CN110262582B - High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network - Google Patents

High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network Download PDF

Info

Publication number
CN110262582B
CN110262582B CN201910696867.8A CN201910696867A CN110262582B CN 110262582 B CN110262582 B CN 110262582B CN 201910696867 A CN201910696867 A CN 201910696867A CN 110262582 B CN110262582 B CN 110262582B
Authority
CN
China
Prior art keywords
temperature
temperature control
dye liquor
neural network
formula
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.)
Active
Application number
CN201910696867.8A
Other languages
Chinese (zh)
Other versions
CN110262582A (en
Inventor
魏苗苗
刘洲峰
李春雷
张爱华
朱永胜
李碧草
杨艳
徐庆伟
林漫漫
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.)
Zhongyuan University of Technology
Original Assignee
Zhongyuan University of 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 Zhongyuan University of Technology filed Critical Zhongyuan University of Technology
Priority to CN201910696867.8A priority Critical patent/CN110262582B/en
Publication of CN110262582A publication Critical patent/CN110262582A/en
Application granted granted Critical
Publication of CN110262582B publication Critical patent/CN110262582B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • D—TEXTILES; PAPER
    • D06—TREATMENT OF TEXTILES OR THE LIKE; LAUNDERING; FLEXIBLE MATERIALS NOT OTHERWISE PROVIDED FOR
    • D06B—TREATING TEXTILE MATERIALS USING LIQUIDS, GASES OR VAPOURS
    • D06B23/00—Component parts, details, or accessories of apparatus or machines, specially adapted for the treating of textile materials, not restricted to a particular kind of apparatus, provided for in groups D06B1/00 - D06B21/00
    • D06B23/20—Arrangements of apparatus for treating processing-liquids, -gases or -vapours, e.g. purification, filtration or distillation
    • D06B23/22—Arrangements of apparatus for treating processing-liquids, -gases or -vapours, e.g. purification, filtration or distillation for heating
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B11/00—Automatic controllers
    • G05B11/01—Automatic controllers electric
    • G05B11/36—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential
    • G05B11/42—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential for obtaining a characteristic which is both proportional and time-dependent, e.g. P. I., P. I. D.
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D23/00—Control of temperature
    • G05D23/19—Control of temperature characterised by the use of electric means
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Textile Engineering (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention discloses a high-temperature and high-pressure jig dyeing machine temperature control method based on an improved RBF neural network, which comprises the following steps: s1, establishing a dye liquor temperature change curve model; s2, calculating temperature control deviation and setting a threshold value according to the actual measurement value of the dye liquor temperature and the dye liquor temperature in the dye liquor temperature change curve model corresponding to the sampling time; and S3, respectively adopting a PD controller, a PID controller and/or a PID controller based on an improved RBF neural network to control the dye liquor temperature in sections according to the relation between the temperature control deviation and the set threshold value until the numerical value of the temperature control deviation is zero. The invention realizes the self-adaptive control and adjustment of the jig dyeing machine temperature control system under the severe application environment by combining the improved RBF neural network-based feature learning method and the PID controller, ensures the high precision and the high efficiency of the control process, and has short control and adjustment period and high control efficiency.

