WO2020199231A1 - 面向盘磨系统粉体粒度的随机分布控制实验装置及方法 - Google Patents

面向盘磨系统粉体粒度的随机分布控制实验装置及方法 Download PDF

Info

Publication number
WO2020199231A1
WO2020199231A1 PCT/CN2019/081827 CN2019081827W WO2020199231A1 WO 2020199231 A1 WO2020199231 A1 WO 2020199231A1 CN 2019081827 W CN2019081827 W CN 2019081827W WO 2020199231 A1 WO2020199231 A1 WO 2020199231A1
Authority
WO
WIPO (PCT)
Prior art keywords
control
particle size
disc
sampling
mill
Prior art date
Application number
PCT/CN2019/081827
Other languages
English (en)
French (fr)
Inventor
周平
赵向志
李明杰
王宏
Original Assignee
东北大学
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 东北大学 filed Critical 东北大学
Priority to US16/483,706 priority Critical patent/US11860078B2/en
Publication of WO2020199231A1 publication Critical patent/WO2020199231A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B02—CRUSHING, PULVERISING, OR DISINTEGRATING; PREPARATORY TREATMENT OF GRAIN FOR MILLING
    • B02C—CRUSHING, PULVERISING, OR DISINTEGRATING IN GENERAL; MILLING GRAIN
    • B02C25/00—Control arrangements specially adapted for crushing or disintegrating
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B02—CRUSHING, PULVERISING, OR DISINTEGRATING; PREPARATORY TREATMENT OF GRAIN FOR MILLING
    • B02C—CRUSHING, PULVERISING, OR DISINTEGRATING IN GENERAL; MILLING GRAIN
    • B02C7/00—Crushing or disintegrating by disc mills
    • B02C7/02—Crushing or disintegrating by disc mills with coaxial discs
    • B02C7/06—Crushing or disintegrating by disc mills with coaxial discs with horizontal axis
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B02—CRUSHING, PULVERISING, OR DISINTEGRATING; PREPARATORY TREATMENT OF GRAIN FOR MILLING
    • B02C—CRUSHING, PULVERISING, OR DISINTEGRATING IN GENERAL; MILLING GRAIN
    • B02C7/00—Crushing or disintegrating by disc mills
    • B02C7/11—Details
    • B02C7/14—Adjusting, applying pressure to, or controlling distance between, discs
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N1/00—Sampling; Preparing specimens for investigation
    • G01N1/28—Preparing specimens for investigation including physical details of (bio-)chemical methods covered elsewhere, e.g. G01N33/50, C12Q
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • 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

