RU2813252C1 - Control system and method for pelletizer control - Google Patents

Abstract

FIELD: pelletizers.

SUBSTANCE: group of inventions relates to a control system and a method for controlling a pelletizer. The system includes a rotation speed controller, an inclination angle controller, a water supply controller, a material supply controller and a central processing unit. The central processor is configured to carry out the following steps: predicting classification indices of raw pellets based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material, the amount of supplied water, the type and proportion of each component in the mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material to obtain predicted classification indices of raw pellets in multiple forecast periods; and optimizing, using a moving average optimization model, the pelletizer rotation speed, the pelletizer plate inclination angle, the amount of material supplied and the amount of water supplied based on the predicted classification index of the green pellets in each forecast period and the specified target classification index of the raw pellets in each forecast period, to control, in real time, the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material supplied to the pelletizer, and the amount of supplied water supplied to the pelletizer.

EFFECT: actual classification index of the raw pellets can meet the predetermined standard, and thus the pelletizing quality of the pelletizer can be improved.

10 cl, 7 dwg, 1 tbl

Description

TECHNICAL FIELD

[0001] The present application relates to the field of technology relating to iron and steel smelting, and in particular to a control system and method for controlling a pelletizer.

BACKGROUND OF THE ART

[0002] In the metallurgical industry, the production of iron ore pellets is currently a widely used iron ore processing technology. The pelletizing process is an important process in the iron ore pellet production line. The stability and improvement of yield and quality of raw pellets mainly depend on the pelletizing process. As the main equipment in the pelletizing process, pelletizers mainly include disc pelletizers and drum pelletizers. The drum pelletizer is generally used in large-scale mass production lines. However, the disc pelletizer is more widely used as small- and medium-scale pellet production lines now predominate.

[0003] When the pelletizer operates, materials are moved along different directions in the pelletizer to form green pellets of different diameters. After certain conditions are reached, the raw pellets are discharged from the pelletizer and enter the subsequent raw pellet receiving device. The size of the pelletizer granules is a key parameter in the pelletizing process. A higher raw pellet classification index indicates a higher pelletizer performance.

[0004] Once the raw material parameters are determined, the factors affecting pelletizing mainly include the rotation speed of the pelletizer, the angle of inclination of the pelletizer, the amount of materials entering the pelletizer, the amount of water added to the material in the pelletizer, the initial moisture content of the material entering the pelletizer , and so on. The pelletizing operator on the job site generally controls the pelletizer production process by pre-setting the appropriate rotation speed of the pelletizer, the inclination angle of the pelletizer, the amount of material supplied by the feed conveyor belt, and the amount of water supplied by the water supply device, based on the raw material parameters of the above ore mixing process combined with the actual situations of pellet production. Thus, it is expected that the yield and quality of the produced raw pellets can meet production requirements, and optimal iron ore pellets can be obtained.

[0005] However, when operating a pelletizer, many uncertain factors may be encountered in the pelletizing process, such as equipment wear, unsatisfactory raw material parameters, inconsistent quantity ratio and raw material moisture degree. These factors may cause the pellet size of the raw pellets, especially the yield of the specified raw pellets, to not meet the requirements of the process, thereby affecting the yield and quality of the pellet production line, and increasing the energy consumption of the process as well as operating costs.

BRIEF DESCRIPTION OF THE INVENTION

[0006] This application provides a control system and method for controlling a pelletizer that can be used to solve the prior art technical problem of a deterioration in pelletizing quality of a pelletizer due to the fact that the actual classification index does not meet a given standard.

[0007] According to a first aspect, an embodiment of the present application provides a control system for a pelletizer, the system including a pelletizer, a water supply device, and a belt weigher; the water outlet point of the water supply device is located at the feed point of the pelletizer and in the area of pellet formation in the pelletizer, and is configured to supply water to the pelletizer; the belt weigher is configured to supply mixed material to the pelletizer; the material falling point of the belt weigher is the feed point of the pelletizer; and the system further includes a rotation speed controller connected to the pelletizer, an inclination angle controller connected to the pelletizer, a water supply controller connected to the water supply device, a material supply controller connected to the belt weigher, and a central processing unit respectively connected to the controller rotation speed, tilt angle controller, water supply controller and material supply controller,

[0008] wherein the central processor is configured to carry out the following steps:

[0009] obtaining the rotation speed of the pelletizer sent by the rotation speed controller, the inclination angle of the pelletizer plate sent by the inclination angle controller, the amount of supplied water sent by the water supply controller, and the type and proportion of each component, the amount of mixed material supplied, the percentage of binder in the mixed material and initial moisture content of the mixed material sent by the material feed controller;

[0010] predicting the percentage of each fraction of raw pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of fed mixed material, the percentage of binder in the mixed material, and the initial moisture content of the mixed material material for obtaining predicted classification indices of raw pellets in multiple forecast periods;

[0011] calculating a deviation value of the raw pellet classification index in each forecast period based on the predicted raw pellet classification index in each forecast period and a predetermined target raw pellet classification index in each forecast period;

[0012] inputting the deviation values of the classification indexes of the raw pellets in a plurality of forecast periods into the moving average optimization model to obtain an adjustable pelletizer rotation speed, an adjustable pelletizer tray angle, an adjustable material supply amount, and an adjustable water supply amount; And

[0013] controlling the rotation speed controller to regulate the rotation speed of the pelletizer as the rotation speed of the pelletizer to be adjusted, controlling the inclination angle controller to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted, controlling the material supply controller to regulate the amount of feed material supplied to the pelletizer as a material supply amount to be controlled, and controlling the water supply controller to control the amount of supply water supplied to the pelletizer as the water supply amount to be controlled,

[0014] wherein the plurality of forecast periods includes the current period and periods following the current period; and the moving average optimization model is configured to calculate the corresponding rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material and the amount of supplied water when the standard deviation of the classification indices of raw pellets in multiple forecast periods is minimal, provided that the type and proportion of each component in a mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material remain unchanged.

[0015] In combination with the first aspect, in a possible embodiment of the first aspect, the classification index of the green pellets is determined based on the predicted percentage values of the green pellets of various parameters, and the predicted percentage values of the green pellets of the various parameters include the predicted percentage value of qualified large pellets, the predicted percentage value substandard large pellets, the predicted percentage of qualified medium pellets, the predicted percentage of qualified small pellets and the predicted percentage of substandard small pellets; And

[0016] the deviation value of the classification index of raw pellets in each forecast period is obtained according to the following formula:

[0017] E(k|k+j)={(r ₁ ((j)-y ₁ (k|k+j), r ₂ (j)-y ₂ (k|k+j),…,r _i (j)-y _i (k|k+j)},

[0018] where E(k|k+j) is the deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecast period and the reference value; r _i (j) represents the target percentage value of raw pellets of the i- ^th parameter with the j ^-th step; y _i (k|k+j) represents the predicted percentage value of raw pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; and j=1, 2, …, m, where m is an integer that is greater than or equal to 1.

[0019] In combination with the first aspect, in a possible embodiment of the first aspect, the standard deviation of the classification indices of raw pellets in a plurality of forecast periods is obtained according to the following formula:

[0020]

[0021] where σ (k|k+j) represents the root mean square error of the value

deviations between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _i (j) represents the target percentage value of small pellets of the i- ^th parameter with the j ^-th prediction step; y _i (k|k+j) represents the predicted percentage value of small pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; j=0, 1, 2, …, m, where m is an integer that is greater than or equal to 0; and k=1, 2, …, n, where n is an integer that is greater than or equal to 1.

[0022] In combination with the first aspect, in a possible embodiment of the first aspect, predicting the percentage of each fraction of green pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of supplied of the mixed material, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to obtain predicted classification indices of green pellets in multiple forecast periods, in particular, is carried out according to the following steps:

[0023] quantifying the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied and the amount of material supplied in the same interval based on the corresponding scaling factors;

[0024] obtaining a characteristic vector influencing pelletization based on the quantified rotation speed of the pelletizer, the quantified tilt angle of the pelletizer plate, the quantified amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the mixed material, the percentage binder in the mixed material and the initial moisture content of the mixed material; And

[0025] introducing a pelletizing characteristic vector into a percentage prediction model to obtain predicted percentage values of green pellets of various parameters, wherein the percentage prediction model includes displaying a relationship between the pelletizing characteristic vector and predicted percentages of raw pellets of various parameters.

[0026] In combination with the first aspect, in a possible embodiment of the first aspect, the scaling factor corresponding to the rotation speed of the pelletizer is the maximum rotation speed of the pelletizer;

[0027] the scaling factor corresponding to the water supply amount represents the maximum water supply amount of the water supply pipe in the water supply system;

[0028] the scaling factor corresponding to the feeding amount is the maximum feeding amount of the feeding belt; And

[0029] The scaling factor corresponding to the inclination angle of the pelletizer tray is the maximum inclination angle of the pelletizer.