Description

High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network
Technical Field
The invention relates to the field of automatic control of jig dyeing processes in the textile industry, in particular to a high-temperature and high-pressure jig dyeing machine temperature control method based on an improved RBF neural network.
Background
The jig dyeing link is a necessary link in the textile production process and is vital to the control and management of the textile quality. In the jig dyeing process, the accurate control of the dye liquor temperature is ensured to directly influence the jig dyeing quality of the fabric, and the method is the core work in the design of a jig dyeing machine control system. With the rapid development of automatic control technology and the continuous and deep research of control algorithms, the control algorithms based on the neural network are widely applied in the field of industrial control, and the control algorithms based on the neural network replace the traditional control algorithms, so that not only can the parameter control precision be improved, but also the feedback regulation speed can be improved, the overall control efficiency of a control system is improved, the phenomenon of uneven fabric dyeing is reduced, the fabric quality is improved, and the development of textile technology and export of textiles in China are facilitated.
The existing algorithm used in the control field of the jig dyeing machine mainly takes a PID control algorithm as a main part, the control algorithm comprises 3 types of control components of proportion, integral and differential, and a PID controller controls a controlled object by taking the linear combination of the proportion, the integral and the differential of the difference value of a preset value and an actual output value as a comprehensive control quantity. [ reference [1] Mahmood Q A, Nawaf AT, Esmael M N, et al. PID Temperature Control of refined Water Tank [ J ]. IOP conference series Materials Science and Engineering,2018,454:012031 ] during the use of PID Control systems, the parameter determination of PID Control algorithms is a difficult point. In practical application, the parameters are influenced by specific working environments, so that the traditional PID controller has the problems of rough parameters, poor adaptability, poor control effect and the like, and is difficult to adapt to complex working conditions. The conventional PID controller is adopted, and the problems of poor anti-interference capability of a system, poor system adaptability to different mathematical models, large overshoot and the like exist [ reference [2] El-Samahy A, Shamseldin MA.Brush DC motor tracking control using self-tuning fuzzy-tuning PID control and model feedback adaptive control [ J ]. Ain Shams Engineering Journal,2016: S209044791630003X ].
Aiming at the problem, fuzzy control and PID control are combined, a fuzzy self-adaptive PID control algorithm is provided, parameters of PID are adjusted on line by using a fuzzy rule, and then the temperature is adjusted by a controller. The effects of fast response speed, small overshoot and strong adaptability to systems of different mathematical models are achieved [ reference [3]: Du Wen, Ding. (2010) [ the resistance-heated Fuel Temperature Control Based on CMAC-Fuzzy image Control. applied Mechanics and materials.29-32.10.4028/www.scientific.net/AMM.29-32.407 ]. Although the algorithm adjusts the calculation link of the control parameters and improves the control efficiency, the algorithm has low control precision, and still continues to use a PID multiple feedback regulation mechanism, thereby failing to radically shorten the control period.
The algorithm proposed at present achieves the goal of adaptive control to a certain extent, but still has many problems which are not solved, and further research is urgently needed: 1) with the development of textile technology, the variety of fabrics is various, the fabric material is various, and the dyeing process is different (for example: cotton, silk, rayon, etc.), so a dye liquor temperature control model should be established for different fabrics respectively to realize the fine control of the jig dyeing process; 2) with the continuous development of textile technology, the competition of the textile industry is fierce, how to further improve the control reaction speed, reduce the fabric loss in the jig dyeing process and improve the production efficiency of textiles is also an industrial problem which needs to be solved urgently. 3) The existing control algorithm needs a plurality of feedback adjustment processes, the control adjustment period is too long, the control efficiency is not high, the requirement on temperature control is gradually improved along with the development of the textile technology, and the control precision of the existing control algorithm is limited and cannot meet the higher production requirement.
In recent years, a feature learning algorithm based on a neural network is rapidly developed and widely applied to the fields of intelligent information, intelligent medical treatment, intelligent transportation, industrial production and the like. The algorithm extracts the change characteristics of the control parameters through a large amount of collected prior data, modifies the network weight, trains a network model, and is used as a control parameter core algorithm to realize the real-time and rapid adjustment of the control parameters. Even if the fabric types are various and the control process is complex, the control algorithm based on the neural network can switch different weight models according to the actual situation, so that the adaptability adjustment of different application situations is realized, and the self-adaptive control characteristic is good. [ reference document [3]]:Liu Y.,Wang J.,Yang J.,Li Q.(2018)An Improved PID AlgorithmBased on BP Neural Network of Ambient Temperature Controller.In:Qiao F.,Patnaik S.,Wang J.(eds)Recent Developments in Mechatronics and IntelligentRobotics.ICMIR 2017.Advances in Intelligent Systems and Computing,vol690.Springer,Cham]And [ reference [4 ]]:Flavio
Figure BDA0002149610920000021
Sanchez E N,Xia Y,et al.Real-time neural inverse optimal control for indoor air temperature and humidityin a direct expansion(DX)air conditioning(A/C)system[J].International Journalof Refrigeration,2017,79.]The control algorithm based on the neural network is applied to the fields of greenhouse temperature control and automobile manufacturing, and a good control effect is achieved. However, the neural network control technology for dye liquor temperature parameters in the jig dyeing process under the high-temperature and high-pressure environment in the textile industry needs to be deeply researched.
Disclosure of Invention
Aiming at the technical problem that the control precision of the existing control method cannot meet the production requirement, the invention improves the network structure on the Basis of the jig dyeing machine temperature control technology based on the BP neural network, and uses the improved Radial Basis Function (RBF) to replace the BP network, thereby providing the high-temperature and high-pressure jig dyeing machine temperature control method based on the improved RBF neural network. The temperature control deviation value is divided according to the empirical threshold value, the segmented control is carried out by combining two algorithms of an improved RBF neural network and an integral separation PID controller, the improved RBF neural network is superior to a BP neural network in learning capacity and learning rate, and the improved RBF neural network has good approximation effect on linear and nonlinear change curves, so that the technology can realize high-efficiency and high-precision control on the dye liquor temperature in the jig dyeing process while considering the calculation complexity.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