Definitions

  • the invention belongs to the technical field of distribution control, and specifically relates to a random distribution control experimental device and method for powder particle size of a disc mill system.
  • the two-dimensional quality distribution is a key process indicator to measure the quality of paper. Therefore, the papermaking process can be regarded as a typical dynamic random distribution system, and the key requirement of control is to make the two-dimensional quality distribution of paper as far as possible to meet the expected given distribution shape .
  • molecular weight distribution is often regarded as a key quality indicator in product quality control and process optimization. Therefore, the main purpose of polymerization process control is to make the molecular weight distribution shape of the polymerized product conform to the given distribution shape.
  • the flame temperature field distribution is often used as an important indicator of the benefit of the combustion process, and the purpose of combustion process control is to make the shape of the flame temperature field distribution meet the given requirements by selecting the appropriate fuel input and process parameters.
  • the size and shape of the processed food particles conform to the desired distribution shape, so that the distribution shape of the crushed food particles meets the requirements of subsequent food processing procedures, thereby improving the control quality and production efficiency of the entire system.
  • the present invention provides a powder particle size-oriented disc mill system. Random distribution control experimental device and method.
  • the present invention provides a random distribution control experimental device and method for the particle size of the disc mill system.
  • the disc system is adopted, and the experimental device includes: spiral feeding Material control device, mill speed adjustment device, grinding disc gap servo control device, sampling device, laser particle size analyzer, control cabinet and host computer;
  • the disc mill system adopts a vertical double disc mill, including a movable grinding disc and a corresponding static grinding disc;
  • the screw feeding control device includes a screw feeder, a DC motor 1 and an electronic scale; the screw feeder is connected with the DC motor 1; the electronic scale is connected with a host computer for measuring the actual screw feeding amount;
  • the speed adjustment device of the mill includes a frequency converter and a three-phase asynchronous motor; the three-phase asynchronous motor is connected to the movable grinding disc of the disc grinding system. After the material enters the grinding zone, the movable grinding disc is driven by the three-phase asynchronous motor to rotate;
  • the device is installed in the control cabinet;
  • the grinding disc gap servo control device includes an AC servo motor, a reducer and a displacement sensor; the AC servo motor is connected to the reducer, the reducer is connected to the movable grinding disc of the disc grinding device, and the position sensor is installed in the disc grinding system to connect the drive of the movable grinding disc On the shaft and connected with the host computer;
  • the sampling device includes a DC motor 2, a sampling sleeve and a powder recovery box; the sampling sleeves are respectively connected to the powder outlet of the grinding disc system and the powder recovery box;
  • the control cabinet is equipped with a PLC; the PLC is respectively connected to the DC motor 1, the DC motor 2 of the sampling device, and the AC servo motor of the disc grinding gap of the spiral feeding control device; the frequency converter is connected to the rotation speed adjustment device of the mill Three-phase asynchronous motor connected;
  • Both the PLC and the frequency converter are connected with the upper computer, and receive instructions sent by the upper computer.
  • the screw feeding control device controls the screw feeding amount according to the screw speed of the screw feeder, that is, the screw feeding amount is set by the upper computer to realize the quantitative feeding of the disc mill system;
  • the mill speed adjusting device controls the mill speed by adjusting the frequency of the frequency converter
  • the grinding disc gap servo control device drives the moving grinding disc to move horizontally, taking the position of the static grinding disc as the zero position, collecting the position signal of the moving grinding disc in real time through a displacement sensor, and feeding it back to the host computer system for calculating the grinding disc gap, and driving the moving grinding disc to adjust according to the calculation result Position, so as to realize the precise positioning of the position movement of the movable grinding disc, and realize the control of the grinding disc gap through the precise positioning of the movable grinding disc;
  • the DC motor 2 performs sampling in the starting state, and controls the sampling amount of the powder according to the length of the motor's starting time, and sends it to the laser particle size analyzer through the sampling sleeve; in the sampling device, the DC motor 2 is in the stopped state, The produced powder enters the powder recovery box to realize intermittent sampling; the sampling amount in each detection cycle is the same, and the sampling amount is adjusted by controlling the start time of the DC motor 2 in the sampling device;
  • the host computer contains a random distribution control algorithm.
  • the initial screw feed amount, mill speed, grinding disc gap, PDF shape of the target powder particle size distribution, start time and stop of the DC motor 2 in the sampling device are set by the host computer.
  • the random distribution control algorithm is used to update the set values of the screw feeding amount, the mill speed, and the grinding disc gap;
  • the AC servo motor is used to drive the position of the movable grinding disc, and the control of the grinding disc gap is realized through the precise positioning of the movable grinding disc;
  • the random distribution control algorithm is used to calculate the updated setting values of the screw feeding amount, the mill speed, and the grinding disc gap; the laser The PDF shape curve of the powder particle size distribution detected by the particle size analyzer is available for users to query through the upper computer.
  • the laser particle size analyzer adopts the Sinpatec powder laser particle size analyzer to detect the particle size distribution of the powder sample obtained by the sampling device to obtain the PDF shape of the powder particle size distribution.
  • the control method of the present invention for the random distribution control experimental device for the particle size of the disc mill system adopts the random distribution control experimental device for the particle size of the disc mill system, including the following steps:
  • Step 1 Perform initial settings
  • start time and stop time of the sampling device are set according to the required detection time of the laser particle size analyzer
  • the initial mill speed is set according to the actual material hardness and the production efficiency of the disc mill system
  • the control process of the spiral feeding amount is: adjusting the spiral speed according to the actual spiral feeding amount measured by the electronic scale to realize the quantitative feeding of the spiral feeding control device, and then feeding the material into the disc grinding system
  • the control process of the mill speed is: controlling the mill speed of the disc grinding system by adjusting the frequency of the frequency converter until the actual mill speed reaches the set mill speed;
  • the control process of the grinding disc gap is: the host computer system controls the speed of the servo motor to drive the disc mill to move horizontally, takes the position of the static grinding disc as the zero position, adjusts the moving speed through the reducer during the movement, and the displacement sensor collects the position signal feedback of the moving disc in real time. Go to the host computer and compare with the set grinding disc gap to determine whether the movable grinding disc reaches the specified position, and control the grinding disc gap servo control device to adjust the position of the movable grinding disc accordingly, so as to reach the set grinding disc gap;
  • Step 2 Start the disc grinding system
  • Step 3 The disc mill system performs sampling according to the start time and stop time of the sampling device
  • the control process of the sampling device is as follows: the sampling motor starts when the set start time of the sampling motor is reached, sampling is performed, and the powder enters the laser particle size analyzer; when the set stop time of the sampling motor is reached, the sampling motor stops and the powder enters The powder recovery box realizes intermittent sampling; the sampling amount in each detection cycle is the same, and the sampling amount is judged by adjusting the length of the start-up time;
  • Step 4 The laser particle size analyzer obtains the PDF shape of the actual powder particle size distribution according to the detection cycle, and feeds it back to the upper computer 8;
  • Step 5 If the PDF shape of the powder particle size distribution does not meet the PDF shape of the target powder particle size distribution, the set values of the screw feed volume, the disc gap and the mill speed are updated through the random distribution control algorithm; otherwise, the powder The PDF shape of the volume distribution meets the production requirements, and the control process ends.
  • y ⁇ [a, ⁇ ] be the uniformly bounded random variable describing the output of the dynamic random distribution system, that is, the output random variable;
  • u(k) ⁇ R m is the control input of the random distribution system at time k, which means that at any sampling time k, the random variable y is described by its PDF shape, and its definition is as follows:
  • ⁇ (y,u(k)) is the PDF of the output random variable y, that is, the output PDF;
  • P(a ⁇ y ⁇ ,u(k)) means that the random distributed system controls the input u(k) at time k
  • the probability that the output random variable y falls within the interval [a, ⁇ ] under the action, that is, the shape of the output PDF ⁇ (y,u(k)) is controlled by the input u(k);
  • the control input u(k) is the screw feed amount, the disc gap and the mill speed
  • the random variable y is the particle size of the powder
  • the output PDF ⁇ (y, u(k)) is the particle size distribution shape of the powder
  • a neural network to approximate the output PDF at any moment, that is, to approximate the output PDF by using a fixed structure of the neural network, such as B-spline neural network, RBF neural network, and link the weight of the neural network with the control input u(k) Up, that is, control the output PDF by controlling the weight of the neural network;
  • ⁇ i (u(k)) is the weight of the B-spline neural network at time k
  • B i (y) is the corresponding B-spline basis function
  • e(y,u(k)) is the approximation error, Ignore it
  • C(y) [B 1 (y),B 2 (y),...,B n-1 (y)]
  • V(k) [ ⁇ 1 (k), ⁇ 2 (k), ..., ⁇ n-1 (k)] T
  • h(V(k)) is the function expression of the first n-1 weights
  • V(k+1) f(V(k),u(k)) (4)
  • f( ⁇ ) is the functional relationship between the control input and the weight, which is a conventional linear function or non-linear function.
  • the system composed of formula (3) and formula (4) describes the control input of a random distributed system The relationship between u(k) and the output PDF; therefore, control the shape of the output PDF by designing a suitable control input u(k);
  • the PDF shape of the target powder particle size distribution is controlled by designing different control inputs u(k);
  • g(y) is the PDF shape of the target powder particle size distribution
  • the performance index of formula (7) is optimized to obtain the controller input u(k):
  • J is the performance index used to design the controller input u(k)
  • R 1 is the control input weight
  • the invention builds a random distribution control experimental device for the particle size of the disc mill system.
  • the device can not only be used to verify a series of random distribution control algorithms, such as fixed structure PDF control algorithm, iterative learning PDF control algorithm, multi-step predictive PDF control Research on algorithms and robust PDF control algorithms provide a better experimental platform for teaching and scientific research.
  • the present invention proposes a random distribution control method for powder particle size of a disc mill system.
  • the method of the present invention can not only effectively realize the control of the non-Gaussian distribution shape of the output random variables of the random distribution system, but also effectively control the mean value and variance of the output random variables.
  • the invention is reasonable in design, easy to realize, and has good practical value.
  • Figure 1 is a schematic diagram of random distribution control in a specific embodiment of the present invention.
  • FIG. 2 is a front view of the structure of an experimental device for random particle size distribution control of a disc mill system in a specific embodiment of the present invention
