagation (BP). However, as the industrial data is more and more complex, the solution speed of the BP neural network is more and more difficult to meet the requirements of people due to the fully-connected characteristic of the BP neural network.
Disclosure of Invention
In order to solve the limitations and defects in the prior art, the invention provides a soft measurement method of CNN heating furnace operation variables based on memory enhancement optimization, which comprises the following steps:
obtaining training samples (X, Y);
-normalizing said training samples (X, Y) using a normalization formula;
initializing each parameter value of the convolutional neural network, wherein the parameter value comprises the number of network layers, the number of nodes and the numerical value of a convolutional kernel;
updating the weight of the convolution layer of the convolution neural network;
updating the weight of the downsampling layer of the convolutional neural network;
optimizing the weight of the convolutional neural network by using a memory enhancement optimization algorithm to form a convolutional neural network model;
and predicting the heating furnace operation variables by using the convolutional neural network model.
Optionally, the step of optimizing the weight of the convolutional neural network by using a memory-enhanced optimization algorithm includes:
when the error cost function of the convolutional neural network is minimum, obtaining the weight parameter of the convolutional neural network, wherein the error cost function of the convolutional neural network is as follows:
wherein,is as followsThe value of the error cost function when the weight value is updated for k +1 times,and epsilon is a constant parameter, and is a value of the sensitivity of the (k + 1) th time.
Optionally, the step of updating the weight of the convolutional layer of the convolutional neural network includes:
obtaining convolutional layer sensitivity of the convolutional neural network according to a convolutional layer sensitivity formula, wherein the convolutional layer sensitivity formula is as follows:
wherein, o is the multiplication of corresponding elements, W is the weight βjUp represents an upsampling operation as the value of the convolution kernel;
calculating the partial derivative of the error cost function to the convolution kernel k according to the sensitivity of the convolution layer:
wherein,is composed ofAndthe convolved patch has a position (u, v) that is the center of the patch, and the value of the position (u, v) in the output feature is determined by the value of the position (u, v) in the input feature and the convolution kernel kijAnd (4) performing convolution.
Optionally, the step of updating the weight of the downsampling layer of the convolutional neural network includes:
obtaining the downsampling layer sensitivity of the convolutional neural network according to a downsampling layer sensitivity formula, wherein the downsampling layer sensitivity formula is as follows:
wherein, o is the multiplication of corresponding elements, and rot180(·) is the anticlockwise rotation of the matrix by 180 degrees;
and calculating the partial derivative of the error cost function to the bias b according to the sensitivity of the down-sampling layer.
Optionally, the normalization formula is:
wherein,
Yminand YmaxThe minimum and maximum values of the output pattern vector Y, respectively.
The invention has the following beneficial effects:
the soft measurement method of CNN heating furnace operation variables based on memory enhancement optimization provided by the invention comprises the steps of firstly obtaining training sample data, preprocessing the training sample data, presetting the number of layers and the number of nodes of each layer of a convolutional neural network according to a convolutional neural network algorithm, initializing the weight and the offset of each layer, then optimizing the weight of the convolutional neural network by using a memory enhancement optimization algorithm to obtain an optimal weight parameter, thereby improving the modeling precision of the model, finally obtaining the measured value of the heating furnace operation variables according to the convolutional neural network model formed by training, and keeping the thermal efficiency of the heating furnace in an optimal state all the time through the measured value of the operation variables, thereby improving the industrial production benefit. The technical scheme provided by the invention can obtain a stable network structure, excellent generalization performance and higher precision, thereby improving the quality of products and ensuring that the heating furnace can be stably operated at a higher heat efficiency level all the time.
Drawings
Fig. 1 is a schematic structural diagram of a refinery reduced pressure heating furnace provided in the first embodiment of the present invention.
Fig. 2 is a schematic flow chart of a measurement model of an operating variable of a heating furnace according to an embodiment of the present invention.
Fig. 3 is a diagram illustrating a distribution of predicted results of a generalization procedure according to an embodiment of the present invention.
Fig. 4 is an error diagram of a measurement model of an operating variable of a heating furnace according to an embodiment of the present invention.
Detailed Description
In order to make those skilled in the art better understand the technical solution of the present invention, the soft measurement method of CNN furnace operation variables based on memory enhancement optimization provided by the present invention is described in detail below with reference to the accompanying drawings.
Example one
Fig. 1 is a schematic structural diagram of a refinery reduced pressure heating furnace provided in the first embodiment of the present invention. As shown in fig. 1, the heating furnace is an important device in the modern petrochemical production process, and is also a large energy consumer. The operating thermal efficiency of a furnace is affected by many factors, such as furnace design, equipment conditions, combustion regulation, process operation, operating load, and the like. Good operation of the heating furnace is an important way to improve the thermal efficiency. The opening degree of the air valve is often used as an operation variable, and relevant factors such as pressure, flow rate and temperature are used as detection targets and control targets for modeling and predicting the operation variable of the heating furnace. If the manipulated variables are always kept in the optimum state, the heating furnace can always maintain the optimum level of thermal efficiency.