a high-temperature high-pressure jig dyeing machine temperature control method based on an improved RBF neural network comprises the following steps:
s1: establishing a dye liquor temperature change curve model;
analyzing the heat exchange process of the dye liquor according to an energy conversion law, considering the influence of factors such as heat dissipation, convection radiation and the like, and establishing a dye liquor temperature change curve model;
s2: calculating temperature control deviation and setting a threshold value;
s2.1: calculating temperature control deviation | e (n) | according to the actual measurement value of the dye liquor temperature and the dye liquor temperature in the dye liquor temperature change curve model corresponding to the sampling moment;
s2.2: setting the threshold value according to specific application environment and dyeing experience1And2and is and1＜2；
s3: carrying out sectional control on the actual dye liquor temperature according to the size of the temperature control deviation | e (n) |;
the temperature control deviation | e (n) | and the threshold value1And2by comparison, when | e (n) | >2Controlling the temperature of the dye solution by adopting a PD controller; when in use1＜|e(n)|≤2Controlling the temperature of the dye liquor by adopting a PID controller; when 0 < | e (n) | ≦1Controlling the dye liquor temperature by adopting a PID controller based on an improved RBF neural network;
s4: and (4) calculating a temperature control deviation according to the dye liquor temperature control output value of the PD controller, the PID controller or the PID controller based on the improved RBF neural network obtained in the step (S3) and the actual dye liquor temperature, updating the temperature control deviation | e (n) |, and circulating the steps (S3-S4) until | e (n) | 0.
Preferably, in step S1, a temperature control curve is preset, and then a dye liquor temperature change curve model is established according to the temperature control curve; the temperature control curve includes at least one temperature raising section, at least one temperature lowering section, and at least one temperature retaining section.
Preferably, in step S1, the dye liquor temperature change curve model is:
Figure BDA0002149610920000031
wherein u represents the flow rate of steam in the dye vat, T represents the temperature of the dye liquor, and T represents the time;
solving the formula (1) to obtain:
Figure BDA0002149610920000032
in the formula, C1Indicating the initial temperature of the dye liquor.
Preferably, in step S3, the calculation formula of the PD controller is:
Figure BDA0002149610920000033
in the formula, kP,kDRespectively representing a proportional parameter and a differential parameter, TsThe sampling interval of the training sample is represented, e (n) represents the temperature control difference value of the training sample at the nth sampling moment, e (n-1) represents the temperature control difference value of the training sample at the n-1 th sampling moment, and u (n) represents the dye liquor temperature control output value of the PD controller at the nth sampling moment.
Preferably, in step S3, the calculation formula of the PID controller is:
Figure BDA0002149610920000041
in the formula, TsRepresenting the sampling interval, k, of the training samplesP,kI,kDThe control parameters of a feedback control loop adopting a PID controller are shown, e (n) shows the temperature control difference value of the training sample at the nth sampling moment, e (n-1) shows the temperature control difference value of the training sample at the n-1 th sampling moment, u (n) shows the dye liquor temperature control output value of the PID controller at the nth sampling moment, e (m) shows the temperature control difference value of the training sample at the mth sampling moment, m is 1,2, … n, and n shows the number of the sampling moments.
Preferably, in step S3, the control step of the PID controller based on the modified RBF neural network is as follows:
s3.1: calculating the control parameters of the PID controller according to the improved RBF neural network;
s3.1.1: selecting Gaussian function as the realization form of radial basis function, and hiding the output of layer
Figure BDA0002149610920000042
The calculation formula of (2) is as follows:
Figure BDA0002149610920000043
where i is 1,2, … M, M indicates the number of hidden layer neurons, j is 1,2, … N, N indicates the number of training samples of the input layer, ci(n) denotes the center of the ith neuron of the hidden layer selected at the nth time, ej(n) denotes the temperature control difference of the jth training sample at the nth sampling instant, dmRepresents the maximum distance between centers;
wherein the maximum distance dmThe variance σ is calculated by the following formula:
Figure BDA0002149610920000044
s3.1.2, using the minimum value reached by the target optimization function ξ (n) as the training target, using the gradient descent method to realize the network training, the network output y of the corresponding output layerkThe calculation formula of (n) is:
Figure BDA0002149610920000045
in the formula, wi(n) represents the weight from the hidden layer to the output layer of the ith neuron in the hidden layer at the nth time, k is 1,2, …, L, and L represents the number of neurons in the output layer;
s3.1.3: calculating the value of the network parameter by adopting a gradient descent method, wherein the weight w of the ith neuron of the hidden layer from the hidden layer to the output layeriThe calculation formula of (n) is:
Figure BDA0002149610920000046
in the formula (I), the compound is shown in the specification,
Figure BDA0002149610920000051
a Gaussian function representing the ith neuron of the hidden layer;
the update formula of equation (10) is:
Figure BDA0002149610920000052
in the formula, η denotes a learning speed, α denotes a momentum factor, and wi(n +1) represents the weight of the ith neuron in the hidden layer from the hidden layer to the output layer at the (n +1) th moment, wi(n-1) representing the weight of the ith neuron of the hidden layer from the hidden layer to the output layer at the (n-1) th moment;
s3.1.4: calculating the center value c of the radial basis functioni(n) the calculation formula is:
Figure BDA0002149610920000053
in the formula, σi(n) representing a variance value of a Gaussian function of an ith neuron of the hidden layer at an nth moment;
the update formula of equation (12) is:
Figure BDA0002149610920000054
in the formula, ci(n +1) denotes the center of the i-th neuron of the hidden layer selected at the time n +1, ci(n-1) represents the center selected by the ith neuron of the hidden layer at the nth-1 moment;
s3.1.5: calculating the variance value sigma of the Gaussian functioni(n) the calculation formula is:
Figure BDA0002149610920000055
the update formula of equation (14) is:
Figure BDA0002149610920000056
in the formula, σi(n +1) denotes the variance value of the Gaussian function of the ith neuron in the hidden layer at the (n +1) th time, sigmai(n-1) representing a Gaussian function variance value of an ith neuron of the hidden layer at an nth-1 moment;