  • Figure 3 is a side view of the structure of an experimental device for random particle size distribution control of the disc mill system in the specific embodiment of the present invention
  • Figure 4 is a diagram of a spiral feeding control device for a disc mill system in a specific embodiment of the present invention.
  • Fig. 5 is a diagram of a mill speed control device and a disc gap servo control device of the disc grinding system in the specific embodiment of the present invention
  • Figure 6 is a diagram of a sampling device of a disc mill system in a specific embodiment of the present invention.
  • Fig. 7 is a control flow chart of an experimental device for random distribution control of powder particle size of a disc mill system in a specific embodiment of the present invention.
  • Figure 8 is a control flow chart of a screw feeding control device in a specific embodiment of the present invention.
  • Figure 9 is a control flow chart of a mill speed adjusting device in a specific embodiment of the present invention.
  • Fig. 10 is a flow chart of the servo control of the grinding disc gap in a specific embodiment of the present invention.
  • Figure 11 is a control flow chart of a sampling device in a specific embodiment of the present invention.
  • Fig. 12 is a PDF three-dimensional diagram of particle size distribution of powder in a specific embodiment of the present invention.
  • the present invention proposes a random distribution control experimental device for the particle size of the disc mill system.
  • the disc system is adopted.
  • the main purpose is to design different random distribution control algorithms.
  • the control of the particle size distribution shape of the grinding system. Therefore, the grinding disc system proposed by the present invention can be regarded as a typical random distribution control system, as shown in Figures 2 and 3.
  • the experimental device includes: a spiral feeding control device 1 , Mill speed adjustment device 2, grinding disc gap servo control device 3, sampling device 4, laser particle size analyzer 6, control cabinet 7 and host computer 8;
  • the disc mill system adopts a vertical double disc mill, including a movable grinding disc 5 and a corresponding static grinding disc;
  • the screw feeding control device 1 is shown in Fig. 4 and includes a screw feeder, a DC motor 1 10 and an electronic scale; the screw feeder is connected to the DC motor 1; the electronic scale is connected to the upper computer, To measure the actual screw feed volume;
  • the screw feeding control device 1 controls the screw feeding amount according to the screw speed of the screw feeder, that is, the upper computer 8 sets the screw feeding amount to realize the quantitative feeding of the disc mill system;
  • the mill speed adjusting device 2 is shown in Fig. 5, including a frequency converter and a three-phase asynchronous motor 11.
  • the three-phase asynchronous motor 11 is connected to the movable grinding disc 5 of the disc grinding system. After the material enters the grinding zone, the movable grinding disc 5 is in three positions. Phase asynchronous motor 11 drives down rotation; the frequency converter is installed in the control cabinet;
  • the mill speed adjusting device 2 controls the mill speed by adjusting the frequency of the frequency converter
  • the grinding disc gap servo control device 3 is shown in Fig. 5, including an AC servo motor 12, a reducer 13 and a displacement sensor 14.
  • the AC servo motor 12 is connected to the reducer 13, and the reducer 13 is connected to the movable grinding disc 5 of the disc grinding device.
  • the sensor is installed on the drive shaft connected to the movable grinding disc 5 in the disc grinding system, and connected with the upper computer 8.
  • the position signal of the movable grinding disc 5 is collected in real time, and the position signal is fed back to the upper computer 8 to calculate the disc gap.
  • the motor is connected to the PLC, and the PLC is connected to the upper computer 8.
  • the grinding disc gap servo control device 3 drives the movable grinding disc 5 to move horizontally, taking the position of the static grinding disc as the zero position, and collecting the position signal of the movable grinding disc 5 in real time through the displacement sensor 14 and feeding it back to the host computer system to calculate the grinding disc gap, according to the calculation result
  • the movable grinding disc 5 is driven to adjust the position, so as to realize the precise positioning of the position movement of the movable grinding disc 5, and the control of the grinding disc gap is realized through the precise positioning of the movable grinding disc 5.
  • the detailed control process of the grinding disc gap is shown in FIG.
  • the sampling device 4 as shown in Fig. 6, includes a DC motor 2 16, a sampling sleeve 15 and a powder recovery box 9; the sampling sleeves are respectively connected to the powder outlet of the grinding disc system and the powder recovery box;
  • the DC motor 2 16 in the sampling device 4 performs sampling in the startup state, and controls the sampling amount of the powder according to the length of the motor startup time, and sends it to the laser particle size analyzer through the sampling sleeve 15; the DC motor 2 16 in the sampling device 4 stops In the state, the produced powder enters the powder recovery box 9 to realize intermittent sampling; the control process of intermittent sampling is shown in Figure 11.
  • the sampling amount in each detection period is the same, and the sampling amount is adjusted by controlling the length of the start-up time of the DC motor 2 16 in the sampling device 4;
  • the laser particle size analyzer adopts the Simpatec powder laser particle size analyzer, which is used to detect the particle size distribution of the powder sample obtained by the sampling device 4 to obtain the PDF shape of the powder particle size distribution;
  • the control cabinet 7 is equipped with a PLC; the PLC is respectively connected to the DC motor 1 10 of the spiral feeding control device 1, the DC motor 2 16 of the sampling device 4, and the AC servo motor 12 of the disc grinding gap; the inverter is connected to The three-phase asynchronous motor of the mill speed adjusting device 2 is connected;
  • the PLC and the frequency converter are both connected to the upper computer 8 and receive instructions sent by the upper computer 8;
  • the host computer 8 includes a random distribution control algorithm.
  • the host computer 8 sets the initial screw feed volume, mill speed, grinding disc gap, PDF shape of the target powder particle size distribution, and the DC motor 2 16 in the sampling device 4 Start time and stop time, and receive the set value of spiral feed volume, mill speed, and disc gap obtained by using random distribution control algorithm;
  • the random distribution control algorithm is used to update the set values of the screw feeding amount, the mill speed, and the grinding disc gap;
  • the AC servo motor 12 is used to drive the position of the movable grinding disc 5 to move, and the control of the grinding disc gap is realized through the precise positioning of the movable grinding disc 5;
  • the random distribution control algorithm is used to calculate the updated setting values of the screw feeding amount, the mill speed, and the grinding disc gap; the laser The PDF shape curve of the powder particle size distribution detected by the particle size analyzer is available for users to query through the upper computer 8;
  • the present invention proposes a control method of a random distribution control experimental device for the particle size of the disc mill system. As shown in Figure 7, the random distribution control experimental device for the particle size of the disc mill system is adopted, and the specific content is:
  • Step 1 Perform initial settings
  • start time and stop time of the sampling device 4 are set according to the detection time required by the laser particle size analyzer
  • the initial mill speed is set according to the actual material hardness and the production efficiency of the disc mill system
  • the production efficiency is 15kg/h
  • the initial grinding disc gap is 0.6mm
  • the mill speed is 3000r/min
  • the start time and stop time of the sampling device 4 are 10s and 30s, respectively;
  • the control process of the screw feeding amount is shown in Figure 8.
  • the screw speed is adjusted according to the actual screw feeding amount measured by the electronic scale to realize the quantitative feeding of the screw feeding control device 1, and then the material Send to the grinding area of the disc grinding system;
  • the control process of the mill speed is shown in Figure 9.
  • the mill speed of the disc mill system is controlled by adjusting the frequency of the frequency converter until the actual mill speed reaches the set mill speed;
  • the control process of the grinding disc gap is shown in Fig. 10.
  • the host computer system controls the servo motor speed to drive the disc mill to move horizontally, taking the position of the static grinding disc as the zero position. During the movement, the speed is adjusted by the reducer 13 and the displacement sensor 14 collects in real time.
  • the position signal of the movable grinding disc 5 is fed back to the host computer 8, and compared with the set grinding disc gap, it is judged whether the movable grinding disc reaches the designated position, and the servo control device 3 of the grinding disc gap is controlled to adjust the position of the movable grinding disc 5 accordingly, so as to achieve the setting Fixed grinding disc gap;
  • Step 2 Start the disc grinding system
  • Step 3 The disc mill system performs sampling according to the start time and stop time of the sampling device 4;
  • the control process of the sampling device 4 is shown in Fig. 11.
  • the sampling motor start time is reached, the sampling motor is started, sampling is performed, and the powder enters the laser particle size analyzer; when the sampling motor stop time is reached, the sampling motor starts. Stop, the powder enters the powder recovery box 9 to realize intermittent sampling; the sampling amount in each detection cycle is the same, and the sampling amount is judged by adjusting the length of the start time.
  • Step 4 The laser particle size analyzer obtains the PDF shape of the actual powder particle size distribution according to the detection cycle, and feeds it back to the upper computer 8;
  • Step 5 If the PDF shape of the powder particle size distribution does not meet the PDF shape of the target powder particle size distribution, the set values of the screw feed volume, the disc gap and the mill speed are updated through the random distribution control algorithm; otherwise, the powder The PDF shape of the volume size distribution meets the production requirements, and the control process ends;
  • y ⁇ [a, ⁇ ] be the uniformly bounded random variable describing the output of the dynamic random distribution system, that is, the output random variable;
  • u(k) ⁇ R m is the control input of the random distribution system at time k, which means that at any sampling time k, the random variable y is described by its PDF shape, and its definition is as follows:
  • ⁇ (y,u(k)) is the PDF of the output random variable y, that is, the output PDF;
  • P(a ⁇ y ⁇ ,u(k)) means that the random distributed system controls the input u(k) at time k
  • the probability that the output random variable y falls within the interval [a, ⁇ ] under the action, that is, the shape of the output PDF ⁇ (y,u(k)) is controlled by the input u(k);
  • the control input u(k) is the screw feed amount, the disc gap and the mill speed
  • the random variable y is the particle size of the powder
  • the output PDF ⁇ (y, u(k)) is the shape of the particle size distribution of the powder
  • the PDF three-dimensional diagram of the particle size distribution of the powder in this embodiment is shown in FIG. 12;
  • a neural network to approximate the output PDF at any moment, that is, to approximate the output PDF by using a fixed structure of the neural network, such as B-spline neural network, RBF neural network, and link the weight of the neural network with the control input u(k) Up, that is, control the output PDF by controlling the weight of the neural network;
  • the B-spline neural network is used to approximate the output PDF to obtain the following:
  • ⁇ i (u(k)) is the weight of the B-spline neural network at time k
  • B i (y) is the corresponding B-spline basis function
  • e(y,u(k)) is the approximation error, In this embodiment, it is ignored;
  • C(y) [B 1 (y),B 2 (y),...,B n-1 (y)]
  • V(k) [ ⁇ 1 (k), ⁇ 2 (k), ..., ⁇ n-1 (k)] T
  • h(V(k)) is the function expression of the first n-1 weights
  • V(k+1) f(V(k),u(k)) (4)
  • f( ⁇ ) is the functional relationship between the control input and the weight, which is a conventional linear function or non-linear function.
  • the system composed of formula (3) and formula (4) describes the control input of a random distributed system The relationship between u(k) and the output PDF; therefore, the shape of the output PDF can be controlled by designing a suitable control input u(k).
  • the PDF shape of the target powder particle size distribution is controlled by designing different control inputs u(k);
  • g(y) is the PDF shape of the target powder particle size distribution
  • J is the performance index used to design the controller input u(k)
  • R 1 is the control input weight