[0030] In combination with the first aspect, in a possible embodiment of the first aspect, a percentage prediction model is obtained as follows:

[0031] obtaining sampling data in N historical forecasting periods, wherein the sampling data in each historical forecasting period includes a historical pelletizer rotation speed, a historical pelletizer tray angle, a historical amount of water supplied, historical pelletizer raw material information for obtaining sample raw pellets, and also measured percentage values of selected raw pellets of various parameters; the historical raw material information includes the historical quantity of feed material, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material; the measured percentage values of the sample raw pellets of various parameters are obtained through analysis and calculation after obtaining and processing images of the sample raw pellets in each historical forecast period using a computer vision method;

[0032] quantifying the historical pelletizer rotation speed, the historical pelletizer tray angle, the historical feed amount of water, and the historical feed amount of material in the same interval based on appropriate scaling factors;

[0033] obtaining characteristic vectors of N samples influencing pelletization based on the quantified historical rotation speed of the pelletizer, the quantified historical tilt angle of the pelletizer plate, the quantified historical amount of water supplied, the quantified amount of material supplied, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material;

[0034] using the characteristic vectors of N samples influencing pelletization as input to the prediction model, and using the actual percentage values of sample raw pellets of various parameters in N historical prediction periods as output of the prediction model to train the percentage prediction model using error backpropagation method;

[0035] continuously updating the weight parameter, bias parameter, and learning coefficient of the percentage prediction model through iterative learning; And

[0036] if the difference between the predicted percentage value of the percentage prediction model for sample raw pellets of various parameters and the actual percentage value of sample raw pellets of various parameters is within a specified range of acceptable values, or if the percentage prediction model reaches a specified maximum number of iterations during an iteration operation , completes the training and saves the weight parameter, bias parameter, and training coefficient that were most recently updated.

[0037] In combination with the first aspect, in a possible embodiment of the first aspect, a percentage prediction model is determined based on a prediction model using a long short-term memory neural network LSTM.

[0038] In combination with the first aspect, in a possible embodiment of the first aspect, the system further includes an image acquisition device and an image processing device, the image acquisition device is located in the discharge opening of the pelletizer and connected to the image processing device, and the processing device images are connected to the central processor;

[0039] the image acquisition device is configured to perform the following step: obtaining image data of the pelletizer pellet outlet, and sending image data of the pellet outlet to the image processing device; And

[0040] The image processing device is configured to perform the following steps:

[0041] performing image pre-processing using the pellet outlet image data to separate the sample raw pellet image data and the background image data;

[0042] obtaining a central bright spot of the sample raw pellet based on the image data of the sample raw pellet;

[0043] determining the outline of the sample green pellet based on the image data of the sample raw pellet and the background image data;

[0044] obtaining a particle size of the sample raw pellet based on the central bright spot of the sample raw pellet and the outline of the sample raw pellet;

[0045] determining a parameter of the sample raw pellet based on the particle size of the sample raw pellet and a predetermined correspondence between the particle fraction and the parameter of the raw pellet;

[0046] calculating the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters; And

[0047] determining a measured classification index of the sample raw pellets based on the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters, and sending the measured classification index of the sample raw pellets to the central processing unit.

[0048] In combination with the first aspect, in a possible embodiment of the first aspect, the rotation speed controller is configured to perform the following step:

[0049] determining the rotation speed of the pelletizer in the current period, and sending the rotation speed of the pelletizer to the central processor;

[0050] the inclination angle controller is configured to perform the following step:

[0051] determining the inclination angle of the pelletizer tray in the current period, and sending the inclination angle of the pelletizer tray to the central processor;

[0052] the water supply controller is configured to perform the following step:

[0053] determining the amount of supply water supplied by the water supply device to the pelletizer in the current period, and sending the amount of supply water to the central processor; And

[0054] The material supply controller is configured to perform the following step:

[0055] determining the type and proportion of each component in the mixed material supplied by the belt weigher to the pelletizer in the current period, the quantity of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material, and sending the type and proportion of each component in the mixed material material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to the central processor.

[0056] According to a second aspect, an embodiment of the present application provides a method for controlling a pelletizer, the method including:

[0057] predicting the percentage of each fraction of raw pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of fed mixed material, the percentage of binder in the mixed material, and the initial moisture content of the mixed material material for obtaining predicted classification indices of raw pellets in multiple forecast periods;

[0058] calculating a deviation value of the raw pellet classification index in each forecast period based on the predicted raw pellet classification index in each forecast period and a predetermined target raw pellet classification index in each forecast period;

[0059] inputting the deviation values of the classification indexes of the raw pellets in a plurality of forecast periods into the moving average optimization model to obtain an adjustable pelletizer rotation speed, an adjustable pelletizer tray angle, an adjustable material supply amount, and an adjustable water supply amount; And

[0060] controlling the rotation speed controller to regulate the rotation speed of the pelletizer as the rotation speed of the pelletizer to be adjusted, controlling the inclination angle controller to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted, controlling the material supply controller to regulate the amount of feed material supplied to the pelletizer as a material supply amount to be controlled, and controlling the water supply controller to control the amount of supply water supplied to the pelletizer as the water supply amount to be controlled,

[0061] wherein the plurality of forecast periods includes the current period and periods following the current period; and the moving average optimization model is configured to calculate the corresponding rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material and the amount of supplied water when the standard deviation of the classification indices of raw pellets in multiple forecast periods is minimal, provided that the type and proportion of each component in a mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material remain unchanged.

[0062] In combination with the second aspect, in a possible embodiment of the second aspect, the classification index of the green pellets is determined based on the predicted percentage values of the green pellets of various parameters, and the predicted percentage values of the green pellets of the various parameters include the predicted percentage value of qualified large pellets, the predicted percentage value substandard large pellets, the predicted percentage of qualified medium pellets, the predicted percentage of qualified small pellets and the predicted percentage of substandard small pellets; And

[0063] The deviation value of the classification index of raw pellets in each forecast period is obtained according to the following formula:

[0064] E(k|k+j)={(r ₁ (j)-y ₁ (k|k+j)), r ₂ (j)-y ₂ (k|k+j), …, r _i (j)-y _i (k|k+j)},

[0065] where E(k|k+j) is the deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecast period and the reference value; r _i (j) represents the target percentage value of raw pellets of the i- ^th parameter with the j ^-th step; y _i (k|k+j) represents the predicted percentage value of raw pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; and j=1, 2, …, m, where m is an integer that is greater than or equal to 1.

[0066] In combination with the second aspect, in a possible embodiment of the second aspect, the standard deviation of the classification indices of green pellets in a plurality of forecast periods is obtained according to the following formula:

[0067]

[0068] where σ (k|k+j) represents the root mean square error of the value

deviations between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _t (j) represents the target percentage value of small pellets of the i- ^th parameter with the j ^-th prediction step; y _i (k|k+j) represents the predicted percentage value of small pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; j=0, 1, 2, …, m, where m is an integer that is greater than or equal to 0; and k=1, 2, …, n, where n is an integer that is greater than or equal to 1.

[0069] In combination with the second aspect, in a possible embodiment of the second aspect, predicting the percentage of each fraction of green pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of of the mixed material, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to obtain predicted classification indices of green pellets in multiple forecast periods, in particular, is carried out according to the following steps:

[0070] quantifying the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied and the amount of material supplied in the same interval based on the corresponding scaling factors;

[0071] obtaining a characteristic vector influencing pelletization based on the quantified rotation speed of the pelletizer, the quantified tilt angle of the pelletizer plate, the quantified amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the mixed material, the percentage binder in the mixed material and the initial moisture content of the mixed material; And

[0072] introducing a pelletizing characteristic vector into a percentage prediction model to obtain predicted percentage values of green pellets of various parameters, wherein the percentage prediction model includes displaying a relationship between the pelletizing characteristic vector and predicted percentages of raw pellets of various parameters.

[0073] In combination with the second aspect, in a possible embodiment of the second aspect, the scaling factor corresponding to the rotation speed of the pelletizer is the maximum rotation speed of the pelletizer;

[0074] the scaling factor corresponding to the water supply amount represents the maximum water supply amount of the water supply pipe in the water supply system;

[0075] the scaling factor corresponding to the feeding amount is the maximum feeding amount of the feeding belt; And

[0076] The scaling factor corresponding to the inclination angle of the pelletizer tray is the maximum inclination angle of the pelletizer.