s3.1.6 according to the target optimization function ξ (n) and the center value ci(n) variance value σi(n) and weight wi(n), the computational network output is:
Figure BDA0002149610920000061
in the formula, KP,KI,KDPID control parameters generated based on the improved RBF neural network are adopted;
s3.2: and regulating the dye liquor temperature by adopting a PID controller according to the PID control parameters generated by the RBF neural network.
Preferably, in step S3.2, the calculation formula for the PID controller to adjust the dye liquor temperature is:
Figure BDA0002149610920000062
in the formula, TsRepresenting the sampling interval of the training samples, KP,KI,KDThe PID control parameters generated by the improved RBF neural network are shown, e (n) shows the temperature control difference value of the training sample at the nth sampling moment, e (n-1) shows the temperature control difference value of the training sample at the n-1 th sampling moment, u (n) shows the dye liquor temperature control output value based on the RBF neural network and adopting a PID controller at the nth sampling moment, e (m) shows the temperature control difference value of the training sample at the mth sampling moment, m is 1,2, … n, and n shows the number of the sampling moments.
Preferably, the objective optimization function ξ (n) is calculated as:
Figure BDA0002149610920000063
where j is 1,2, … N, N denotes the number of training samples of the input layer, and ej(n) represents the temperature control difference for the jth training sample at the nth time instant.
Preferably, in step S3, the formula for calculating the dye liquor temperature control output value u (n) using the PD controller, the PID controller and the PID controller based on the modified RBF neural network is:
Figure BDA0002149610920000064
in the formula, kP,kI,kDIndicating a control parameter, T, using a PID feedback control loopsRepresenting the sampling interval of the training sample, e (n) representing the temperature control difference value of the training sample at the nth sampling moment, e (n-1) representing the temperature control difference value of the training sample at the n-1 th sampling moment, u (n) representing the dye liquor temperature control output value of the controller at the nth sampling moment, KP,KI,KDThe PID control parameters generated by the improved RBF neural network are shown, e (m) shows the temperature control difference value of the training sample at the mth sampling moment, m is 1,2, … n, and n shows the number of the sampling moments.
The invention has the beneficial effects that:
the invention relates to a dye liquor temperature control method in a high-temperature high-pressure jig dyeing machine, which utilizes a characteristic learning method and a PID control algorithm based on an improved RBF neural network to realize self-adaptive control and adjustment of a jig dyeing machine temperature control system in a severe application environment; the invention establishes temperature control curves for different fabrics, establishes corresponding dye liquor temperature change curve models according to the dye liquor temperature change rule, collects a large amount of prior data to realize the pre-correction of weight coefficients among layers of an improved RBF neural network, obtains network models under different application scenes, switches the network models in real time according to a dye jigger material and a specific application scene to effectively realize the high-precision control of the dye liquor temperature change in the whole dye jigger process under a high-temperature and high-pressure environment, adopts the concepts of sectional control and simulation modeling, not only ensures the high precision and high efficiency of the control process, but also has short control regulation period and high control efficiency.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic view of the working process of the present invention.
Fig. 2 is a structural diagram of an improved RBF neural network of the present invention.
FIG. 3 is a comparison graph of response curves of control algorithms based on different neural networks according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
A temperature control method of a high-temperature and high-pressure jig dyeing machine based on an improved RBF neural network is shown in figure 1 and comprises the following steps:
s1: and establishing a dye liquor temperature change curve model.
S1.1: presetting a temperature control curve, and segmenting the temperature control curve according to the type of the fabric, wherein the temperature control curve comprises at least one temperature rising section, at least one heat preservation section and at least one temperature reduction section, and the time length and the temperature change speed of each section are different.
S1.2: analyzing the dye liquor heat exchange process according to an energy conversion law, considering the influence of factors such as heat dissipation, convection radiation and the like, and calculating the dye liquor temperature, wherein the calculation formula of the dye liquor temperature T is as follows:
Figure BDA0002149610920000081
wherein u represents the flow rate of steam in the cylinder, T represents the temperature of the dye liquor, and T represents time;
solving the formula (1) to obtain:
Figure BDA0002149610920000082
in the formula, C1Denotes the initial temperature of the dye liquor, in this example C1≈200887069.2857；
Under the condition of knowing the steam flow rate and the initial temperature of the dye liquor in the vat, the corresponding dye liquor temperature can be obtained, and a dye liquor temperature change curve model is established by combining a preset temperature control curve.
S1.3: and establishing a dye liquor temperature change curve model by combining a preset temperature control curve.
S2: the temperature control deviation | e (n) | is calculated and a threshold value is set.
S2.1: the calculation formula of the temperature control deviation | e (n) | is as follows:
|e(n)|＝|r(n)-T|； (3)
wherein r (n) represents the actual measurement value of the dye liquor temperature at the nth sampling moment, and T represents the dye liquor temperature value at the same sampling moment as r (n) in the dye liquor temperature change curve model.
S2.2: setting a threshold value according to specific application environment and past dyeing experience1And2and is and1＜2。
s3: and (3) combining a dye liquor temperature change curve model, and performing segmented control on the actual dye liquor temperature r (n) according to the temperature control deviation | e (n) | in the temperature rising section, the temperature preservation section and the temperature reduction section respectively.
S3.1: the temperature control deviation | e (n) | and the threshold value1And2comparing; when the temperature control deviation is | e (n) | >2Execute byStep S4, controlling the dye liquor temperature by adopting a PD controller to ensure the control speed; when temperature control deviation1＜|e(n)|≤2Step S5 is executed to control the dye liquor temperature by adopting a PID controller so as to adjust the control precision; when the temperature control deviation is 0 < | e (n) | ≦1Step S6 is executed to fine tune the dye liquor temperature by using a PID controller based on the modified RBF neural network, so as to further improve the control accuracy.