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Food Science & Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Molecular Biology (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Mathematical Optimization (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Operations Research (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Algebra (AREA)
  • Crushing And Grinding (AREA)
  • Disintegrating Or Milling (AREA)

Abstract

一种面向盘磨系统粉体粒度的随机分布控制实验装置及方法，属于分布控制技术领域。实验装置包括螺旋喂料控制装置（1）、磨机转速调节装置（2）、磨盘间隙伺服控制装置（3）、取样装置（4）、激光粒度仪（6）、控制柜（7）和上位机（8）；实验装置的控制方法包括：1)初始设置；2)启动盘磨系统；3)取样装置（4）进行取样；4)激光粒度仪（6）检测实际粉体粒度分布的PDF形状；5)通过随机分布控制算法对初始设置进行更新。该装置和方法可以用于验证一系列随机分布控制算法，为教学和科研提供较好的实验平台；可以有效实现随机分布系统输出随机变量的非高斯分布形状的控制，而且也可以对输出随机变量的均值和方差进行有效控制。

Description

面向盘磨系统粉体粒度的随机分布控制实验装置及方法 技术领域
本发明属于分布控制技术领域，具体涉及一种面向盘磨系统粉体粒度的随机分布控制实验装置及方法。
背景技术
早期的随机系统控制的成果集中于对系统变量本身的统计特性，这些成果最典型的例子是线性高斯二次型、最小方差控制、具有马尔柯夫阶跃参数系统的随机控制等，控制的目标是系统输出的一阶和二阶统计特性即均值和方差。当系统受到的干扰为高斯噪声时，其方差和均值可以决定输出概率密度函数(Probability Density Function，PDF)的形状，然而，很多实际工业过程不满足高斯输出的假设条件，这时输出随机变量的PDF形状往往是非对称、多峰值的，此时采用传统的均值和方差并不能准确反映输出变量的随机分布特性。因此，从1996年开始，王宏教授提出直接设计控制器以使系统输出PDF形状跟踪给定目标PDF形状的思想，并系统地建立了许多建模及控制方法，这一研究框架称为随机分布控制(Stochastic Distribution Control，SDC，H.Wang.Bounded dynamic stochastic distributions modelling and control[M].Springer–Verlag(London)Ltd,2000)，这类控制从某种意义上说概括了常规随机系统中关于输出变量的均值和方差的控制。
另一方面，在控制工程实践中，关于某些变量的分布控制问题一直是一个具有挑战性的问题，而SDC的研究一直源于一些复杂工业过程控制需求，例如：在造纸过程中，纸张的二维质量分布是衡量纸品质量的关键工艺指标，因此，造纸过程可以看作一个典型的动态随机分布系统，而控制的关键要求是使纸张的二维质量分布尽可能满足期望给定分布形状。在化工聚合过程中，分子量分布常被看作产品质量控制和工艺优化中的关键质量指标，因此，聚合过程控制的主要目的是使聚合产品分子量分布形状符合给定分布形状。在燃烧过程中，火焰温度场分布常作为燃烧过程效益的重要指标，而燃烧过程控制的目的是通过选择合适的燃料输入量和过程参数，使得火焰温度场分布的形状满足给定要求。在粮食加工过程中，通常希望加工后的粮食颗粒大小分布形状符合期望的分布形状，使得碾碎后的粮食颗粒分布形状符合后续的食品加工工序要求，从而提高整个系统的控制质量及生产效率。
对于这种新颖的控制技术，目前缺乏验证其有效性和可行性的实验装置，使得该控制方法在实际工业过程中推广应用带来诸多不便，本发明提供一种面向盘磨系统粉体粒度的随机分布控制实验装置及方法。
发明内容
针对目前缺乏验证随机分布控制方法有效性和可行性的实验装置，本发明提供一种面向盘磨系统粉体粒度的随机分布控制实验装置及方法，采用磨盘系统，所述实验装置包括：螺旋喂料控制装置、磨机转速调节装置、磨盘间隙伺服控制装置、取样装置、激光粒度仪、控制柜和上位机；
所述盘磨系统采用立式双盘磨粉机，包括动磨盘和相对应的静磨盘；
所述螺旋喂料控制装置包括螺旋喂料器、直流电机1和电子秤；所述螺旋喂料器与直流电机1连接；所述电子秤与上位机连接，用于测量实际螺旋喂料量；
所述磨机转速调节装置，包括变频器和三相异步电动机；三相异步电动机连接盘磨系统的动磨盘，在物料进入磨区后，动磨盘在三相异步电动机带动下转动；所述变频器安装在控制柜内；
所述磨盘间隙伺服控制装置包括交流伺服电机、减速器和位移传感器；交流伺服电机连接减速器，减速器连接盘磨装置的动磨盘，位置传感器安装到所述盘磨系统中连接动磨盘的驱动轴上，并与上位机连接；
所述取样装置包括直流电机2、取样套管和粉体回收箱；所述取样套管分别磨盘系统的出粉口和粉体回收箱连接；
根据激光粒度仪的检测周期，设定直流电机2的启动时间和停止时间；
所述控制柜内部设置PLC；所述PLC分别与螺旋喂料控制装置的直流电机1、取样装置的直流电机2和盘磨间隙的交流伺服电机相连；所述变频器与磨机转速调节装置的三相异步电机相连；
所述PLC和变频器均与上位机连接，接收上位机发送的指令。
所述螺旋喂料控制装置根据螺旋喂料器的螺旋转速控制螺旋喂料量，即通过上位机设置螺旋喂料量，实现所述盘磨系统的定量喂料；
所述磨机转速调节装置通过调节变频器频率控制磨机转速；
所述磨盘间隙伺服控制装置驱动动磨盘水平移动，将静磨盘位置作为零点位置，通过位移传感器实时采集动磨盘的位置信号，反馈到上位机系统进行磨盘间隙的计算，根据计算结果驱动动磨盘调整位置，从而实现动磨盘的位置移动的精确定位，通过动磨盘的精确定位实现磨盘间隙的控制；
所述取样装置中直流电机2在启动状态下进行取样，并根据电机的启动时间长短控制粉体的取样量，通过取样套管送入激光粒度仪；取样装置中直流电机2在停止状态下，生产的 粉体进入粉体回收箱，实现间歇式取样；每个检测周期内的取样量是一样的，取样量通过控制取样装置中直流电机2启动时间的长短进行调节；
所述上位机中包含随机分布控制算法，通过上位机设定初始的螺旋喂料量、磨机转速、磨盘间隙、目标粉体粒度分布的PDF形状、取样装置中直流电机2的启动时间和停止时间以及接收利用随机分布控制算法得到的螺旋喂料量、磨机转速、磨盘间隙的设定值；
根据实际粉体粒度分布的PDF形状与设定目标粉体粒度分布的PDF形状，利用所述随机分布控制算法更新螺旋喂料量、磨机转速、磨盘间隙的设定值；
通过控制螺旋喂料控制装置中的螺旋转速实现螺旋喂料量的控制；
通过调节变频器频率控制磨机转速；
利用交流伺服电机驱动动磨盘的位置移动，通过动磨盘的精确定位实现磨盘间隙的控制；
根据激光粒度仪检测的粉体粒度分布的PDF形状与目标粉体粒度分布的PDF形状，利用随机分布控制算法计算螺旋喂料量、磨机转速、磨盘间隙的更新设定值；所述的激光粒度仪检测的粉体粒度分布的PDF形状曲线，供使用者通过上位机查询。
所述激光粒度仪采用新帕泰克粉体激光粒度仪，用于对取样装置得到的粉体样本的粒度分布进行检测，得到粉体粒度分布的PDF形状。
本发明的面向盘磨系统粉体粒度的随机分布控制实验装置的控制方法，采用所述的面向盘磨系统粉体粒度的随机分布控制实验装置，包括以下步骤：
步骤1，进行初始设置；
根据盘磨系统的生产效率，通过上位机系统设置目标粉体粒度分布的PDF形状，设置初始的螺旋喂料量、取样装置的启动时间和停止时间，初始的磨盘间隙以及初始的磨机转速；
其中，所述取样装置的启动时间和停止时间，根据激光粒度仪所需检测时间设置；