[0077] In combination with the second aspect, in a possible embodiment of the second aspect, a percentage prediction model is obtained as follows:

[0078] obtaining sampling data in N historical forecasting periods, wherein the sampling data in each historical forecasting period includes a historical pelletizer rotation speed, a historical pelletizer tray angle, a historical quantity of water supplied, historical pelletizer raw material information for obtaining sample raw pellets, and also measured percentage values of selected raw pellets of various parameters; the historical raw material information includes the historical quantity of feed material, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material; the measured percentage values of the sample raw pellets of various parameters are obtained through analysis and calculation after obtaining and processing images of the sample raw pellets in each historical forecast period using the computer vision method;

[0079] quantifying the historical pelletizer rotation speed, the historical pelletizer tray angle, the historical amount of water supplied, and the historical amount of material supplied in the same interval based on the corresponding scaling factors;

[0080] obtaining characteristic vectors of N samples influencing pelletization based on the quantified historical rotation speed of the pelletizer, the quantified historical tilt angle of the pelletizer plate, the quantified historical amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material;

[0081] using the characteristic vectors of N samples influencing pelletization as input to the prediction model, and using the actual percentage values of sample raw pellets of various parameters in N historical prediction periods as output of the prediction model to train the percentage prediction model using error backpropagation method;

[0082] continuously updating the weight parameter, bias parameter, and learning coefficient of the percentage prediction model through iterative learning; And

[0083] if the difference between the predicted percentage value of the percentage prediction model for sample raw pellets of various parameters and the actual percentage value of sample raw pellets of various parameters is within a specified range of acceptable values, or if the percentage prediction model reaches a specified maximum number of iterations during an iteration operation , completes the training and saves the weight parameter, bias parameter, and training coefficient that were most recently updated.

[0084] In combination with the second aspect, in a possible embodiment of the second aspect, a percentage prediction model is determined based on a prediction model using a long short-term memory LSTM neural network.

[0085] In combination with the second aspect, in a possible embodiment of the second aspect, the system further includes an image acquisition device and an image processing device, the image acquisition device is located in the discharge opening of the pelletizer and connected to the image processing device, and the processing device images are connected to the central processor;

[0086] the image acquisition device is configured to perform the following step: obtaining image data of the pelletizer pellet outlet, and sending image data of the pellet outlet to the image processing device; And

[0087] The image processing device is configured to perform the following steps:

[0088] performing image pre-processing using the pellet outlet image data to separate the sample raw pellet image data and the background image data;

[0089] obtaining a central bright spot of the sample raw pellet based on the image data of the sample raw pellet;

[0090] determining the outline of the sample green pellet based on the image data of the sample raw pellet and the background image data;

[0091] obtaining a particle size of the sample green pellet based on the central bright spot of the sample raw pellet and the outline of the sample green pellet;

[0092] determining a parameter of the sample raw pellet based on the particle size of the sample raw pellet and a predetermined correspondence between the particle fraction and the parameter of the raw pellet;

[0093] calculating the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters; And

[0094] determining a measured classification index of the sample raw pellets based on the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters, and sending the measured classification index of the sample raw pellets to the central processing unit.

[0095] In combination with the second aspect, in a possible embodiment of the second aspect, the rotation speed controller is configured to perform the following step:

[0096] determining the rotation speed of the pelletizer in the current period, and sending the rotation speed of the pelletizer to the central processor;

[0097] the inclination controller is configured to perform the following step:

[0098] determining the inclination angle of the pelletizer tray in the current period, and sending the inclination angle of the pelletizer tray to the central processor;

[0099] the water supply controller is configured to perform the following step:

[00100] determining the amount of supply water supplied by the water supply device to the pelletizer in the current period, and sending the amount of supply water to the central processor; And

[00101] The material supply controller is configured to perform the following step:

[00102] determining the type and proportion of each component in the mixed material supplied by the belt weigher to the pelletizer in the current period, the quantity of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material, and sending the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to the central processor.

[00103] In embodiments of the present application, the classification indices of the green pellets are predicted based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of material fed, the amount of water fed, the type and proportion of each component in the mixed material, the percentage of binder in the mixed material, and the starting material. moisture content of the mixed material to obtain predicted classification indices of raw pellets over multiple forecast periods. The rotation speed of the pelletizer, the inclination angle of the pelletizer plate, the amount of material supplied and the amount of supplied water are optimized using a moving average optimization model based on the predicted classification index of the raw pellets in each forecast period and the specified target classification index of the raw pellets in each forecast period for control, in the mode real time, the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of feed material supplied to the pelletizer, and the amount of feed water supplied to the pelletizer. In this way, the actual classification index of the raw pellets can meet the predetermined standard, and thus the pelletizing quality of the pelletizer can be improved.

BRIEF DESCRIPTION OF THE DRAWINGS

[00104] FIG. 1 is a schematic block diagram of a pelletizing process according to the prior art;

[00105] FIG. 2 is a schematic block diagram of a pelletizer control system according to an embodiment of the present application;

[00106] FIG. 3 is a schematic diagram of an operating block diagram of a pelletizer control system according to an embodiment of the present application;

[00107] FIG. 4 is a flowchart corresponding to a method for predicting the classification index of green pellets according to an embodiment of the present application;

[00108] FIG. 5 is a flowchart corresponding to a method for creating a percentage prediction model according to an embodiment of the present application;

[00109] FIG. 6 is a schematic diagram of a particle size analysis operating flowchart of a pelletizer control system according to an embodiment of the present application; And

[00110] in FIG. 7 is a block diagram corresponding to a control method for a pelletizer according to an embodiment of the present application as an example.

DETAILED DESCRIPTION OF IMPLEMENTATION OPTIONS

[00111] To explain the objectives, technical solutions and advantages of the present application, embodiments of the present application are further described in detail with reference to the accompanying drawings.

[00112] Referring to FIG. 1, FIG. 1 is a schematic block diagram of a pelletizing process according to the prior art.

[00113] As shown in FIG. 1, the pelletizing process includes a disc pelletizer 11, a water supply device 21 and a belt weigher 31. The disc pelletizer 11 includes a drive 111, a central shaft 112, a base 113, a disc 114, a scraper frame 115, a scraper 116 and a raw pellet conveyor belt 117 The water supply device 21 includes a water valve 211, a water supply pipe 212 and a water outlet 213. The belt weigher 31 includes a material supply valve 311, a feed hopper 312 and a feed conveyor belt 313.

[00114] In the above structure, the central shaft 112 allows the angle of the plate 114 to be adjusted; regulation of the speed of rotation of the drive 111 allows you to change the speed of rotation of the plate 114; the scraper frame 115 is configured to rest on the plate 114; and the scraper 116 is configured to push the mixed material onto the tray 114 to move and prevent sticking.

[00115] The mixed material from the feed hopper 312 is transported to the tray 114 by the feed conveyor belt 313, and the flow rate (i.e., the amount of material supplied) of the mixed material can be adjusted by the material supply valve 311.

[00116] Water from the water supply pipe 212 may be discharged, through the water outlet 213, at a position in which the mixed material falls on the tray 114; or may be discharged, through water outlet 213, to an area where raw pellets are formed on tray 114. The volume of water can be adjusted by the water valve 211. After leaving the tray 114, the raw pellet enters the green pellet conveyor belt 117, and the position of the raw pellet conveyor belt 117 can be considered as the pellet dispensing area of the plate pelletizer 11.

[00117] It should be noted that the disc pelletizer shown in FIG. 1 can be replaced by a drum pelletizer, which is not limited in this application.

[00118] Based on FIG. 1, in an embodiment of the present application, a control system for a pelletizer is presented. FIG. 2 is a schematic block diagram of a pelletizer control system according to an embodiment of the present application. As shown in FIG. 2, the system mainly includes a pelletizer 11, a rotation speed controller 12, an inclination angle controller 13, a water supply device 21, a water supply controller 22, a belt weigher 31, a material supply controller 32 and a central processing unit 5. Rotation speed controller 12 and the inclination angle controller are respectively connected to the pelletizer 11. The water supply controller 22 is connected to the water supply device 21. The material supply controller 32 is connected to the belt weigher 31. The central processor 5 is connected to the rotation speed controller 12, the inclination angle controller 13, water supply controller 22 and material supply controller 32, respectively.

[00119] In particular, the water outlet point of the water supply device 21 is located at the feed point of the pelletizer 11 and in the pelletizing area of the pelletizer, and is configured to supply water to the pelletizer 11.

[00120] The belt weigher 31 is configured to supply mixed material to the pelletizer. The material drop point of the belt weigher 31 is the feed point of the pelletizer 11, and is configured to supply mixed material to the pelletizer 11.

[00121] In the running process of the control system, various devices interact with each other, thereby realizing control of the pelletizer. In particular, referring to FIG. 3, in FIG. 3 shows, as an example, a schematic diagram of an operating block diagram of a pelletizer control system according to an embodiment of the present application.

[00122] The rotation speed controller 12 may be configured to implement the following steps S301 and S302.

[00123] Step S301. Determination of the rotation speed of the pelletizer in the current period.