S3.2: and (3) calculating a temperature control deviation | e (n) | ═ r (n) < u > (n) | according to a dye liquor temperature control output value u (n) obtained by a PD controller, a PID controller or a PID controller based on an improved RBF neural network and the actual dye liquor temperature r (n), updating the temperature control deviation | e (n) |, and circularly executing the step S3.1 until | e (n) | is 0, wherein the actual dye liquor temperature r (n) is consistent with the dye liquor temperature at the corresponding sampling time in the dye liquor temperature change curve model.
S4: controlling the temperature of the dye solution by adopting a PD controller;
executing a PD controller to adjust the dye liquor temperature, wherein the input is e (n), and the calculation formula of the PD controller is as follows:
Figure BDA0002149610920000091
in the formula, kP,kDRespectively representing a proportional parameter and a differential parameter, TsThe sampling time period of the training sample is represented, e (n) represents the temperature control difference value of the training sample at the nth sampling moment, e (n-1) represents the temperature control difference value of the training sample at the n-1 th sampling moment, and u (n) represents the dye liquor temperature control output value of the PD controller at the nth sampling moment.
S5: a PID controller is adopted to carry out coarse adjustment on the temperature of the dye liquor;
and (3) executing a PID controller to adjust the dye liquor temperature, wherein the input is e (n), and the calculation formula of the PID controller is as follows:
Figure BDA0002149610920000092
in the formula, TsWhich represents the sampling interval of the training samples,e (n) represents the temperature control difference of the training sample at the nth sampling time, e (n-1) represents the temperature control difference of the training sample at the nth-1 sampling time, m is 1,2, … n, n represents the number of the sampling time, u (n) represents the dye liquor temperature control output value of the PID controller at the nth sampling time, e (m) represents the temperature control difference of the training sample at the mth sampling time, kP,kI,kDThe three control parameters are preset according to prior experience, namely fixed values.
S6: calculating control parameters of a PID controller based on an improved RBF neural network, and finely adjusting the dye liquor temperature by combining the PID controller;
s6.1: calculating control parameters of the PID controller based on the improved RBF neural network;
as shown in FIG. 2, the RBF neural network comprises an input layer, a hidden layer and an output layer, wherein the input of the input layer is e (n), and the output of the hidden layer is
Figure BDA0002149610920000093
The output of the output layer is yk(n)；
In the formula, ci(n) denotes the central point selected by the ith neuron of the hidden layer, ej(n) represents the temperature control difference of the jth training sample at the nth time instant;
s6.1.1 selecting Gaussian function as the implementation form of radial basis function, hiding the output of layer
Figure BDA0002149610920000094
The calculation formula of (2) is as follows:
Figure BDA0002149610920000095
where i is 1,2, … M, M indicates the number of hidden layer neurons, j is 1,2, … N, N indicates the number of training samples of the input layer, and d indicates the number of hidden layer neuronsmRepresents the maximum distance between centers;
wherein the maximum distance dmThe number M being related to the variance σ ofThe calculation formula is as follows:
Figure BDA0002149610920000101
s6.1.2 training the network by using the minimum value reached by the target optimization function and gradient descent method, and outputting the network output y of the corresponding output layerkThe calculation formula of (n) is:
Figure BDA0002149610920000102
where k is 1,2, …, L indicates the number of neurons in the output layer, and w indicates the number of neurons in the output layeri(n) represents the weight of the ith neuron in the hidden layer from the hidden layer to the output layer;
s6.1.3, selecting an objective optimization function, wherein the objective optimization function xi (n) is calculated by the formula:
Figure BDA0002149610920000103
s6.1.4 calculating the value of the network parameter by gradient descent method, and the weight w of the ith neuron in the hidden layer from the hidden layer to the output layeriThe calculation formula of (n) is:
Figure BDA0002149610920000104
in the formula (I), the compound is shown in the specification,
Figure BDA0002149610920000105
a Gaussian function representing the ith neuron of the hidden layer;
the update formula of equation (10) is:
Figure BDA0002149610920000106
in the formula, η denotes a learning speed, α denotes a momentum factor, and wi(n +1) represents the weight from the hidden layer to the output layer of the ith neuron in the hidden layer at the time of (n +1), wi(n-1) represents a concealed regionThe weight from the hidden layer to the output layer of the ith neuron at the (n-1) th moment;
s6.1.5 calculating the center value c of the radial basis functioni(n) the calculation formula is:
Figure BDA0002149610920000107
in the formula, σi(n) representing a variance value of a Gaussian function of an ith neuron of the hidden layer at an nth moment;
the update formula of equation (12) is:
Figure BDA0002149610920000111
in the formula, ci(n +1) represents the central point selected by the ith neuron of the hidden layer at the (n +1) th moment, ci(n-1) represents the central point selected by the ith neuron of the hidden layer at the (n-1) th moment;
s6.1.6 calculating the variance value sigma of the Gaussian functioni(n) the calculation formula is:
Figure BDA0002149610920000112
the update formula of equation (14) is:
Figure BDA0002149610920000113
in the formula, σi(n +1) denotes the variance value of the Gaussian function of the ith neuron in the hidden layer at the (n +1) th time, sigmai(n-1) representing a Gaussian function variance value of an ith neuron of the hidden layer at an nth-1 moment;
s6.1.7 according to the target optimization function ξ (n), the center value ci(n) variance value σi(n) and weight wi(n), obtaining a network output of:
Figure BDA0002149610920000114
KP,KI,KDrepresenting the control parameters of a PID controller generated based on the improved RBF neural network;
s6.2: according to the control parameters of a PID controller generated by the improved RBF neural network, the dye liquor temperature is adjusted by adopting the PID controller, and the calculation formula of the PID controller is as follows:
Figure BDA0002149610920000115
in the formula, TsRepresents the sampling interval of the training sample, m is 1,2, … n, n represents the number of the sampling time, e (n) represents the temperature control difference value of the training sample at the nth sampling time, e (n-1) represents the temperature control difference value of the training sample at the n-1 th sampling time, u (n) represents the dye liquor temperature control output value of the PID controller at the nth sampling time, e (m) represents the temperature control difference value of the training sample at the mth sampling time, KP,KI,KDThe three control parameters are preset according to prior experience, namely fixed values.
As shown in figure 3, the invention is superior to other neural network structures in learning ability and learning rate, and the improved RBF network has good approximation effect on linear and nonlinear change curves, so that the technology can realize high-efficiency and high-precision control on the dye liquor temperature in the jig dyeing process while considering the calculation complexity.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (8)