所述初始的磨机转速根据实际物料硬度以及盘磨系统的生产效率进行设置；
所述螺旋喂料量的控制流程为：根据电子秤测量的实际的螺旋喂料量值对螺旋转速进行调节，实现所述螺旋喂料控制装置的定量喂料，然后将物料送入盘磨系统的磨区；
所述磨机转速的控制流程为：通过调节变频器频率控制盘磨系统的磨机转速，直至实际磨机转速达到设定的磨机转速；
所述磨盘间隙的控制流程为：上位机系统控制伺服电机转速带动盘磨水平移动，将静磨盘位置作为零点位置，移动过程中通过减速器调节移动速度，位移传感器实时采集动磨盘的位置信号反馈到上位机，并与设定的磨盘间隙的对比，判断动磨盘是否到达指定位置，并控制磨盘间隙伺服控制装置对动磨盘位置进行相应调整，从而达到设定的磨盘间隙；
步骤2，启动盘磨系统；
步骤3，盘磨系统根据取样装置的启动时间和停止时间进行取样；
所述取样装置的控制流程为：达到设定的取样电机的启动时间时取样电机启动，进行取样，粉体进入激光粒度仪；达到设定的取样电机的停止时间时取样电机停止，粉体进入粉体回收箱，实现间歇式取样；每个检测周期内的取样量是一样的，取样量通过调节启动时间的长短进行判断；
步骤4，激光粒度仪按检测周期来获得实际粉体粒度分布的PDF形状，并反馈给上位机8；
步骤5，若粉体粒度分布的PDF形状不满足目标粉体粒度分布的PDF形状，则通过随机分布控制算法对螺旋喂料量、磨盘间隙和磨机转速的设定值进行更新；反之，粉体粒度分布的PDF形状满足生产要求，控制流程结束。
根据随机分布控制理论，随机分布系统的建模和控制方法描述如下：
记y∈[a,ξ]为描述动态随机分布系统输出的一致有界随机变量，即输出随机变量；u(k)∈R m为k时刻的随机分布系统控制输入，这表明在任一采样时刻k，随机变量y通过其PDF形状来描述，其定义式如下：
Figure PCTCN2019081827-appb-000001
式中，γ(y,u(k))为输出随机变量y的PDF，即输出PDF；P(a≤y＜ξ,u(k))表示随机分布系统在k时刻控制输入u(k)作用下输出随机变量y落在区间[a,ξ]内的概率，也即输出PDFγ(y,u(k))形状由输入u(k)控制；
所述控制输入u(k)为螺旋喂料量、磨盘间隙和磨机转速；
所述随机变量y为粉体粒度，输出PDFγ(y,u(k))为粉体粒度分布形状；
利用神经网络来逼近任一时刻的输出PDF，即通过采用固定结构的神经网络来逼近输出PDF，如B样条神经网络、RBF神经网络，将神经网络的权值和控制输入u(k)联系起来，即通过控制神经网络的权值来实现对输出PDF的控制；
采用B样条神经网络逼近输出PDF得到如下：
Figure PCTCN2019081827-appb-000002
式中，ω i(u(k))为k时刻B样条神经网络的权值，B i(y)为相应的B样条基函数；e(y,u(k))为逼近误差，对其忽略不计；
因为输出PDF在其定义域上的积分应该始终等于1，这表明在n个权值中有n-1个所述权值是相互独立的，因此得到如下公式：
γ(y,u(k))＝C(y)V(k)+h(V(k))B n(y)                  (3)
式中，C(y)＝[B 1(y),B 2(y),…,B n-1(y)]，V(k)＝[ω 1(k),ω 2(k),…,ω n-1(k)] T，h(V(k))为前n-1个权值的函数表达式；考虑到如下神经网络逼近权值和控制输入u(k)之间存在如下动态关系：
V(k+1)＝f(V(k),u(k))                            (4)
式中，f(·)为控制输入与权值之间的函数关系，其为常规的线性函数或非线性函数，式(3)和式(4)所组成的系统描述随机分布系统的控制输入u(k)和输出PDF的关系；因此，通过设计合适的控制输入u(k)实现对输出PDF形状控制；
基于式(3)和式(4)随机分布系统的输出PDF动态模型表示如下：
Figure PCTCN2019081827-appb-000003
根据随机分布控制原理，基于如式(6)所示的跟踪PDF误差，通过设计不同的控制输入u(k)实现对目标粉体粒度分布的PDF形状的控制；
e(y,u(k))＝g(y)-γ(y,u(k))                          (6)
式中，g(y)为目标粉体粒度分布的PDF形状；
为了获得合适的控制器输入u(k)，如采用优化如式(7)性能指标得到控制器输入u(k)：
Figure PCTCN2019081827-appb-000004
式中，J分别为设计控制器输入u(k)采用的性能指标，R 1为控制输入权值。
本发明的有益效果：
本发明搭建面向盘磨系统粉体粒度的随机分布控制实验装置，该装置不但可以用于验证一系列随机分布控制算法，如固定结构的PDF控制算法，迭代学习PDF控制算法，多步预测PDF控制算法和鲁棒PDF控制算法等研究，为教学和科研提供较好的实验平台。
本发明提出一种面向盘磨系统粉体粒度的随机分布控制方法，对于实际工业过程不满足高斯输出的假设条件时，也即输出随机变量的PDF形状具有典型的非对称、多峰值、非高斯 分布特征，本发明所提方法不但可以有效实现随机分布系统输出随机变量的非高斯分布形状的控制，而且也可以对输出随机变量的均值和方差的进行有效控制。
本发明设计合理，易于实现，具有很好的实用价值。
附图说明
图1是本发明具体实施方式中随机分布控制原理图；
图2是本发明具体实施方式中盘磨系统粉体粒度随机分布控制实验装置结构正视图；
图3是本发明具体实施方式中盘磨系统粉体粒度随机分布控制实验装置结构侧视图；
图4是本发明具体实施方式中盘磨系统螺旋喂料控制装置图；
图5是本发明具体实施方式中盘磨系统磨机转速控制装置及磨盘间隙伺服控制装置图；
图6是本发明具体实施方式中盘磨系统取样装置图；
图7是本发明具体实施方式中盘磨系统粉体粒度的随机分布控制实验装置的控制流程图；
图8是本发明具体实施方式中螺旋喂料控制装置控制流程图；
图9是本发明具体实施方式中磨机转速调节装置控制流程图；
图10是本发明具体实施方式中磨盘间隙伺服控制流程图；
图11是本发明具体实施方式中取样装置控制流程图；
图12是本发明具体实施方式中粉体粒度分布的PDF三维图。
图中：1、螺旋喂料控制装置；2、磨机转速调节装置；3、磨盘间隙伺服控制装置；4、取样装置；5、动磨盘；6、激光粒度仪；7、控制柜；8、上位机；9、粉体回收箱；10、直流电机1；11、三相异步电动机；12、交流伺服电机；13、减速器；14、位移传感器；15、取样套管；16、直流电机2。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施实例，对本发明做出进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本发明，并不用于限定本发明。
为了验证随机分布控制方法的有效性和可行性，本发明提出一种面向盘磨系统粉体粒度的随机分布控制实验装置，采用磨盘系统，其主要目的是通过设计不同的随机分布控制算法实现盘磨系统粒度分布形状的控制，因此，本发明所提出的磨盘系统可以看作一种典型的随机分布控制系统，如图2和图3所示，所述实验装置包括：螺旋喂料控制装置1、磨机转速调节装置2、磨盘间隙伺服控制装置3、取样装置4、激光粒度仪6、控制柜7和上位机8；
所述盘磨系统采用立式双盘磨粉机，包括动磨盘5和相对应的静磨盘；
所述螺旋喂料控制装置1如图4所示，包括螺旋喂料器、直流电机1 10和电子秤；所述螺旋喂料器与直流电机1连接；所述电子秤与上位机连接，用于测量实际螺旋喂料量；
所述螺旋喂料控制装置1根据螺旋喂料器的螺旋转速控制螺旋喂料量，即通过上位机8设置螺旋喂料量，实现所述盘磨系统的定量喂料；
所述磨机转速调节装置2如图5所示，包括变频器和三相异步电动机11，三相异步电动机11连接盘磨系统的动磨盘5，在物料进入磨区后，动磨盘5在三相异步电动机11带动下转动；所述变频器安装在控制柜内；
所述磨机转速调节装置2通过调节变频器频率控制磨机转速；