[00124] During the process of determining the rotation speed of the pelletizer, if the detected rotation speed is stable, the stable rotation speed is stored and used as determination data in the next process. If the detected rotation speed changes, the stored rotation speed is updated in real time.

[00125] In an embodiment of the present application, the rotation speed of the pelletizer can be measured using measurement methods such as optical reflection method, magnetoelectric method, grating method or Hall effect switch detection method.

[00126] Step S302. Sending the rotation speed of the pelletizer to the central processor.

[00127] The inclination angle controller 13 may be configured to perform the following steps S303 and S304.

[00128] Step S303. Determination of the tilt angle of the pelletizer plate in the current period.

[00129] The inclination angle of the pelletizer tray is generally provided by a hydraulic inclination angle adjusting device, and the inclination angle is directly detected by an inclination angle sensor.

[00130] Step S304. Sending the tilt angle of the pelletizer plate to the central processor.

[00131] The water supply controller 22 may be configured to perform the following steps S305 and S306.

[00132] Step S305. Determination of the amount of supplied water supplied by the device for supplying water to the pelletizer in the current period.

[00133] Step S306. Sends the amount of water supplied to the central processor.

[00134] The material supply controller 32 may be configured to perform the following steps S307 and S308.

[00135] Step S307. Determination of the type and proportion of each component in the mixed material supplied by the belt weigher to the pelletizer in the current period, the amount of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material.

[00136] There are many methods for determining the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material. A possible determination method is to determine the type and proportion of each component, the amount of mixed material supplied, the percentage of binder in the mixed material, and the initial moisture content of the mixed material once at a given time interval, and replace the previously obtained parameters of the mixed material with the newly obtained parameters of the mixed material.

[00137] As used herein, it should be noted that the following three terms referred to herein have the same meaning: the type and proportion of each component, the type and proportion of each component of the mixed material, and the type and proportion of each component in the mixed material.

[00138] Another possible determination method is to determine various parameters of the mixed material in real time, and determine the degree of change of each parameter of the mixed material separately. If the degree of change is relatively small and within the specified error range, the previously obtained mixed material parameters are still used. If the degree of change is relatively high and outside the specified error range, the previously obtained mixed material parameters are replaced with the newly obtained mixed material parameters. It should be noted that the parameters of the mixed material mentioned herein refer to the type and proportion of each component, the amount of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material.

[00139] Step S308. Sends the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to the central processor.

[00140] The central processing unit 5 may be configured to carry out the following steps S309-S312

[00141] Step S309. Obtaining the rotation speed of the pelletizer sent by the rotation speed controller, the tilt angle of the pelletizer plate sent by the tilt angle controller, the amount of supplied water sent by the water supply controller, and the type and proportion of each component in the mixed material, the amount of fed mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material sent by the material feed controller.

[00142] Step S310. Predicting the classification index of green pellets based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material and the initial moisture content of the mixed material, to obtain the predicted classification index raw pellet indices over multiple forecast periods.

[00143] It should be noted that predicting the classification index of raw pellets and predicting the percentage of each fraction of raw pellet particles have the same meaning.

[00144] In a particular implementation process, the classification index of the raw pellets can be predicted using a percentage prediction model. Reference may be made to FIG. 4, and in FIG. 4 is a flowchart corresponding to a method for predicting the classification index of green pellets according to an embodiment of the present application as an example. Specifically, the following steps are included.

[00145] Step S401. Quantify the rotation speed of the pelletizer, the tilt angle of the pelletizer plate, the amount of water supplied and the amount of mixed material supplied in the same interval based on the corresponding scaling factors, to obtain the characteristic vector affecting pelletization.

[00146] It should be noted that the inclination angle of the pelletizer tray and the inclination angle of the pelletizer tray mentioned in this application have the same meaning. The supply amount and the mixed material supply amount mentioned in this application have the same meaning.

[00147] Data volumes and data types of the amount of feed material, the proportion of each component in the mixed material, the percentage of binder, the initial moisture content of the mixed material, the amount of water supplied, the rotation speed of the pelletizer and the angle of inclination of the pelletizer plate are different at the same time. Therefore, operations cannot be performed directly on these data volumes. In this embodiment of the present application, considering that the proportion of each component, the percentage of binder and the initial moisture content of the mixed material are proportions expressed as percentages, that is, the values are within the interval (0, 1), the current rotation speed of the pelletizer, the current the amount of water supplied and the current amount of mixed material supplied can also be quantified in the interval (0, 1). Thus, operations can be performed on data volumes of the proportion of each component, the percentage of binder, the initial moisture content of the mixed material, the amount of material supplied, the amount of water supplied, the rotation speed of the pelletizer and the angle of inclination of the pelletizer tray.

[00148] In particular, when quantifying the rotation speed of the pelletizer (or the angle of inclination of the pelletizer tray), the corresponding scaling factor may be represented by the maximum rotation speed of the pelletizer (or the maximum angle of inclination of the pelletizer tray). In particular, to calculate the ratio of the rotation speed of the pelletizer to the maximum rotation speed of the pelletizer, reference can be made to formula (1):

[00149]

[00150] where Norm(n) is the quantified rotation speed of the pelletizer, n is the rotation speed of the pelletizer, and n is the maximum rotation speed of the pelletizer. According to another embodiment, Norm(n) is a quantified pelletizer tray angle, n is a pelletizer tray angle, and n is a maximum pelletizer tray angle.

[00151] When quantifying the water supply amount, the corresponding scaling factor can be represented by the maximum water supply amount of the water supply pipe in the water supply device. In particular, to calculate the ratio of the amount of supplied water to the maximum amount of supplied water, reference may be made to formula (2):

[00152]

[00153] where Norm(s) represents the quantity of supplied water, s represents the quantity of supplied water, and s _max represents the maximum amount of supplied water.

[00154] It should be noted that the water supply system and the water supply device mentioned in this application have the same meaning.

[00155] When quantifying the feed quantity of the mixed material, the corresponding scaling factor represents the maximum feed quantity of the belt weigher. In particular, to calculate the ratio of the amount of mixed material supplied to the maximum amount of supplied material of the feed hopper, reference may be made to formula (3):

[00156]

[00157] where Norm(m) represents the quantified mixed material feed amount, m represents the mixed material feed amount, and m _max represents the maximum feed amount of the belt weigher.

[00158] It should be noted that the feeding belt and the supply belt mentioned in this application have the same meaning. [00159] Once quantified, the characteristic vector influencing the key factors of pelletization is as follows:

[00160] X (k)=(N, x ₁ (k) _, x ₂ (k), x ₃ (k), x ₄ (k), x ₅ (k), x ₆ (k), x ₇ ( k))=Norm(n, s, m, w, r, p, q)

[00161] where X (k) is a characteristic vector influencing the key factors of pelletization; x ₁ (k), x ₂ (k), x ₃ (k), x ₄ (k), x ₅ (k), x ₆ (k), and x ₇ (k), respectively, represent the quantified speed rotation of the pelletizer, the quantified angle of inclination of the pelletizer plate, the quantified amount of water supplied, the quantified amount of mixed material supplied, the proportion of each component in the mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material; N represents the number of component types in the mixed material. x5(k) contains information about the type and proportion of each component. In practical applications, each component may be numbered. In x5(k), the component number corresponds to the component's share.

[00162] For example, if there are three types of components in a mixed material, respectively numbered 1, 2 and 3, the characteristic vector influencing the key factors of pelletization can be represented as follows:

[00163]

[00164] where the first number 3 indicates that the mixed material contains three other components in addition to the binder; x ₁ (k) represents the quantified current rotation speed of the pelletizer; x ₂ (k) represents the quantity of water supplied; x ₃ (k) represents the quantified amount of mixed material supplied; x ₄ (k) represents the quantity of mixed material supplied; represents the proportion of component number 1 in the mixed material; represents the proportion of component number 2 in the mixed material; represents the proportion of component number 3 in the mixed material; x ₆ (k) represents the percentage of binder in the mixed material; and x ₇ (k) represents the initial moisture content of the mixed material.

[00165] Step S402. Introducing the characteristic vector influencing pelletization into the percentage prediction model to obtain predicted classification indices of raw pellets in multiple forecast periods based on the dependence mapping specified in the percentage prediction model.

[00166] The predetermined dependence display is a dependence display between the characteristic vector influencing pelletization and the predicted classification indices of the green pellets in a plurality of forecast periods.

[00167] In this embodiment of the present application, the classification index of the green pellets can be determined based on the predicted percentage values of the green pellets of various parameters, and the predicted percentage values of the green pellets of various parameters include the predicted percentage of qualifying large pellets, the predicted percentage of substandard large pellets, the predicted percentage of qualifying medium pellets, the predicted percentage of qualifying small pellets, and the predicted percentage of substandard small pellets. That is, in this embodiment of the present application, the classification index of the raw pellets can be predicted by predicting the percentage of the raw pellets of various parameters.