1. A high-temperature high-pressure jig dyeing machine temperature control method based on an improved RBF neural network is characterized by comprising the following steps:
s1: establishing a dye liquor temperature change curve model;
establishing a dye liquor temperature change curve model according to an energy conversion law;
s2: calculating temperature control deviation and setting a threshold value;
s2.1: calculating temperature control deviation | e (n) | according to the actual measurement value of the dye liquor temperature and the dye liquor temperature in the dye liquor temperature change curve model corresponding to the sampling moment;
s2.2: setting a threshold value1And2and is and1＜2；
s3: carrying out sectional control on the actual dye liquor temperature according to the size of the temperature control deviation | e (n) |;
the temperature control deviation | e (n) | and the threshold value1And2by comparison, when | e (n) | >2Controlling the temperature of the dye solution by adopting a PD controller; when in use1＜|e(n)|≤2Controlling the temperature of the dye liquor by adopting a PID controller; when 0 < | e (n) | ≦1Controlling the dye liquor temperature by adopting a PID controller based on an improved RBF neural network;
the control steps of the PID controller based on the improved RBF neural network are as follows:
s3.1: calculating the control parameters of the PID controller according to the improved RBF neural network;
s3.1.1: selecting Gaussian function as the realization form of radial basis function, and hiding the output of layer
Figure FDA0002603103860000011
The calculation formula of (2) is as follows:
Figure FDA0002603103860000012
where i is 1,2, … M, M indicates the number of hidden layer neurons, j is 1,2, … N, N indicates the number of training samples of the input layer, ci(n) denotes the center of the ith neuron of the hidden layer selected at the nth time, ej(n) denotes the temperature control difference of the jth training sample at the nth sampling instant, dmRepresents the maximum distance between centers;
wherein the maximum distance dmAnd the number M and the varianceThe calculation formula of the variance sigma is as follows:
Figure FDA0002603103860000013
s3.1.2, using the minimum value reached by the target optimization function ξ (n) as the training target, using the gradient descent method to realize the network training, the network output y of the corresponding output layerkThe calculation formula of (n) is:
Figure FDA0002603103860000014
in the formula, wi(n) represents the weight from the hidden layer to the output layer of the ith neuron in the hidden layer at the nth time, k is 1,2, …, L, and L represents the number of neurons in the output layer;
s3.1.3: calculating the value of the network parameter by adopting a gradient descent method, wherein the weight w of the ith neuron of the hidden layer from the hidden layer to the output layeriThe calculation formula of (n) is:
Figure FDA0002603103860000021
in the formula (I), the compound is shown in the specification,
Figure FDA0002603103860000022
a Gaussian function representing the ith neuron of the hidden layer;
the update formula of equation (10) is:
Figure FDA0002603103860000023
in the formula, η denotes a learning speed, α denotes a momentum factor, and wi(n +1) represents the weight of the ith neuron in the hidden layer from the hidden layer to the output layer at the (n +1) th moment, wi(n-1) representing the weight of the ith neuron of the hidden layer from the hidden layer to the output layer at the (n-1) th moment;
s3.1.4: calculating the center value c of the radial basis functioni(n) the calculation formula is:
Figure FDA0002603103860000024
in the formula, σi(n) representing a variance value of a Gaussian function of an ith neuron of the hidden layer at an nth moment;
the update formula of equation (12) is:
Figure FDA0002603103860000025
in the formula, ci(n +1) denotes the center of the i-th neuron of the hidden layer selected at the time n +1, ci(n-1) represents the center selected by the ith neuron of the hidden layer at the nth-1 moment;
s3.1.5: calculating the variance value sigma of the Gaussian functioni(n) the calculation formula is:
Figure FDA0002603103860000026
the update formula of equation (14) is:
Figure FDA0002603103860000027
in the formula, σi(n +1) denotes the variance value of the Gaussian function of the ith neuron in the hidden layer at the (n +1) th time, sigmai(n-1) representing a Gaussian function variance value of an ith neuron of the hidden layer at an nth-1 moment;
s3.1.6 according to the target optimization function ξ (n) and the center value ci(n) variance value σi(n) and weight wi(n), the computational network output is:
Figure FDA0002603103860000031
in the formula, KP,KI,KDPID control parameters generated based on the improved RBF neural network are adopted;
s3.2: regulating the dye liquor temperature by adopting a PID controller according to PID control parameters generated by the RBF neural network;
s4: and (4) calculating a temperature control deviation according to the dye liquor temperature control output value of the PD controller, the PID controller or the PID controller based on the improved RBF neural network obtained in the step (S3) and the actual dye liquor temperature, updating the temperature control deviation | e (n) |, and circulating the steps (S3-S4) until | e (n) | 0.