所述磨盘间隙伺服控制装置3如图5所示，包括交流伺服电机12、减速器13和位移传感器14；交流伺服电机12连接减速器13，减速器13连接盘磨装置的动磨盘5，位置传感器安装到所述盘磨系统中连接动磨盘5的驱动轴上，并与上位机8连接，实时采集动磨盘5的位置信号，将该位置信号反馈到上位机8进行磨盘间隙的计算，伺服电机与PLC连接，PLC与上位机8连接。
所述磨盘间隙伺服控制装置3驱动动磨盘5水平移动，将静磨盘位置作为零点位置，通过位移传感器14实时采集动磨盘5的位置信号，反馈到上位机系统进行磨盘间隙的计算，根据计算结果驱动动磨盘5调整位置，从而实现动磨盘5的位置移动的精确定位，通过动磨盘5的精确定位实现磨盘间隙的控制，详细的磨盘间隙的控制流程如图10所示。
所述取样装置4如图6所示，包括直流电机2 16、取样套管15和粉体回收箱9；所述取样套管分别磨盘系统的出粉口和粉体回收箱连接；
根据激光粒度仪的检测周期，设定取样装置4中直流电机2 16的启动时间和停止时间；
取样装置4中直流电机2 16在启动状态下进行取样，并根据电机的启动时间长短控制粉体的取样量，通过取样套管15送入激光粒度仪；取样装置4中直流电机2 16在停止状态下，生产的粉体进入粉体回收箱9，实现间歇式取样；间歇式取样的控制流程如图11所示。每个检测周期内的取样量是一样的，取样量通过控制取样装置4中直流电机2 16启动时间的长短进行调节；
所述激光粒度仪采用新帕泰克粉体激光粒度仪，用于对取样装置4得到的粉体样本的粒度分布进行检测，得到粉体粒度分布的PDF形状；
所述控制柜7内部设置PLC；所述PLC分别与螺旋喂料控制装置1的直流电机1 10、取样装置4的直流电机2 16和盘磨间隙的交流伺服电机12相连；所述变频器与磨机转速调节装置2的三相异步电机相连；
所述PLC和变频器均与上位机8连接，接收上位机8发送的指令；
所述上位机8中包含随机分布控制算法，通过上位机8设定初始的螺旋喂料量、磨机转速、磨盘间隙、目标粉体粒度分布的PDF形状、取样装置4中直流电机2 16的启动时间和停止时间以及接收利用随机分布控制算法得到的螺旋喂料量、磨机转速、磨盘间隙的设定值；
根据实际粉体粒度分布的PDF形状与设定目标粉体粒度分布的PDF形状，利用所述随机分布控制算法更新螺旋喂料量、磨机转速、磨盘间隙的设定值；
通过控制螺旋喂料控制装置1中的螺旋转速实现螺旋喂料量的控制；
通过调节变频器频率控制磨机转速；
利用交流伺服电机12驱动动磨盘5的位置移动，通过动磨盘5的精确定位实现磨盘间隙的控制；
根据激光粒度仪检测的粉体粒度分布的PDF形状与目标粉体粒度分布的PDF形状，利用随机分布控制算法计算螺旋喂料量、磨机转速、磨盘间隙的更新设定值；所述的激光粒度仪检测的粉体粒度分布的PDF形状曲线，供使用者通过上位机8查询；
本发明提出一种面向盘磨系统粉体粒度的随机分布控制实验装置的控制方法，如图7所示，采用所述面向盘磨系统粉体粒度的随机分布控制实验装置，具体内容为：
步骤1，进行初始设置；
根据盘磨系统的生产效率，通过上位机系统设置目标粉体粒度分布的PDF形状，设置初始的螺旋喂料量、取样装置4的启动时间和停止时间，初始的磨盘间隙以及初始的磨机转速；
其中，所述取样装置4的启动时间和停止时间，根据激光粒度仪所需检测时间设置；
所述初始的磨机转速根据实际物料硬度以及盘磨系统的生产效率进行设置；
本实施例中，生产效率为15kg/h，初始的磨盘间隙为0.6mm，磨机转速为3000r/min，取样装置4的启动时间和停止时间分别为10s和30s；
所述螺旋喂料量的控制流程如图8所示，根据电子秤测量的实际的螺旋喂料量值对螺旋转速进行调节，实现所述螺旋喂料控制装置1的定量喂料，然后将物料送入盘磨系统的磨区；
所述磨机转速的控制流程如图9所示，通过调节变频器频率控制盘磨系统的磨机转速，直至实际磨机转速达到设定的磨机转速；
所述磨盘间隙的控制流程如图10所示，上位机系统控制伺服电机转速带动盘磨水平移动，将静磨盘位置作为零点位置，移动过程中通过减速器13调节移动速度，位移传感器14实时采集动磨盘5的位置信号反馈到上位机8，并与设定的磨盘间隙的对比，判断动磨盘是否到达指定位置，并控制磨盘间隙伺服控制装置3对动磨盘5位置进行相应调整，从而达到设定的磨盘间隙；
步骤2，启动盘磨系统；
步骤3，盘磨系统根据取样装置4的启动时间和停止时间进行取样；
所述取样装置4的控制流程如图11所示，达到设定的取样电机的启动时间时取样电机启动，进行取样，粉体进入激光粒度仪；达到设定的取样电机的停止时间时取样电机停止，粉体进入粉体回收箱9，实现间歇式取样；每个检测周期内的取样量是一样的，取样量通过调节启动时间的长短进行判断。
步骤4，激光粒度仪按检测周期来获得实际粉体粒度分布的PDF形状，并反馈给上位机8；
步骤5，若粉体粒度分布的PDF形状不满足目标粉体粒度分布的PDF形状，则通过随机分布控制算法对螺旋喂料量、磨盘间隙和磨机转速的设定值进行更新；反之，粉体粒度分布的PDF形状满足生产要求，控制流程结束；
根据随机分布控制理论，随机分布系统的建模和控制方法如图1所示，描述如下：
记y∈[a,ξ]为描述动态随机分布系统输出的一致有界随机变量，即输出随机变量；u(k)∈R m为k时刻的随机分布系统控制输入，这表明在任一采样时刻k，随机变量y通过其PDF形状来描述，其定义式如下：
Figure PCTCN2019081827-appb-000005
式中，γ(y,u(k))为输出随机变量y的PDF，即输出PDF；P(a≤y＜ξ,u(k))表示随机分布系统在k时刻控制输入u(k)作用下输出随机变量y落在区间[a,ξ]内的概率，也即输出PDFγ(y,u(k))形状由输入u(k)控制；
所述控制输入u(k)为螺旋喂料量、磨盘间隙和磨机转速；
所述随机变量y为粉体粒度，输出PDFγ(y,u(k))为粉体粒度分布形状；本实施方式中粉体粒度分布的PDF三维图如图12所示；
利用神经网络来逼近任一时刻的输出PDF，即通过采用固定结构的神经网络来逼近输出PDF，如B样条神经网络、RBF神经网络，将神经网络的权值和控制输入u(k)联系起来，即通过控制神经网络的权值来实现对输出PDF的控制；
本实施例中，采用B样条神经网络逼近输出PDF得到如下：
Figure PCTCN2019081827-appb-000006
式中，ω i(u(k))为k时刻B样条神经网络的权值，B i(y)为相应的B样条基函数；e(y,u(k))为逼近误差，本实施例中，对其忽略不计；
因为输出PDF在其定义域上的积分应该始终等于1，这表明在n个权值中有n-1个所述权值是相互独立的，因此得到如下公式：
γ(y,u(k))＝C(y)V(k)+h(V(k))B n(y)                   (3)
式中，C(y)＝[B 1(y),B 2(y),…,B n-1(y)]，V(k)＝[ω 1(k),ω 2(k),…,ω n-1(k)] T，h(V(k))为前n-1个权值的函数表达式；考虑到如下神经网络逼近权值和控制输入u(k)之间存在如下动态关系：
V(k+1)＝f(V(k),u(k))                             (4)
式中，f(·)为控制输入与权值之间的函数关系，其为常规的线性函数或非线性函数，式(3)和式(4)所组成的系统描述随机分布系统的控制输入u(k)和输出PDF的关系；因此，通过设计合适的控制输入u(k)实现对输出PDF形状控制。
基于式(3)和式(4)随机分布系统的输出PDF动态模型表示如下：
Figure PCTCN2019081827-appb-000007
根据随机分布控制原理，基于如式(6)所示的跟踪PDF误差，通过设计不同的控制输入u(k)实现对目标粉体粒度分布的PDF形状的控制；
e(y,u(k))＝g(y)-γ(y,u(k))                           (6)
式中，g(y)为目标粉体粒度分布的PDF形状；
为了获得合适的控制器输入u(k)，采用式(7)优化性能指标得到控制器输入u(k)：
Figure PCTCN2019081827-appb-000008
式中，J为设计控制器输入u(k)采用的性能指标，R 1为控制输入权值。