[00168] Here, the dependence display may also be a dependence display between the characteristic vector influencing pelletization and the predicted percentage values of the green pellets of various parameters.

[00169] In particular, the relationship display includes the predicted percentage value of green pellets in the same period and in different steps. The whole pelletizing process of the pelletizer can be divided into many steps according to the actual technical requirements, for example, one step is ten seconds. Thus, the percentage prediction model produces the following predicted percentages of multiple steps based on the dependency mapping:

[00170] y(k|k), y(k|k+1), y(k|k+2), ..., and y(k|k+j)

[00171] where j represents the prediction step and k represents the specific prediction instant.

[00172] In an embodiment of the present application, the predicted percentage values of the raw pellets of various parameters include the predicted percentage of qualifying large pellets, the predicted percentage of substandard large pellets, the predicted percentage of qualifying medium pellets, the predicted percentage of qualifying small pellets, and the predicted percentage of substandard small pellets pellets.

[00173] According to the diameter of the green pellets, the green pellets can be grouped according to five parameters, i.e. standard large pellets, substandard large pellets, standard medium pellets, standard small pellets, substandard small pellets. As shown in Table 1, a number of examples of percentage ranges of green pellets of various parameters that meet the requirements for the pelletizing process are presented. When the diameter of the green pellet (denoted as d in Table 1) is greater than or equal to 5 mm and less than 8 mm, the green pellet is a substandard fine pellet. When the diameter of the green pellet is greater than or equal to 8 mm and less than 11 mm, the green pellet is a qualified fine pellet. When the diameter of the green pellet is greater than or equal to 11 mm and less than 14 mm, the green pellet is a qualified medium pellet. When the diameter of the green pellet is greater than or equal to 14 mm and less than 16 mm, the green pellet is a qualified large pellet. When the diameter of the green pellet is greater than or equal to 16 mm, the green pellet is a substandard large pellet.

[00174] The percentage target range of substandard fine pellets is less than or equal to 7%; the percentage specified range of qualified small pellets is less than or equal to 30%; the percentage specified range of conditioned average pellets is less than or equal to 55%; the percentage specified range of qualified large pellets is less than or equal to 20%; and the percentage target range of substandard large pellets is less than or equal to 15%.

[00175] When the percentage of green pellets for each of the various parameters corresponds to the specified ranges in Table 1, and the sum of all percentages is 100%, the raw pellets can be considered to be of good quality. Conversely, the quality of the raw pellet can be considered to be inadequate.

[00176] In an embodiment of the present application, a percentage prediction model may be determined based on a prediction model using an LSTM neural network. As shown in FIG. 5, FIG. 5 is a flowchart corresponding to a method for creating a percentage prediction model according to an embodiment of the present application. Specifically, the following steps are included.

[00177] Step S501. Obtaining sample data in N historical forecasting periods.

[00178] The sample data in each historical forecast period includes the historical pelletizer rotation speed, the historical pelletizer tray angle, the historical amount of water supplied, the historical pelletizer feedstock information for producing sample raw pellets, and the measured percentage values of sample raw pellets of various parameters. Historical feedstock information includes the historical feed quantity, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material. The measured percentage values of sample raw pellets of various parameters are obtained by post-image calculation and analysis of sample raw pellets in each historical forecast period using a particle size analysis method.

[00179] In particular, the measured classification index of the sample raw pellets can be represented by the measured percentage value of the sample raw pellets of various parameters. There are many methods to obtain the measured percentage value of sample raw pellets of various parameters using the granulometric analysis method. For example, particle size analysis can be performed using a manual sampling method. As another example, granulometric analysis can be performed using a machine vision method.

[00180] The content of performing granulometric analysis using a machine vision method is specifically described below.

[00181] As shown in FIG. 2, the control system may further include an image acquisition device 41 and an image processing device 42. The image acquisition device 41 is located in the discharge opening of the pelletizer and is connected to the image processing device 42. The image processing device 42 is connected to the central processing unit 5.

[00182] During the granulometric analysis process of the control system, the image acquisition device 41 interacts with the image processing device 42 to obtain measured percentage values of sample raw pellets of various parameters. In particular, referring to FIG. 6, in FIG. 6 is an example of a schematic diagram of the operating flowchart of a granulometric analysis of a pelletizer control system according to an embodiment of the present application.

[00183] The image acquisition device 41 may be configured to perform the following steps S601 and S602.

[00184] Step S601. Obtaining image data of the pelletizer pellet outlet.

[00185] It should be noted that the pelletizer discharge port and the pelletizer pellet outlet referred to in this application have the same meaning.

[00186] Step S602. Send image data of the pellet outlet to the image processing device.

[00187] The image processing apparatus 42 may be configured to perform the following steps S603-S610.

[00188] Step S603. Perform image pre-processing using the pellet outlet image data to separate the sample raw pellet image data and the background image data.

[00189] Step S604. Obtaining the central bright spot of a sample raw pellet from image data of a sample raw pellet.

[00190] Step S605. Determining the outline of a sample green pellet based on the sample raw pellet image data and the background image data.

[00191] Step S606. Obtaining the particle size of the sample raw pellet based on the central bright spot of the sample raw pellet and the outline of the sample raw pellet.

[00192] Step S607. Determination of the sample raw pellet parameter based on the particle size of the sample raw pellet and a specified correspondence between the particle fraction and the raw pellet parameter.

[00193] Step S608. Calculation of the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters.

[00194] Step S609. Determination of the measured classification index of sample raw pellets based on the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters.

[00195] Step S610. Sending the measured classification index of selected raw pellets to the central processor.

[00196] Step S502. Quantify historical pelletizer rotation speed, historical pelletizer tray angle, historical water feed quantity, and historical mixed material feed quantity in the same interval based on appropriate scaling factors.

[00197] It should be noted that step S502 in FIG. 5 in relation to the historical drive rotation speed, the historical quantity of water supplied and the historical quantity of mixed material supplied, the quantification of the historical rotation speed of the pelletizer, the historical tilt angle of the pelletizer plate, the historical quantity of water supplied and the historical quantity of mixed material supplied in the same interval based on respective scaling factors has the same meaning as step S502 in describing the quantification of the historical rotation speed of the pelletizer, the historical tilt angle of the pelletizer plate, the historical water supply amount, and the historical mixed material supply amount in the same interval based on the respective scaling factors.

[00198] It should be noted that for a particular quantification method, reference may be made to the above description, and the detailed description will not be presented again herein.

[00199] Step S503. Obtaining characteristic vectors of N samples that influence pelletization, based on the quantified historical rotation speed of the pelletizer, the quantified historical inclination angle of the pelletizer plate, the quantified historical amount of supplied water, the quantified amount of supplied historical mixed material, the type and share of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the initial moisture content of the mixed material in the historical mixed material.

[00200] It should be noted that for a particular quantification method, reference may be made to the above description, and the detailed description will not be presented again herein.

[00201] Step S504. Using the characteristic vectors of N samples influencing pelletization as the input of the forecasting model, and using the actual percentage values of sample raw pellets of various parameters in N historical forecasting periods as the output of the forecasting model, to train the percentage forecasting model using the inverse method error propagation.

[00202] The percentage prediction learning module trains the LSTM neural network model using backpropagation based on the training set input data and the training set output data. Backpropagation is a learning algorithm suitable for a multilayer neural network that causes the multilayer neural network to iterate through propagation and modification of weights until the response (output) to the input falls within a given target range.

[00203] Step S505. Continuously update the weight parameter, bias parameter, and learning coefficient of the percentage prediction model through iterative learning.

[00204] Step S506. If the difference between the predicted percentage value of the percentage prediction model for sample raw pellets of various parameters and the actual percentage value of sample raw pellets of various parameters is within the specified range of acceptable values, or if the percentage prediction model reaches the specified maximum number of iterations during the iteration operation, end training and storing the weight parameter, bias parameter, and learning coefficient updated most recently. Thus, a dynamic display of the relationship between the characteristic vector influencing pelletization and the predicted percentage values of raw pellets of various parameters or a dynamic prediction model based on the LSTM neural network is obtained.

[00205] During this process, the forward signal transmission at time k (that is, the LSTM output signal at time k) is expressed as follows:

[00206]

[00207] where Y _k-1 is the output signal at time k-1, X _k is the input vector at time k, σ is a sigmoid function, W _f and b _f are the weight vector and the forgetting gate bias component, W _i and b _i represent the weight vector and bias component of the input gate, W _c and b _c represent the weight vector and bias component of the neuron state, W _o and b _o represent the weight vector and bias component of the output gate, c _k represents the current state , and c _k-1 represents the state at the previous moment.

[00208] Considering device wear, transfer of operating conditions, and changes in the determination point in the actual process, the characteristic vector affecting pelletization may make the percentage prediction model inapplicable. To increase the usable range and accuracy of the prediction model, an embodiment of the present application further provides a method for updating a percentage prediction model in an always-on Internet connection.