Convolutional Neural Networks (CNN) is a deep learning method, and is generally used for processing rasterized data and time series data, and the characteristics of weight sharing and sparse connection greatly reduce the complexity of the algorithm and shorten the operation time of the algorithm. The object of this embodiment is a tubular heating furnace, and the production process data has time series characteristics, so this embodiment adopts the convolutional neural network method to perform soft measurement modeling on the operation variables of the production process.
Although the convolutional neural network can well process time sequence industrial data and model operation variables of the process production process, due to the fact that a large number of iterative computations exist, when weight values are updated in each iteration, the algorithm structure is not changed, and only parameters are updated. Aiming at the structural characteristics of the algorithm of the convolutional neural network, the embodiment provides a convolutional neural network method for memory enhancement optimization, which reserves the empirical parameters of previous iterative computation, and uses a self-adaptive method to find out the weight parameter which enables the output error to be minimum from the historical computation experience, so as to optimize the convolutional neural network model.
In order to solve the problem of complex modeling of the heating furnace at present, the embodiment provides a memory-enhanced optimization-based soft measurement method for CNN heating furnace operating variables, which is used for guiding the production of the heating furnace, improving the precision and the thermal efficiency of the operating variables, and further increasing the production benefits. The technical scheme provided by the embodiment comprises the following steps: and acquiring data, performing data preprocessing, initializing each parameter value of the CNN, optimizing the weight of the CNN by using a memory enhancement optimization algorithm, and modeling the operation variable of the heating furnace. The method of soft measurement of the operating variables of the heating furnace is described in detail below.
This embodiment obtains data and performs data preprocessing. Specifically, in this embodiment, missing data, abnormal data, and noise data existing in Pure Terephthalic Acid (PTA) data collected in the field are processed to finally obtain I samples { (X)i,Yi) 1, 2., I }, where X isi=[xi1,xi2,…,xin]∈RnRepresents the ith input sample, xinRepresents the ith input sample XiThe nth element of (1). Referring to Table 1, n elements correspond to pressure, feed amount, temperature, etc. in the production of the heating furnace, respectively, and Y isiE R represents the output vector-air valve opening. The present embodiment uses a multi-layered convolutional neural network to process the high non-linearity of petrochemical industry data to improve performance in terms of accuracy and stability. The embodiment realizes modeling of the operation variables of the heating furnace in the chemical industry based on the convolutional neural network algorithm. The technical scheme provided by the embodiment obtains a stable network structure, excellent generalization performance and higher precision, so that the heating furnace always keeps stable operation at a higher heat efficiency level.
TABLE 1 input-output variables
The present embodiment models the heating furnace operation variable — the air valve opening. Specifically, in this embodiment, the proposed memory-enhanced optimization-based convolutional neural network method is used to train a network model, and a network prediction value can be obtained by inputting related variable parameters, that is, a measurement value of the opening of the air valve can be obtained.
The technical scheme provided by the embodiment is different from the traditional convolutional neural network model, and the prior calculation experience is retained aiming at the repeatability and similarity characteristics of the weight value updating problem in the convolutional neural network, and the weight value parameter which enables the output error to be minimum is found out from the historical calculation experience by using a self-adaptive method, so that the convolutional neural network model is optimized, and the precision and the stability of the convolutional neural network model are improved. The technical scheme provided by the embodiment is easy to use, has obvious effect, can improve the precision of model training, and can be widely applied to modeling of the heating furnace.
Fig. 2 is a schematic flow chart of a measurement model of an operating variable of a heating furnace according to an embodiment of the present invention. As shown in fig. 2, the present embodiment provides a memory-enhanced optimization-based CNN furnace operation variable soft measurement method, including: obtaining training samples (X, Y); -normalizing said training samples (X, Y) using a normalization formula; initializing each parameter value of the convolutional neural network, wherein the parameter value comprises the number of network layers, the number of nodes and the numerical value of a convolutional kernel; updating the weight of the convolution layer of the convolution neural network; updating the weight of the downsampling layer of the convolutional neural network; optimizing the weight of the convolutional neural network by using a memory enhancement optimization algorithm to form a convolutional neural network model; and predicting the heating furnace operation variables by using the convolutional neural network model.
The embodiment obtains the training samples (X, Y) and carries out normalization processing on the training samples, and eliminates the influence of the dimension on the model. Wherein, the normalization process is shown as formula (1) and formula (2):
wherein,
ymin and Ymax are the minimum and maximum values of the output pattern vector Y, respectively.
In this embodiment, a convolutional neural network is initialized, and the number of network layers, the number of nodes, and the number of convolutional cores are set. Then, the present embodiment defines the error cost function as follows:
wherein N is the number of samples, c is the dimension of label,label t representing the nth samplenThe (c) th dimension of (a),the kth dimension of the nth sample net input. The purpose of updating the network weight is to make the network output y closer to the true value t, even if the error cost function E is minimized. When considering a single sample, the error cost function for the nth sample is:
the weight update of the convolutional neural network provided by this embodiment involves two parts: some are updating of convolutional layer weight, and some are updating of down-sampling layer, in this embodiment, the following derivation is firstly performed on the convolutional layer weight updating:
feature map x for each output of convolutional layerjIs provided with
Wherein M isjRepresenting selected input feature combinations, kijIs the convolution kernel used for the connection between the ith feature of the input and the jth feature of the output, bjIs the bias corresponding to the jth feature, and f is the activation function.
In this embodiment, convolutional layer sensitivity of the convolutional neural network is obtained according to a convolutional layer sensitivity formula, where the convolutional layer sensitivity formula is:
where "o" is the multiplication of the corresponding element, since l +1 layer is the sampling layer, it is equivalent to convolution. For example, down-sampling of scale 2 is to use each value of 2 x 2 asThe convolution kernel of (a) is convolved, so that the weight W is actually the 2 x 2 convolution kernel whose value is βj. Up represents the upsampling operation because the sensitivity matrix of the l +1 sampling layer is of the l-layer sensitivity matrix size(scale 2), the sensitivity matrix of the l +1 layer is up-sampled here so that they are of uniform size.
This embodiment sums all nodes of sensitivity of l layers, where (u, v) represents the element position in the sensitivity matrix, forming the following equation:
in this embodiment, the partial derivative of the error cost function on the convolution kernel k is calculated according to the sensitivity of the convolution layer:
wherein,is composed ofAnd k isijThe convolved patch has a position (u, v) that is the center of the patch, and the value of the position (u, v) in the output feature is determined by the value of the position (u, v) in the input feature and the convolution kernel kijAnd (4) performing convolution.
The weight value updating formula of the convolutional neural network downsampling layer is derived as follows:
firstly, calculating the output of l layers, and outputting each kind of output characteristic x of sampling layer in the convolutional neural networkjComprises the following steps:
where down denotes downsampling, β is multiplicative bias, and b is additive bias.
Obtaining the downsampling layer sensitivity of the convolutional neural network according to a downsampling layer sensitivity formula, wherein the downsampling layer sensitivity formula is as follows:
where "o" is the multiplication of the corresponding elements, and rot180(·) is the counterclockwise rotation of the matrix by 180 degrees. Finally, the present embodiment calculates the partial derivative of the error cost function to the bias b according to the sensitivity of the down-sampling layer. Thus, the embodiment obtains the weight updating formula of the convolutional neural network.
In the embodiment, the weight of the convolutional neural network is optimized by using a memory enhancement optimization algorithm, so that the establishment of a network model is completed. In the method for keeping the previous calculation experience, the weight parameter which enables the output error to be minimum is found out from the historical calculation experience by using an adaptive method.
When the error cost function of the convolutional neural network is minimum, obtaining the weight parameter of the convolutional neural network, wherein the error cost function of the convolutional neural network is as follows:
wherein,the value of the error cost function when the weight value is updated for the (k + 1) th time,and epsilon is a constant parameter, and is a value of the sensitivity of the (k + 1) th time.
The embodiment predicts the operation variables of the heating furnace by using the convolution neural network model. Specifically, the optimized convolutional neural network model is used for predicting the opening of an air valve as a heating furnace operation variable.
TABLE 2 comparison of optimized convolutional neural networks with conventional convolutional neural networks
As can be seen from table 2, the Average Relative Error (ARE) and the average Root Mean Square Error (RMSE) of the optimized convolutional neural network model provided in this embodiment ARE both smaller than those of the conventional convolutional neural network model, so that the integration method provided in this embodiment is more accurate.
Fig. 3 is a diagram illustrating a distribution of predicted results of a generalization process according to a first embodiment of the present invention, and fig. 4 is a diagram illustrating an error of a measurement model of manipulated variables of a heating furnace according to a first embodiment of the present invention. As shown in fig. 3 and 4, the experiment provided in this embodiment compares the optimized convolutional neural network method with the conventional convolutional neural network method, the back propagation method, and the like, which indicates that the memory-enhanced optimized convolutional neural network method provided in this embodiment can improve the accuracy of the model while ensuring the speed, and has good robustness, thereby improving the quality of the product, and enabling the heating furnace to always operate at a higher thermal efficiency.
The memory-enhanced optimization-based soft measurement method for the CNN heating furnace operating variables comprises the steps of firstly obtaining training sample data, preprocessing the training sample data, presetting the number of layers and the number of nodes of each layer of a convolutional neural network according to a convolutional neural network algorithm, initializing the weight and the offset of each layer, then optimizing the weight of the convolutional neural network by using a memory-enhanced optimization algorithm to obtain an optimal weight parameter, so that the modeling precision of the model is improved, finally obtaining the measured value of the heating furnace operating variables according to the convolutional neural network model formed by training, and enabling the thermal efficiency of the heating furnace to be always kept in an optimal state through the measured value of the operating variables, so that the industrial production benefit is improved. The technical scheme provided by the embodiment can obtain a stable network structure, excellent generalization performance and higher precision, thereby improving the quality of products and enabling the heating furnace to always keep stable operation at a higher heat efficiency level.
It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.