Claims (5)

  1. 一种面向盘磨系统粉体粒度的随机分布控制实验装置，采用盘磨系统，其特征在于，所述实验装置包括：螺旋喂料控制装置、磨机转速调节装置、磨盘间隙伺服控制装置、取样装置、激光粒度仪、控制柜和上位机；
    所述盘磨系统采用立式双盘磨粉机，包括动磨盘和相对应的静磨盘；
    所述螺旋喂料控制装置包括螺旋喂料器、直流电机1和电子秤；所述螺旋喂料器与直流电机1连接；所述电子秤与上位机连接，用于测量实际螺旋喂料量；
    所述磨机转速调节装置，包括变频器和三相异步电动机；三相异步电动机连接盘磨系统的动磨盘，在物料进入磨区后，动磨盘在三相异步电动机带动下转动；所述变频器安装在控制柜内；
    所述磨盘间隙伺服控制装置包括交流伺服电机、减速器和位移传感器；交流伺服电机连接减速器，减速器连接盘磨装置的动磨盘，位置传感器安装到所述盘磨系统中连接动磨盘的驱动轴上，并与上位机连接；
    所述取样装置包括直流电机2、取样套管和粉体回收箱；所述取样套管分别与盘磨系统的出粉口和粉体回收箱连接；
    根据激光粒度仪的检测周期，设定直流电机2的启动时间和停止时间；
    所述控制柜内部设置PLC；所述PLC分别与螺旋喂料控制装置的直流电机1、取样装置的直流电机2和盘磨间隙的交流伺服电机相连；所述变频器与磨机转速调节装置的三相异步电机相连；
    所述PLC和变频器均与上位机连接，接收上位机发送的指令。
  2. 根据权利要求1所述的面向盘磨系统粉体粒度的随机分布控制实验装置，其特征在于，所述螺旋喂料控制装置根据螺旋喂料器的螺旋转速控制螺旋喂料量，即通过上位机设置螺旋喂料量，实现所述盘磨系统的定量喂料；
    所述磨机转速调节装置通过调节变频器频率控制磨机转速；
    所述磨盘间隙伺服控制装置驱动动磨盘水平移动，将静磨盘位置作为零点位置，通过位移传感器实时采集动磨盘的位置信号，反馈到上位机系统进行磨盘间隙的计算，根据计算结果驱动动磨盘调整位置，从而实现动磨盘的位置移动的精确定位，通过动磨盘的精确定位实现磨盘间隙的控制；
    所述取样装置中直流电机2在启动状态下进行取样，并根据电机的启动时间长短控制粉体的取样量，通过取样套管送入激光粒度仪；取样装置中直流电机2在停止状态下，生产的粉体进入粉体回收箱，实现间歇式取样；每个检测周期内的取样量是一样的，取样量通过控制取样装置中直流电机2启动时间的长短进行调节；
    所述上位机中包含随机分布控制算法，通过上位机设定初始的螺旋喂料量、磨机转速、磨盘间隙、目标粉体粒度分布的PDF形状、取样装置中直流电机2的启动时间和停止时间以及接收利用随机分布控制算法得到的螺旋喂料量、磨机转速、磨盘间隙的设定值；
    根据实际粉体粒度分布的PDF形状与设定目标粉体粒度分布的PDF形状，利用所述随机分布控制算法更新螺旋喂料量、磨机转速、磨盘间隙的设定值；
    通过控制螺旋喂料控制装置中的螺旋转速实现螺旋喂料量的控制；
    通过调节变频器频率控制磨机转速；
    利用交流伺服电机驱动动磨盘的位置移动，通过动磨盘的精确定位实现磨盘间隙的控制；
    根据激光粒度仪检测的粉体粒度分布的PDF形状与目标粉体粒度分布的PDF形状，利用随机分布控制算法计算螺旋喂料量、磨机转速、磨盘间隙的更新设定值；所述的激光粒度仪检测的粉体粒度分布的PDF形状曲线，供使用者通过上位机查询。
  3. 根据权利要求1所述的面向盘磨系统粉体粒度的随机分布控制实验装置，其特征在于，所述激光粒度仪采用新帕泰克粉体激光粒度仪，用于对取样装置得到的粉体样本的粒度分布进行检测，得到粉体粒度分布的PDF形状。
  4. 一种面向盘磨系统粉体粒度的随机分布控制实验装置的控制方法，其特征在于，采用权力要求1所述的面向盘磨系统粉体粒度的随机分布控制实验装置，包括以下步骤：
    步骤1，进行初始设置；
    根据盘磨系统的生产效率，通过上位机系统设置目标粉体粒度分布的PDF形状，设置初始的螺旋喂料量、取样装置的启动时间和停止时间，初始的磨盘间隙以及初始的磨机转速；
    其中，所述取样装置的启动时间和停止时间，根据激光粒度仪所需检测时间设置；
    所述初始的磨机转速根据实际物料硬度以及盘磨系统的生产效率进行设置；
    所述螺旋喂料量的控制流程为：根据电子秤测量的实际的螺旋喂料量值对螺旋转速进行调节，实现所述螺旋喂料控制装置的定量喂料，然后将物料送入盘磨系统的磨区；
    所述磨机转速的控制流程为：通过调节变频器频率控制盘磨系统的磨机转速，直至实际磨机转速达到设定的磨机转速；
    所述磨盘间隙的控制流程为：上位机系统控制伺服电机转速带动盘磨水平移动，将静磨盘位置作为零点位置，移动过程中通过减速器调节移动速度，位移传感器实时采集动磨盘的位置信号反馈到上位机，并与设定的磨盘间隙的对比，判断动磨盘是否到达指定位置，并控制磨盘间隙伺服控制装置对动磨盘位置进行相应调整，从而达到设定的磨盘间隙；
    步骤2，启动盘磨系统；
    步骤3，盘磨系统根据取样装置的启动时间和停止时间进行取样；
    所述取样装置的控制流程为：达到设定的取样电机的启动时间时取样电机启动，进行取样，粉体进入激光粒度仪；达到设定的取样电机的停止时间时取样电机停止，粉体进入粉体回收箱，实现间歇式取样；每个检测周期内的取样量是一样的，取样量通过调节启动时间的长短进行判断；
    步骤4，激光粒度仪按检测周期来获得实际粉体粒度分布的PDF形状，并反馈给上位机；
    步骤5，若粉体粒度分布的PDF形状不满足目标粉体粒度分布的PDF形状，则通过随机分布控制算法对螺旋喂料量、磨盘间隙和磨机转速的设定值进行更新；反之，粉体粒度分布的PDF形状满足生产要求，控制流程结束。
  5. 根据权利要求4所述的面向盘磨系统粉体粒度的随机分布控制实验装置的控制方法，其特征在于，根据随机分布控制理论，随机分布系统的建模和控制方法描述如下：
    记y∈[a,ξ]为描述动态随机分布系统输出的一致有界随机变量，即输出随机变量；u(k)∈R m为k时刻的随机分布系统控制输入，这表明在任一采样时刻k，随机变量y通过其PDF形状来描述，其定义式如下：
    Figure PCTCN2019081827-appb-100001
    式中，γ(y,u(k))为输出随机变量y的PDF，即输出PDF；P(a≤y＜ξ,u(k))表示随机分布系统在k时刻控制输入u(k)作用下输出随机变量y落在区间[a,ξ]内的概率，也即输出PDFγ(y,u(k))形状由输入u(k)控制；
    所述控制输入u(k)为螺旋喂料量、磨盘间隙和磨机转速；
    所述随机变量y为粉体粒度，输出PDFγ(y,u(k))为粉体粒度分布形状；
    利用神经网络来逼近任一时刻的输出PDF，即通过采用固定结构的神经网络来逼近输出PDF，如B样条神经网络、RBF神经网络，将神经网络的权值和控制输入u(k)联系起来，即通过控制神经网络的权值来实现对输出PDF的控制；
    采用B样条神经网络逼近输出PDF得到如下：
    Figure PCTCN2019081827-appb-100002
    式中，ω i(u(k))为k时刻B样条神经网络的权值，B i(y)为相应的B样条基函数；e(y,u(k))为逼近误差，对其忽略不计；
    因为输出PDF在其定义域上的积分应该始终等于1，这表明在n个权值中有n-1个所述 权值是相互独立的，因此得到如下公式：
    γ(y,u(k))＝C(y)V(k)+h(V(k))B n(y)        (3)
    式中，C(y)＝[B 1(y),B 2(y),…,B n-1(y)]，V(k)＝[ω 1(k),ω 2(k),…,ω n-1(k)] T，h(V(k))为前n-1个权值的函数表达式；考虑到如下神经网络逼近权值和控制输入u(k)之间存在如下动态关系：
    V(k+1)＝f(V(k),u(k))           (4)
    式中，f(·)为控制输入与权值之间的函数关系，其为常规的线性函数或非线性函数，式(3)和式(4)所组成的系统描述随机分布系统的控制输入u(k)和输出PDF的关系；因此，通过设计合适的控制输入u(k)实现对输出PDF形状控制；
    基于式(3)和式(4)随机分布系统的输出PDF动态模型表示如下：
    Figure PCTCN2019081827-appb-100003
    根据随机分布控制原理，基于如式(6)所示的跟踪PDF误差，通过设计不同的控制输入u(k)实现对目标粉体粒度分布的PDF形状的控制；
    e(y,u(k))＝g(y)-γ(y,u(k))         (6)
    式中，g(y)为目标粉体粒度分布的PDF形状；
    为了获得合适的控制器输入u(k)，如采用优化如式(7)性能指标得到控制器输入u(k)：
    Figure PCTCN2019081827-appb-100004
    式中，J分别为设计控制器输入u(k)采用的性能指标，R 1为控制输入权值。
PCT/CN2019/081827 2019-03-29 2019-04-09 面向盘磨系统粉体粒度的随机分布控制实验装置及方法 WO2020199231A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/483,706 US11860078B2 (en) 2019-03-29 2019-04-09 Particle size distribution control in disc milling system based stochastic distribution control experimental device and method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910250610.X 2019-03-29
CN201910250610.XA CN109847921A (zh) 2019-03-29 2019-03-29 面向盘磨系统粉体粒度的随机分布控制实验装置及方法