[00209] In particular, the parameters of the model are adjusted accordingly based on the deviation in the prediction process when using the training model in an always-on Internet mode.

[00210] With respect to the vector influencing pelletization, in an embodiment of the present application, the granulometric index of the model is mainly used to determine whether there is a need to adjust the model and how to adjust the model. Generally, the root mean square error of the predicted value and the measured value can be used as the granulometric index, and then, according to the statistical distribution law of the granulometric index, the statistical confidence limit is preliminarily set to determine the need to initiate updating and the required updating method.

[00211] The recurrent method for estimating model parameters is selected if, based on the result of the index analysis, it is determined that the process characteristic is gradually changing. The prediction model is updated using recurrent identification in a moving window, and the steps are presented as follows.

[00212] It is determined that the sample data set of the original prediction model is S={[X ¹ Y ¹ ],[X ^t ,Y ^t ]}, and t represents the total number of samples. When a new measurement value [X ^m , Y ^m ] is received, it is added to the sample data set and the oldest sample is discarded. The new sample data set is:

[00213] S={[X ² ,Y ² ], ...,[X ^t ,Y ^t ],[X ^m ,Y ^m ]}.

[00214] Next, the learning algorithm is trained using the new sample to obtain a new prediction model.

[00215] The flash learning method is selected if, based on the result of the index analysis, it is determined that the process characteristic changes abruptly. Data samples similar to the current measurement state in the historical sample data are selected to transform the forecasting model.

[00216] Step S311. Calculation of the deviation value of the raw pellet classification index in each forecast period based on the predicted raw pellet classification index in each forecast period and the specified target raw pellet classification index in each forecast period.

[00217] Specifically, based on the predicted percentage value of each type of raw pellets with different prediction steps in each forecast period and a given reference (target) percentage value of each type of raw pellets with different steps in one forecast period, the deviation value between the percentage of each type of raw pellets with different steps in the period and reference value.

[00218] The deviation value between the percentage of each type of green pellets at different steps in each forecast period and the reference value can be obtained according to formula (4):

[00219]

[00220] In formula (4), E(k|k+j) represents the deviation value between the percentage of each type of green pellets with the ^jth step in the ^kth forecast period and the reference value; r _i (j) represents the target percentage value of raw pellets of the i- ^th parameter with the j ^-th step; y _i (k|k+j) represents the predicted percentage value of raw pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; j=1, 2, …, m, where m is an integer greater than or equal to 1.

[00221] Step S312. Inputting the deviation values of the classification indexes of raw pellets in multiple forecast periods into the moving average optimization model to obtain the pelletizer rotation speed to be adjusted, the pelletizer tray angle to be adjusted, the material feed amount to be adjusted, and the water feed amount to be adjusted, controlling the rotation speed controller to regulate the speed. rotation of the pelletizer as the rotation speed of the pelletizer to be adjusted, control of the inclination angle controller to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted, control of the material supply controller to regulate the amount of feed material supplied to the pelletizer as the amount of feed material to be adjusted , and controlling the water supply controller to regulate the amount of supply water supplied to the pelletizer as the supply amount of water to be controlled.

[00222] It should be noted that the supply amount to be controlled and the supply amount to be adjusted mentioned in this application have the same meaning.

[00223] Specifically, the deviation values between the percentages of different types of green pellets with multiple steps in one forecast period and the reference value are input into the moving average optimization model to obtain the pelletizer rotation speed to be controlled, the pelletizer plate angle to be controlled, the feed quantity to be controlled. material and the amount of water supplied to be regulated. The rotation speed controller is controlled to regulate the rotation speed of the pelletizer as the rotation speed of the pelletizer to be controlled. The inclination angle controller is controlled to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted. The material supply controller is controlled to control the amount of supply material supplied to the pelletizer as the supply amount to be controlled. The water supply controller is controlled to control the amount of supply water supplied to the pelletizer as the supply amount of water to be controlled.

[00224] The moving average optimization model is configured to calculate the corresponding pelletizer rotation speed, pelletizer plate inclination angle, amount of material supplied and amount of water supplied when the standard deviation between the percentage of each type of raw pellets in one forecast period and the reference value is minimal, when provided that the type and proportion of component in the mixed material, the percentage of binder and the initial moisture content of the mixed material remain unchanged.

[00225] The deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecast period and the reference value can be obtained according to formula (5):

[00226]

[00227] where error value of the deviation between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _i (j) represents the target percentage value of small pellets of the i- ^th parameter with the j ^-th prediction step; y _i (k|k+j) represents the predicted percentage value of small pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; j=0, 1, 2, …, m, where m is an integer that is greater than or equal to 0; and k=1, 2, …, n, where n is an integer that is greater than or equal to 1.

[00228] It should be noted that in the above-mentioned embodiments, the disc pelletizer is given as an example only. Undoubtedly, the raw pellet quality prediction system presented in the embodiment of the present application is also applicable to the drum pelletizer.

[00229] In embodiments of the present application, the classification indices of the green pellets are predicted based on the rotational speed of the pelletizer, the angle of inclination of the pelletizer tray, the amount of water supplied, the amount of mixed material supplied, the type and proportion of each component in the mixed material, the percentage of binder in the mixed material, and initial moisture content of the mixed material to obtain predicted classification indices of raw pellets in multiple forecast periods. The rotation speed of the pelletizer, the tilt angle of the pelletizer plate, the amount of material supplied and the amount of water supplied are optimized using a moving average optimization model based on the predicted classification index of the raw pellets in each forecast period and the specified target classification index of the raw pellets in each forecast period, for control, in in real time, the speed of rotation of the pelletizer, the angle of inclination of the pelletizer plate, the amount of feed material supplied to the pelletizer, and the amount of feed water supplied to the pelletizer. In this way, the actual classification index of the raw pellets can meet the predetermined standard, and thus the pelletizing quality of the pelletizer can be improved.

[00230] Below is an embodiment of the method according to the present application. For details not disclosed in the embodiment of the method according to the present application, reference may be made to an embodiment of the system according to the present application.

[00231] In FIG. 7 is a block diagram corresponding to a control method for a pelletizer according to an embodiment of the present application as an example. As shown in FIG. 7, the method may include the following steps.

[00232] Step S701. Predicting the percentage of each fraction of green pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material and the initial moisture content of the mixed material to obtain predicted classification indices of raw pellets in multiple forecast periods.

[00233] Step S702. Calculation of the deviation value of the raw pellet classification index in each forecast period based on the predicted raw pellet classification index in each forecast period and the specified target raw pellet classification index in each forecast period.

[00234] Step S703. Introducing the deviation values of the classification indices of the raw pellets in multiple forecast periods into the moving average optimization model to obtain the controllable rotation speed of the pelletizer, the controllable angle of the pelletizer plate, the controllable amount of supplied material, and the controllable amount of supplied water.

[00235] The rotation speed controller is controlled to regulate the rotation speed of the pelletizer as the rotation speed of the pelletizer to be controlled. The inclination angle controller is controlled to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted. The material supply controller is controlled to control the amount of supply material supplied to the pelletizer as the supply amount to be controlled. The water supply controller is controlled to control the amount of supply water supplied to the pelletizer as the supply amount of water to be controlled.

[00236] The plurality of forecast periods includes the current period and periods following the current period; and the moving average optimization model is configured to calculate the corresponding rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material and the amount of supplied water when the standard deviation of the classification indices of raw pellets in multiple forecast periods is minimal, provided that the type and proportion of each component in a mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material remain unchanged.

[00237] If necessary, the classification index of the green pellets is determined based on the predicted percentage values of the green pellets of various parameters, and the predicted percentage values of the green pellets of various parameters include the predicted percentage of qualifying large pellets, the predicted percentage of substandard large pellets, the predicted percentage of qualifying medium pellets , the predicted percentage of qualifying small pellets, as well as the predicted percentage of substandard small pellets.

[00238] The deviation value of the classification index of raw pellets in each forecast period is obtained according to the following formula:

[00239] E(k|k+j)={(r ₁ (j)-y ₁ (k|k+j)), r ₂ (j)-y ₂ (k|k+j), …, r _i (j)-y _i (k|k+j)}

[00240] where E(k|k+j) is the deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecast period and the reference value; r _i (j) represents the target percentage value of raw pellets of the i- ^th parameter with the j ^-th step; y _i (k|k+j) represents the predicted percentage value of raw pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; and j=1, 2, …, m, where m is an integer that is greater than or equal to 1.