2. The improved RBF neural network-based temperature control method for the high-temperature and high-pressure jig dyeing machine as claimed in claim 1, wherein in step S1, a temperature control curve is preset, and then a dye liquor temperature change curve model is established according to the temperature control curve; the temperature control curve includes at least one temperature raising section, at least one temperature lowering section, and at least one temperature retaining section.
3. The improved RBF neural network-based high-temperature high-pressure jig dyeing machine temperature control method as claimed in claim 1 or 2, wherein in step S1, the dye liquor temperature change curve model is:
Figure FDA0002603103860000032
wherein u represents the flow rate of steam in the dye vat, T represents the temperature of the dye liquor, and T represents the time;
solving the formula (1) to obtain:
Figure FDA0002603103860000033
in the formula, C1Indicating the initial temperature of the dye liquor.
4. The improved RBF neural network-based high-temperature high-pressure jig dyeing machine temperature control method as claimed in claim 1, wherein in step S3, the calculation formula of the PD controller is as follows:
Figure FDA0002603103860000034
in the formula, kP,kDRespectively representing a proportional parameter and a differential parameter, TsThe sampling interval of the training sample is represented, e (n) represents the temperature control difference value of the training sample at the nth sampling moment, e (n-1) represents the temperature control difference value of the training sample at the n-1 th sampling moment, and u (n) represents the dye liquor temperature control output value of the PD controller at the nth sampling moment.
5. The improved RBF neural network-based high-temperature high-pressure jig dyeing machine temperature control method according to claim 1, wherein in step S3, the calculation formula of the PID controller is as follows:
Figure FDA0002603103860000041
in the formula, TsRepresenting the sampling interval, k, of the training samplesP,kI,kDThe control parameters of a feedback control loop adopting a PID controller are shown, e (n) shows the temperature control difference value of the training sample at the nth sampling moment, e (n-1) shows the temperature control difference value of the training sample at the n-1 th sampling moment, u (n) shows the dye liquor temperature control output value of the PID controller at the nth sampling moment, e (m) shows the temperature control difference value of the training sample at the mth sampling moment, m is 1,2, … n, and n shows the number of the sampling moments.
6. The improved RBF neural network-based high-temperature high-pressure jig dyeing machine temperature control method as claimed in claim 1, wherein in step S3.2, the calculation formula of the PID controller for adjusting the dye liquor temperature is as follows:
Figure FDA0002603103860000042
in the formula, TsRepresenting the sampling interval of the training samples, KP,KI,KDIndicating the adoption of improved RBF godsAnd (e), (n) represents the temperature control difference of the training sample at the nth sampling moment, e (n-1) represents the temperature control difference of the training sample at the nth-1 sampling moment, m is 1,2, … n, n represents the number of the sampling moments, u (n) represents the dye liquor temperature control output value of the RBF neural network based on the nth sampling moment and adopting a PID controller, and e (m) represents the temperature control difference of the training sample at the mth sampling moment.
7. The improved RBF neural network-based high-temperature high-pressure jig dyeing machine temperature control method as claimed in claim 1, wherein the objective optimization function xi (n) is calculated by the formula:
Figure FDA0002603103860000043
where j is 1,2, … N, N denotes the number of training samples of the input layer, and ej(n) represents the temperature control difference for the jth training sample at the nth time instant.
8. The method for controlling the temperature of a high-temperature high-pressure jig dyeing machine based on the modified RBF neural network as claimed in any one of claims 1 or 4 to 7, wherein in step S3, the calculation formula of the dye liquor temperature control output value u (n) of the PD controller, the PID controller and the PID controller based on the modified RBF neural network is as follows:
Figure FDA0002603103860000051
in the formula, kP,kI,kDIndicating a control parameter, T, using a PID feedback control loopsRepresenting the sampling interval of the training sample, e (n) representing the temperature control difference value of the training sample at the nth sampling moment, e (n-1) representing the temperature control difference value of the training sample at the n-1 th sampling moment, u (n) representing the dye liquor temperature control output value of the controller at the nth sampling moment, KP,KI,KDIndicating PI generated using modified RBF neural networkD control parameter, e (m) represents the temperature control difference of the training sample at the mth sampling time, m is 1,2, … n, n represents the number of sampling times.
CN201910696867.8A 2019-07-30 2019-07-30 High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network Active CN110262582B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910696867.8A CN110262582B (en) 2019-07-30 2019-07-30 High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910696867.8A CN110262582B (en) 2019-07-30 2019-07-30 High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network