Claims (5)

1. A soft measurement method of CNN heating furnace operation variables based on memory enhancement optimization is characterized by comprising the following steps:
obtaining training samples (X, Y);
-normalizing said training samples (X, Y) using a normalization formula;
initializing each parameter value of the convolutional neural network, wherein the parameter value comprises the number of network layers, the number of nodes and the numerical value of a convolutional kernel;
updating the weight of the convolution layer of the convolution neural network;
updating the weight of the downsampling layer of the convolutional neural network;
optimizing the weight of the convolutional neural network by using a memory enhancement optimization algorithm to form a convolutional neural network model;
and predicting the heating furnace operation variables by using the convolutional neural network model.
2. The memory-enhanced optimization-based soft measurement method for CNN heating furnace operation variables according to claim 1, wherein the step of optimizing the weights of the convolutional neural network using a memory-enhanced optimization algorithm comprises:
when the error cost function of the convolutional neural network is minimum, obtaining the weight parameter of the convolutional neural network, wherein the error cost function of the convolutional neural network is as follows:
wherein,the value of the error cost function when the weight value is updated for the (k + 1) th time,and epsilon is a constant parameter, and is a value of the sensitivity of the (k + 1) th time.
3. The memory-enhanced-optimization-based soft measurement method for CNN heating furnace operating variables according to claim 1, wherein the step of updating the weights of the convolutional layers of the convolutional neural network comprises:
obtaining convolutional layer sensitivity of the convolutional neural network according to a convolutional layer sensitivity formula, wherein the convolutional layer sensitivity formula is as follows:
wherein,is multiplied by the corresponding element, W is the weight value, βjUp represents an upsampling operation as the value of the convolution kernel;
calculating the partial derivative of the error cost function to the convolution kernel k according to the sensitivity of the convolution layer:
wherein,is composed ofAnd k isijThe convolved patch has a position (u, v) that is the center of the patch, and the value of the position (u, v) in the output feature is determined by the value of the position (u, v) in the input feature and the convolution kernel kijAnd (4) performing convolution.
4. The memory-enhanced optimization-based soft measurement method for CNN furnace operation variables according to claim 1, wherein the step of updating the weights of the downsampling layer of the convolutional neural network comprises:
obtaining the downsampling layer sensitivity of the convolutional neural network according to a downsampling layer sensitivity formula, wherein the downsampling layer sensitivity formula is as follows:
wherein,rot180 (-) rotates the matrix counterclockwise 18 for the corresponding element multiplication0 degree;
and calculating the partial derivative of the error cost function to the bias b according to the sensitivity of the down-sampling layer.
5. The memory-enhancement-optimization-based soft measurement method for CNN heating furnace operating variables according to claim 1, characterized in that the normalization formula is:
wherein,
Yminand YmaxThe minimum and maximum values of the output pattern vector Y, respectively.
CN201810950550.8A 2018-08-20 2018-08-20 The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization Pending CN109188903A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810950550.8A CN109188903A (en) 2018-08-20 2018-08-20 The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810950550.8A CN109188903A (en) 2018-08-20 2018-08-20 The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization

Publications (1)

Publication Number Publication Date
CN109188903A true CN109188903A (en) 2019-01-11

Family

ID=64918601

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810950550.8A Pending CN109188903A (en) 2018-08-20 2018-08-20 The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization

Country Status (1)

Country Link
CN (1) CN109188903A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110488888A (en) * 2019-07-03 2019-11-22 太原理工大学 A kind of resistance-heated furnace temprature control method based on adaptive iteration study
CN112215351A (en) * 2020-09-21 2021-01-12 浙江大学 Enhanced multi-scale convolution neural network soft measurement method
CN113239565A (en) * 2021-05-27 2021-08-10 中南大学 Soft measurement method and device for product quality of fluidized bed roaster
CN116088307A (en) * 2022-12-28 2023-05-09 中南大学 Multi-working-condition industrial process prediction control method, device, equipment and medium based on error triggering self-adaptive sparse identification

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102880905A (en) * 2011-07-15 2013-01-16 中国石油天然气股份有限公司 Online soft measurement method for normal oil dry point
US8468108B1 (en) * 2009-06-30 2013-06-18 Zytek Communications Corporation Modeling efficiency over a range of velocities in underwater vehicles
CN104376380A (en) * 2014-11-17 2015-02-25 北京工业大学 Ammonia concentration predicting method based on recursion self-organization neural network
CN106124373A (en) * 2016-06-17 2016-11-16 中国大唐集团科学技术研究院有限公司华东分公司 A kind of measuring method of coal powder density

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8468108B1 (en) * 2009-06-30 2013-06-18 Zytek Communications Corporation Modeling efficiency over a range of velocities in underwater vehicles
CN102880905A (en) * 2011-07-15 2013-01-16 中国石油天然气股份有限公司 Online soft measurement method for normal oil dry point
CN104376380A (en) * 2014-11-17 2015-02-25 北京工业大学 Ammonia concentration predicting method based on recursion self-organization neural network
CN106124373A (en) * 2016-06-17 2016-11-16 中国大唐集团科学技术研究院有限公司华东分公司 A kind of measuring method of coal powder density