Publications (1)

Publication Number Publication Date
WO2020199231A1 true WO2020199231A1 (zh) 2020-10-08

Family

ID=66902596

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/081827 WO2020199231A1 (zh) 2019-03-29 2019-04-09 面向盘磨系统粉体粒度的随机分布控制实验装置及方法

Country Status (2)

Country Link
CN (1) CN109847921A (zh)
WO (1) WO2020199231A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112169914B (zh) * 2020-10-12 2023-06-02 张宏悦 磨盘自动调节间隙的方法及系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002072310A2 (en) * 2001-03-12 2002-09-19 Norwalk Industrial Components, Llc Method for controlling a disk mill
CN101654887A (zh) * 2009-09-11 2010-02-24 华南理工大学 中浓盘磨机微调控制系统与方法
CN105568732A (zh) * 2015-12-17 2016-05-11 镇江市高等专科学校 盘磨机控制方法
CN107442262A (zh) * 2017-08-04 2017-12-08 东北大学 一种盘磨系统的粉体粒度分布控制装置及方法
CN108536017A (zh) * 2018-05-03 2018-09-14 山东师范大学 基于动态反馈控制的随机分布互联系统协作容错控制方法
CN108846178A (zh) * 2018-05-30 2018-11-20 东北大学 一种盘磨系统的粉体粒度分布形状估计方法及其系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908235A (zh) * 2010-06-23 2010-12-08 苏州科技学院 B样条曲面建模新方法
CN106777807B (zh) * 2017-01-13 2020-09-25 北京航空航天大学 一种粉末冶金随机粒度分布3d有限元建模与仿真方法
CN109190226B (zh) * 2018-08-24 2023-01-31 东北大学 一种磨矿系统溢流粒度指标软测量方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002072310A2 (en) * 2001-03-12 2002-09-19 Norwalk Industrial Components, Llc Method for controlling a disk mill
CN101654887A (zh) * 2009-09-11 2010-02-24 华南理工大学 中浓盘磨机微调控制系统与方法
CN105568732A (zh) * 2015-12-17 2016-05-11 镇江市高等专科学校 盘磨机控制方法
CN107442262A (zh) * 2017-08-04 2017-12-08 东北大学 一种盘磨系统的粉体粒度分布控制装置及方法
CN108536017A (zh) * 2018-05-03 2018-09-14 山东师范大学 基于动态反馈控制的随机分布互联系统协作容错控制方法
CN108846178A (zh) * 2018-05-30 2018-11-20 东北大学 一种盘磨系统的粉体粒度分布形状估计方法及其系统

Also Published As

Publication number Publication date
CN109847921A (zh) 2019-06-07

Similar Documents

Publication Publication Date Title
CN108809167A (zh) 一种基于模糊控制的bp神经网络pid调速控制算法
Xiong et al. Control of deposition height in WAAM using visual inspection of previous and current layers
Huang et al. Modified Smith fuzzy PID temperature control in an oil-replenishing device for deep-sea hydraulic system
CN103336107B (zh) 一种水泥熟料f-CaO含量软测量方法
CN109541168B (zh) 一种煤粉经济细度在线监测与调整方法
CN102944994B (zh) 基于不确定性离散模型的液压活套鲁棒模糊控制方法
CN106391286A (zh) 一种用于煤粉细度控制的系统及方法
WO2020199231A1 (zh) 面向盘磨系统粉体粒度的随机分布控制实验装置及方法
EP3933513A1 (en) Prediction control development device, prediction control development method, and prediction control development program
Gao et al. Process parameters optimization using a novel classification model for plastic injection molding
CN110782067A (zh) 一种基于模糊信息粒化的烧结终点波动范围预测方法
CN111443597B (zh) 用于控制立磨矿粉粒度的装置及方法
CN113048017A (zh) 基于内模控制的风电机组有功功率控制优化方法及系统
JP2013069094A (ja) 制御方法および制御装置
US11860078B2 (en) Particle size distribution control in disc milling system based stochastic distribution control experimental device and method
Wang et al. Research on variable universe fuzzy PID control strategy of pipe lining induction heating system
CN113885310A (zh) 一种用于真空干泵测试的智能控制系统
CN104898563B (zh) 一种球磨机控制方法
Takeuchi et al. Variation model abstraction and adaptive control based on element description method toward smart factory
Lin et al. Modeling of vertical mill raw meal grinding process and optimal setting of operating parameters based on wavelet neural network
Xiao et al. Research on the Control Strategy of Battery Roller Press Deflection Device by Introducing Genetic Algorithm to Optimize Integral Separation PID
Gräler et al. Assisted setup of forming processes: compensation of initial stochastic disturbances
Wang et al. Expert PID control of grinding condition for cement vertical roller mill
Yuan et al. Modeling for cement combined grinding system with closed grinding process based on LS-SVM in typical working condition
Chowdhury A Modeling and Controller Design for Continuous Electric Field Assisted Sintering System

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19922482

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19922482

Country of ref document: EP

Kind code of ref document: A1