Publications (2)

Publication Number Publication Date
CN110262582A CN110262582A (en) 2019-09-20
CN110262582B true CN110262582B (en) 2020-09-11

Family

ID=67912394

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910696867.8A Active CN110262582B (en) 2019-07-30 2019-07-30 High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network

Country Status (1)

Country Link
CN (1) CN110262582B (en)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110554715A (en) * 2019-10-25 2019-12-10 攀钢集团攀枝花钢铁研究院有限公司 RBF neural network-based PID control method for hydrolysis process temperature of titanyl sulfate plus seed crystal
CN111850892B (en) * 2020-07-10 2023-08-11 绍兴数智科技有限公司 Method and device for realizing abnormal alarm in dyeing process of overflow dyeing machine
CN111719322A (en) * 2020-07-16 2020-09-29 浙江流庆纺织有限公司 One-bath dyeing method of polyester-nylon and cationic polyester-nylon composite fiber terry knitted fabric
CN112363556A (en) * 2020-11-10 2021-02-12 贵州华亿绿色纺织产业科技有限公司 Temperature and smoke humidity control system
CN112553817B (en) * 2020-12-11 2022-11-15 河南省纺织产品质量监督检验院 Method and system for real-time control and adjustment of textile dyeing process based on artificial intelligence
CN112680918B (en) * 2020-12-30 2021-12-14 江苏华一机械有限公司 Sanding treatment method and system based on combined carbon fiber sanding machine
CN114137823B (en) * 2021-11-10 2023-12-26 中国科学院上海技术物理研究所 PID parameter setting method for multi-stage temperature control in low-temperature area of mechanical refrigerator
CN114870761A (en) * 2022-04-28 2022-08-09 中国计量大学 Intelligent safety temperature control method applied to calorimetric reaction kettle
CN115328232B (en) * 2022-09-01 2023-12-19 厦门宇昊软件有限公司 Temperature control method and system
CN116219665A (en) * 2023-02-09 2023-06-06 青岛富润达纺织有限公司 Dye liquor process temperature following control method based on temperature speed adjustment
CN116643497B (en) * 2023-05-29 2024-05-10 汕头市鼎泰丰实业有限公司 Dyeing control system and method for cheese

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102393770A (en) * 2011-11-25 2012-03-28 任洪娥 Temperature and humidity control method during a wood drying process based on combination of radial basis function (RBF) nerve network and proportional integral derivative (PID) closed loop control
CN102914623B (en) * 2012-10-19 2014-12-10 南京信息工程大学 Fusing method of temperature compensation of humidity sensor
CN103611734B (en) * 2013-11-25 2015-11-18 华中科技大学 A kind of section cooling temperature-controlled process and system
CN108719516A (en) * 2018-05-22 2018-11-02 湖南师范大学 One kind being based on RBF neural tea machine processing control parameter intelligence setting method
CN108958020A (en) * 2018-05-29 2018-12-07 南京理工大学 A kind of temprature control method based on RBF neural dynamic prediction PID
CN109597449A (en) * 2019-01-30 2019-04-09 杭州庆睿科技有限公司 A kind of ultrasonic wave separating apparatus temprature control method neural network based

Also Published As

Publication number Publication date
CN110262582A (en) 2019-09-20

Similar Documents

Publication Publication Date Title
CN110262582B (en) High-temperature high-pressure jig dyeing machine temperature control method based on improved RBF neural network
CN110008575B (en) Multi-parameter predictive control algorithm for switching process medium multi-temperature target set values of circulating cooling water system
CN110888317A (en) PID controller parameter intelligent optimization method
CN111520878A (en) Air conditioner temperature control system based on RBF neural network and control method thereof
CN109696836A (en) Aircraft steering engine electrohydraulic servo system intelligent control method
CN113253779A (en) Heat pump temperature control system based on particle swarm fuzzy PID algorithm
WO2019205216A1 (en) Rbf neural network predictive control-based control system and control method for double-input double-output ball mill
CN111123871B (en) Prediction function control method for genetic algorithm optimization of chemical process
CN111880489B (en) Regression scheduling method for complex manufacturing system
CN113885324A (en) Intelligent building power utilization control method and system
Zheng et al. Dynamic rolling force prediction of reversible cold rolling mill based on bp neural network with improved pso
CN113050717A (en) Control method of temperature control system based on generalized predictive control
Zhmud et al. The two methods of reverse overshoot suppression in automation systems
CN116307440B (en) Workshop scheduling method based on reinforcement learning and multi-objective weight learning, device and application thereof
CN116430726A (en) Method for controlling turboset based on subtractive clustering and fuzzy neural network
CN108089442B (en) PI controller parameter self-tuning method based on prediction function control and fuzzy control
Wang et al. Model-based predictive control for spatially-distributed systems using dimensional reduction models
CN106647247B (en) A kind of control algolithm suitable for servo controller
Rospawan et al. Adaptive predictive PID control using recurrent fuzzy broad learning system for accurate setpoint tracking of digital nonlinear time-delay dynamic systems
CN112327628A (en) Improved self-adaptive generalized predictive control method for data-driven time-lag system
Zhuang et al. The internal mode fractional-order PID control based on neural network for the temperature of air-conditioned rooms
CN111273563B (en) Prediction control method based on AGC (automatic gain control) comprehensive index of heat supply unit
CN109039166B (en) Method for self-correcting speed loop PI-IP control parameter of permanent magnet synchronous linear servo system
CN112836348B (en) Strip-shaped mechanism regulating quantity optimization method based on genetic algorithm and penalty function method
CN111221245B (en) Setting machine fuzzy PID tension control parameter self-tuning method based on PSO

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