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
YONGJIAN WANG, HONGGUANG LI: "A novel intelligent modeling framework integrating convolutional neural network with an adaptive time-series window and its application to industrial process operational optimization", 《CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS》 *
YONGJIAN WANG, JINGWEN HUANG, CHONG SU, HONGGUANG LI: "Furnace thermal efficiency modeling using an improved convolution neural network based on parameter-adaptive mnemonic enhancement optimization", 《APPLIED THERMAL ENGINEERING》 *
张正江: "一种基于经验增强的实时优化方法MEO", 《化工学报》 *
方学毅: "过程系统记忆增强型实时优化方法", 《中国博士学位论文全文数据库 基础科学辑》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110488888A (en) * 2019-07-03 2019-11-22 太原理工大学 A kind of resistance-heated furnace temprature control method based on adaptive iteration study
CN110488888B (en) * 2019-07-03 2020-11-27 太原理工大学 Resistance heating furnace temperature control method based on adaptive iterative learning
CN112215351A (en) * 2020-09-21 2021-01-12 浙江大学 Enhanced multi-scale convolution neural network soft measurement method
CN113239565A (en) * 2021-05-27 2021-08-10 中南大学 Soft measurement method and device for product quality of fluidized bed roaster
CN113239565B (en) * 2021-05-27 2022-07-12 中南大学 Soft measurement method and device for product quality of fluidized bed roaster
CN116088307A (en) * 2022-12-28 2023-05-09 中南大学 Multi-working-condition industrial process prediction control method, device, equipment and medium based on error triggering self-adaptive sparse identification
CN116088307B (en) * 2022-12-28 2024-01-30 中南大学 Multi-working-condition industrial process prediction control method, device, equipment and medium based on error triggering self-adaptive sparse identification

Similar Documents

Publication Publication Date Title
CN109188903A (en) The flexible measurement method of CNN furnace operation variable based on memory-enhancing effect optimization
CN111353631A (en) Thermal power plant condenser vacuum degree prediction method based on multilayer LSTM
CN111768000A (en) Industrial process data modeling method for online adaptive fine-tuning deep learning
CN108537366B (en) Reservoir scheduling method based on optimal convolution bidimensionalization
CN110018675A (en) Nonlinear system modeling method based on LWDNN-ARX model
CN105512832A (en) Urban water demand combined predication method based on time-varying weight minimum variance
CN110097929A (en) A kind of blast furnace molten iron silicon content on-line prediction method
CN114298134A (en) Wind power prediction method and device and electronic equipment
CN107894710B (en) Principal component analysis modeling method for temperature of cracking reaction furnace
CN113128666A (en) Mo-S-LSTMs model-based time series multi-step prediction method
CN110245398B (en) Soft measurement deep learning method for thermal deformation of air preheater rotor
CN114444790B (en) Method for predicting time sequence of various measuring points on gas turbine based on steady-state characteristic composition
CN105676645A (en) Double-loop water tank liquid level prediction control method based on function type weight RBF-ARX model
Galvão et al. Linear-wavelet networks
CN110909492B (en) Sewage treatment process soft measurement method based on extreme gradient lifting algorithm
CN116861256A (en) Furnace temperature prediction method, system, equipment and medium for solid waste incineration process
CN116292057A (en) Control method, device, equipment and medium for variable-speed pumped storage unit
CN111210877A (en) Method and device for deducing physical property parameters
CN116596396A (en) Industrial polyethylene process quality prediction method based on K nearest neighbor interpolation and SLSTM
CN111222708A (en) Power plant combustion furnace temperature prediction method based on transfer learning dynamic modeling
CN114186771B (en) Mixed regularized random configuration network industrial process operation index estimation method
CN113779724B (en) Intelligent fault prediction method and system for filling packaging machine
CN112416913B (en) GWO-BP algorithm-based aircraft fuel system state missing value supplementing method
CN115018162A (en) Method and system for predicting machining quality in industrial finish machining process in real time
CN111144052B (en) CNN-ARX model-based linear primary inverted pendulum system modeling method and model

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20190111

RJ01 Rejection of invention patent application after publication