[00241] If necessary, the standard deviation of the classification indices of raw pellets in multiple forecast periods is obtained according to the following formula:

[00242]

[00243] where σ (k|k+j) represents the root mean square error of the value

deviations between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _i (j) represents the target percentage value of small pellets of the i- ^th parameter with the j ^-th prediction step; y _i (k|k+j) represents the predicted percentage value of small pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; j=0, 1, 2, …, m, where m is an integer that is greater than or equal to 0; and k=1, 2, …, n, where n is an integer that is greater than or equal to 1.

[00244] Optionally, predicting the percentage of each fraction of green pellet particles based on the rotation speed of the pelletizer, the angle of the pelletizer tray, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to obtain predicted classification indices of raw pellets in multiple forecast periods, in particular, is carried out according to the following steps:

[00245] quantifying the rotation speed of the pelletizer, the tilt angle of the pelletizer plate, the amount of water supplied and the amount of material supplied in the same interval based on the corresponding scaling factors;

[00246] obtaining a characteristic vector influencing pelletization based on the quantified rotation speed of the pelletizer, the quantified tilt angle of the pelletizer plate, the quantified amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the mixed material, the percentage binder in the mixed material and the initial moisture content of the mixed material; And

[00247] introducing a pelletizing characteristic vector into a percentage prediction model to obtain predicted percentage values of green pellets of various parameters, wherein the percentage prediction model includes displaying a relationship between the pelletizing characteristic vector and predicted percentages of raw pellets of various parameters.

[00248] Optionally, the scaling factor corresponding to the rotation speed of the pelletizer is the maximum rotation speed of the pelletizer.

[00249] The scaling factor corresponding to the water supply amount represents the maximum water supply amount of the water supply pipe in the water supply system.

[00250] The scaling factor corresponding to the feed amount is the maximum feed amount of the feed belt.

[00251] The scaling factor corresponding to the tilt angle of the pelletizer tray represents the maximum tilt angle of the pelletizer.

[00252] If necessary, a percentage prediction model is obtained as follows:

[00253] obtaining sampling data in N historical forecasting periods, wherein the sampling data in each historical forecasting period includes historical pelletizer rotation speed, historical pelletizer tray angle, historical amount of water supplied, historical pelletizer raw material information for obtaining sample raw pellets, and also measured percentage values of selected raw pellets of various parameters; the historical raw material information includes the historical quantity of feed material, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material; the measured percentage values of sample raw pellets of various parameters are obtained through analysis and calculation after obtaining and processing images of sample raw pellets in each historical forecast period using a machine vision method;

[00254] quantifying the historical pelletizer rotation speed, the historical pelletizer tray angle, the historical feed amount of water, and the historical feed amount of material in the same interval based on appropriate scaling factors;

[00255] obtaining characteristic vectors of N samples influencing pelletization based on the quantified historical rotation speed of the pelletizer, the quantified historical tilt angle of the pelletizer plate, the quantified historical amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the historical mixed material, the percentage of binder in the historical mixed material, and the original moisture content of the mixed material in the historical mixed material;

[00256] using the characteristic vectors of the N samples influencing pelletization as the input of the prediction model, and using the actual percentage values of the sample raw pellets of various parameters in the N historical prediction periods as the output of the prediction model, to train the percentage prediction model with using the backpropagation method;

[00257] continuously updating the weight parameter, bias parameter, and learning coefficient of the percentage prediction model through iterative learning; And

[00258] if the difference between the predicted percentage value of the percentage prediction model for sample raw pellets of various parameters and the actual percentage value of sample raw pellets of various parameters is within a specified range of acceptable values, or if the percentage prediction model reaches a specified maximum number of iterations during an iteration operation , completes the training and saves the weight parameter, bias parameter, and training coefficient that were most recently updated.

[00259] If necessary, a percentage prediction model is determined based on a prediction model using a long short-term memory LSTM neural network.

[00260] If necessary, the system further includes an image acquisition device and an image processing device. The image acquisition device is located in the discharge opening of the pelletizer and is connected to the image processing device. The image processing device is connected to the central processing unit.

[00261] The image acquisition device is configured to perform the following step: obtaining image data of the pellet outlet outlet, and sending image data of the pellet outlet to the image processing device.

[00262] The image processing device is configured to perform the following steps:

[00263] performing image pre-processing using the pellet outlet image data to separate the sample raw pellet image data and the background image data;

[00264] obtaining a central bright spot of the sample raw pellet based on the image data of the sample raw pellet;

[00265] determining the outline of the sample raw pellet based on the sample raw pellet image data and the background image data;

[00266] obtaining the particle size of the sample raw pellet based on the central bright spot of the sample raw pellet and the outline of the sample raw pellet;

[00267] determining a parameter of the sample raw pellet based on the particle size of the sample raw pellet and a predetermined correspondence between the particle fraction and the parameter of the raw pellet;

[00268] calculating the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters; And

[00269] determining a measured classification index of the sample raw pellets based on the total number of sample raw pellets in historical periods and the number of sample raw pellets of various parameters, and sending the measured classification index of the sample raw pellets to the central processing unit.

[00270] If necessary, the rotation speed controller is configured to perform the following step:

[00271] determining the rotation speed of the pelletizer in the current period, and sending the rotation speed of the pelletizer to the central processor.

[00272] The tilt angle controller is configured to perform the following step:

[00273] determining the inclination angle of the pelletizer tray in the current period, and sending the inclination angle of the pelletizer tray to the central processor.

[00274] The water supply controller is configured to perform the following step:

[00275] determining the amount of supply water supplied by the water supply device to the pelletizer in the current period, and sending the amount of supply water to the central processor.

[00276] The material supply controller is configured to perform the following step:

[00277] determining the type and proportion of each component in the mixed material supplied by the belt weigher to the pelletizer in the current period, the amount of mixed material fed, the percentage of binder in the mixed material and the initial moisture content of the mixed material, and sending the type and proportion of each component in the mixed material material, the amount of mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to the central processor.

[00278] In an embodiment of the present application, the rotation speed of the pelletizer, the angle of the pelletizer tray, the amount of feed material and the amount of feed water are optimized using a moving average optimization model to control, in real time, the rotation speed of the pelletizer, the angle of the pelletizer tray, the amount of feed material supplied to the pelletizer, and the amount of feed water supplied to the pelletizer. In this way, the actual classification index of the raw pellets can meet the predetermined standard, and thus the pelletizing quality of the pelletizer can be improved.

[00279] For like parts, references may be made to each other between the embodiments presented herein. The specific embodiments described above are only some examples in accordance with the general concept of this application and do not constitute any limitation on the scope of legal protection of this application. To one skilled in the art, any other embodiments obtained in accordance with the solutions of this application without the exertion of effective efforts will fall within the scope of protection of this application.

Claims (50)

1. A control system for a pelletizer containing a device for supplying water and a belt weigher, in which the water outlet point of the water supply device is located at the feed point of the pelletizer and in the area of pellet formation in the pelletizer, and is configured to supply water to the pelletizer; the belt weigher is configured to supply mixed material to the pelletizer; the material falling point of the belt weigher is the feed point of the pelletizer; characterized in that the system includes a rotation speed controller connected to the pelletizer, an inclination angle controller connected to the pelletizer, a water supply controller connected to the water supply device, a material supply controller connected to the belt weigher, and a central processing unit respectively connected to a rotation speed controller, an inclination angle controller, a water supply controller and a material supply controller, wherein the central processor is configured to perform the following steps: receiving the rotation speed of the pelletizer sent by the rotation speed controller, the inclination angle of the pelletizer plate sent by the inclination angle controller, the amount of supplied water sent by the water supply controller, and the type and proportion of each component, the amount of mixed material supplied, the percentage of binder in the mixed material and the feed moisture content of the mixed material sent by the material feed controller; predicting the percentage of each fraction of raw pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material to obtain predicted classification indices of raw pellets in multiple forecast periods; calculating the deviation value of the classification index of the raw pellets in each forecasting period based on the predicted classification index of the raw pellets in each forecasting period and a given target classification index of the raw pellets in each forecasting period; introducing the deviation values of the classification indexes of raw pellets in multiple forecast periods into the moving average optimization model to obtain the pelletizer rotation speed to be adjusted, the pelletizer plate inclination angle to be adjusted, the amount of feed material to be adjusted, and the amount of water to be supplied to be adjusted, controlling the rotation speed controller to regulate the speed rotation of the pelletizer as the rotation speed of the pelletizer to be adjusted, control of the inclination angle controller to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted, control of the material supply controller to regulate the amount of feed material supplied to the pelletizer as the amount of feed material to be adjusted , and controlling the water supply controller to regulate the amount of supply water supplied to the pelletizer as the supply amount to be controlled, the plurality of prediction periods including a current period and periods following the current period; and the moving average optimization model is configured to calculate the corresponding rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material and the amount of supplied water when the standard deviation of the classification indices of raw pellets in multiple forecast periods is minimal, provided that the type and proportion of each component in a mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material remain unchanged. 2. The control system according to claim 1, characterized in that the classification index of raw pellets is determined based on the predicted percentage values of raw pellets of various parameters, and the predicted percentage values of raw pellets of various parameters include the predicted percentage value of standard large pellets, the predicted percentage value of substandard large pellets , the predicted percentage of qualifying medium pellets, the predicted percentage of qualifying small pellets, and the predicted percentage of substandard small pellets; And the deviation value of the classification index of raw pellets in each forecasting period is obtained according to the following formula: , where E(k|k+j) represents the deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _i (j) represents the target percentage value of raw pellets of the i- ^th parameter with the j ^-th step; y _i (k|k+j) represents the predicted percentage value of raw pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=1, 2, …, d, where d is an integer that is greater than or equal to 1; and j=1, 2, …, m, where m is an integer that is greater than or equal to 1. 3. The control system according to claim 2, characterized in that the standard deviation of the classification indices of raw pellets in multiple forecast periods is obtained according to the following formula: , where σ (k|k+j) represents the root-mean-square error of the deviation value between the percentage of each type of raw pellets with the ^jth step in the ^kth forecasting period and the reference value; r _i (j) represents the target percentage value of small pellets of the i- ^th parameter with the j ^-th prediction step; y _i (k|k+j) represents the predicted percentage value of small pellets of the i ^-th parameter with the j ^-th step in the k ^-th forecasting period; i=l, 2, …, d, where d is an integer that is greater than or equal to 1; j=0, 1, 2, …, m, where m is an integer that is greater than or equal to 0; and k=1, 2, …, n, where n is an integer that is greater than or equal to 1. 4. The control system according to claim 1, characterized in that the prediction of the percentage content of each fraction of raw pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied water, the type and proportion of each component in the mixed material, the amount of supplied mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material to obtain predicted classification indices of green pellets in multiple forecast periods, in particular, is carried out according to the following steps: quantitative assessment of the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied and the amount of material supplied in the same interval based on the corresponding scaling factors; obtaining a characteristic vector that influences pelletizing, based on the quantitatively expressed rotation speed of the pelletizer, the quantified angle of inclination of the pelletizer plate, the quantified amount of supplied water, the quantified amount of supplied material, the type and proportion of each component in the mixed material, the percentage of binder in the mixed material material and initial moisture content of the mixed material; And introducing the pelletizing characteristic vector into the percentage prediction model to obtain the predicted percentage values of the green pellets of various parameters, wherein the percentage prediction model includes displaying the relationship between the pelletizing characteristic vector and the predicted percentage values of the green pellets of various parameters . 5. The control system according to claim 4, characterized in that the scaling factor corresponding to the rotation speed of the pelletizer is the maximum rotation speed of the pelletizer; the scaling factor corresponding to the supply water amount represents the maximum supply water amount of the water supply pipe in the water supply system; the scaling factor corresponding to the feeding quantity represents the maximum feeding quantity of the feeding belt; And The scaling factor corresponding to the tilt angle of the pelletizer tray is the maximum tilt angle of the pelletizer. 6. The control system according to claim 4, characterized in that the percentage forecasting model is obtained as follows: obtaining sample data in N previous forecast periods, where the sample data in each previous forecast period includes the previous pelletizer rotation speed, the previous pelletizer tray angle, the previous amount of water supplied, information about the previous pelletizer feedstock to produce sample raw pellets, and the measured percentages values of selected raw pellets of various parameters; the previous raw material information includes the previous feed amount, the type and proportion of each component in the previous mixed material, the percentage of binder in the previous mixed material, and the initial moisture content of the mixed material in the previous mixed material; the measured percentage values of the sample raw pellets of various parameters are obtained through analysis and calculation after obtaining and processing images of the sample raw pellets in each previous forecast period using a computer vision method; quantifying the previous rotation speed of the pelletizer, the previous tilt angle of the pelletizer plate, the previous quantity of water supplied and the previous quantity of material supplied in the same interval based on the corresponding scaling factors; obtaining characteristic vectors of N samples influencing pelletization based on the quantified previous rotation speed of the pelletizer, the quantified previous tilt angle of the pelletizer plate, the quantified previous amount of water supplied, the quantified amount of feed material, the type and proportion of each component in the previous mixed material , the percentage of binder in the previous mixed material and the initial moisture content of the mixed material in the previous mixed material; using the characteristic vectors of N samples influencing pelletization as the input of the prediction model, and using the actual percentage values of sample raw pellets of various parameters in the N previous forecast periods as the output of the prediction model to train the percentage prediction model using back propagation method errors; continuously updating the weight parameter, bias parameter and learning coefficient of the percentage prediction model through iterative learning; And If the difference between the predicted percentage value of the percentage prediction model for sample raw pellets of various parameters and the actual percentage value of sample raw pellets of various parameters is within the specified range of acceptable values, or if the percentage prediction model reaches the specified maximum number of iterations during the iteration operation, end training and storing the weight parameter, bias parameter, and learning coefficient updated most recently. 7. The control system according to claim 6, characterized in that the percentage forecasting model is determined based on a forecasting model using a long short-term memory LSTM neural network. 8. The control system according to claim 6, characterized in that the system further includes an image acquisition device and an image processing device, the image acquisition device is located in the discharge opening of the pelletizer and is connected to the image processing device, and the image processing device is connected to central processor; the image acquisition device is configured to perform the following step: obtaining image data of the pelletizing outlet, and sending image data of the pellet outlet to the image processing device; And The image processing device is configured to perform the following steps: performing image pre-processing using the pellet outlet image data to separate the sample raw pellet image data and the background image data; obtaining a central bright spot of the sample raw pellet based on the image data of the sample raw pellet; determining an outline of the sample raw pellet based on the image data of the sample raw pellet and the background image data; obtaining a particle size of the sample raw pellet based on a central bright spot of the sample raw pellet and an outline of the sample raw pellet; determining a parameter of the sample raw pellet based on the particle size of the sample raw pellet and a predetermined correspondence between the particle fraction and the parameter of the raw pellet; calculation of the total number of sample raw pellets in previous periods and the number of sample raw pellets of various parameters; And determining a measured classification index of the sample raw pellets based on the total number of sample raw pellets in previous periods and the number of sample raw pellets of various parameters, and sending the measured classification index of the sample raw pellets to the central processing unit. 9. Control system according to any one of paragraphs. 1-8, characterized in that the rotation speed controller is configured to carry out the following step: determining the rotation speed of the pelletizer in the current period, and sending the rotation speed of the pelletizer to the central processor; The inclination angle controller is configured to perform the following step: determining the inclination angle of the pelletizer tray in the current period, and sending the inclination angle of the pelletizer tray to the central processor; The water supply controller is configured to perform the following stage: determining the amount of supply water supplied by the device for supplying water to the pelletizer in the current period, and sending the amount of supply water to the central processor; And The material supply controller is configured to perform the following step: determining the type and proportion of each component in the mixed material supplied by the belt weigher to the pelletizer in the current period, the quantity of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material, and sending the type and proportion of each component in the mixed material, quantity of the mixed material fed, the percentage of binder in the mixed material, and the initial moisture content of the mixed material to the central processor. 10. A method for controlling a pelletizer, characterized in that the method includes: predicting the percentage of each fraction of raw pellet particles based on the rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of water supplied, the type and proportion of each component in the mixed material, the amount of mixed material supplied, the percentage of binder in the mixed material and the initial moisture content of the mixed material to obtain predicted classification indices of green pellets in multiple forecast periods; calculating the deviation value of the classification index of the raw pellets in each forecasting period based on the predicted classification index of the raw pellets in each forecasting period and a given target classification index of the raw pellets in each forecasting period; introducing the deviation values of the classification indexes of raw pellets in multiple forecast periods into the moving average optimization model to obtain the pelletizer rotation speed to be adjusted, the pelletizer plate inclination angle to be adjusted, the amount of feed material to be adjusted, and the amount of water to be supplied to be adjusted, controlling the rotation speed controller to regulate the speed rotation of the pelletizer as the rotation speed of the pelletizer to be adjusted, control of the inclination angle controller to adjust the inclination angle of the pelletizer tray as the inclination angle of the pelletizer tray to be adjusted, control of the material supply controller to regulate the amount of feed material supplied to the pelletizer as the amount of feed material to be adjusted , and controlling the water supply controller to regulate the amount of supply water supplied to the pelletizer as the amount of supply water to be controlled, wherein the plurality of forecasting periods include the current period and periods following the current period; and the moving average optimization model is configured to calculate the corresponding rotation speed of the pelletizer, the angle of inclination of the pelletizer plate, the amount of supplied material and the amount of supplied water when the standard deviation of the classification indices of raw pellets in multiple forecast periods is minimal, provided that the type and proportion of each component in a mixed material, the percentage of binder in the mixed material and the initial moisture content of the mixed material remain unchanged.