Sintering internal return reduction method and system based on cold screen optimization

Abstract

The invention discloses a sintering internal return reduction method and system based on cold screen optimization, and belongs to the technical field of sintering ore internal return rate reduction. The method comprises the steps that S1, cold screen equipment of a double-layer vibrating screen structure is built, and the cold screen equipment comprises an upper-layer screen mesh and a lower-layer screen mesh; s2, monitoring units are installed on the cold screening equipment and the sintering production line, and the monitoring units transmit monitoring data to a data processing unit; s3, data processing and model prediction; s4, the regulation and control unit dynamically regulates and controls cold screen parameters according to a model prediction result, wherein the cold screen parameters comprise vibration frequency, amplitude and ultrasonic vibration device power; s5, correcting and optimizing the sintering internal return prediction model; s3-S5 form a closed-loop flow of prediction, regulation and feedback correction, and the closed-loop flow can be executed circularly and is allowed to be executed in an overlapping manner in time. The synergistic effect of graded precise screening and effective anti-blocking is achieved, the false screening of qualified materials is reduced, the internal returning amount is remarkably reduced, and the wear resistance and the adaptability to different raw materials / different-scale production lines are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of reducing the internal return rate of sinter, and particularly relates to a sinter internal return reduction method and system based on cold screening optimization. BACKGROUND

[0002] Sintering is a metal ore pretreatment process that fuses small ore particles into larger blocky materials at high temperatures. In metallurgical production, sintering technology is a key process for preparing blocky sintered ore suitable for blast furnace ironmaking from raw materials such as iron ore and returned ore through mixing, granulation, ignition, sintering, cooling, etc.

[0003] In the sintering production process, internal return refers to the part of sintered ore after sintering and cooling that is returned to the sintering batching room to participate in sintering again after being screened by a cold screen before entering the finished product system. Excessive internal return will have many adverse effects on sintering production, such as increasing energy consumption, reducing the utilization coefficient of the sintering machine, and affecting the quality stability of sintered ore, etc.

[0004] Current common ways to reduce internal return include optimizing sintering process parameters and improving raw material quality, but there is insufficient optimization of the cold screening link. Traditional cold screening generally has low screening efficiency and easy plugging of screen holes, and the screen hole specifications and vibration parameters are difficult to adjust in real time with the working conditions, resulting in some qualified sintered ore being mis-screened as internal return, thereby increasing the internal return amount. In existing production, there are also problems such as single-layer screen, fixed screen holes, and lagging control, which together restrict the effective reduction of internal return. SUMMARY

[0005] The present application aims to provide a sinter internal return reduction method and system based on cold screening optimization to solve the technical problems of low screening efficiency and easy plugging of screen holes, difficulty in adjusting screen holes and vibration parameters in real time with working conditions, insufficient monitoring and data modeling means, lagging control and lack of closed-loop control, which cause qualified materials to be mis-screened as internal return, and high and fluctuating internal return amount.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solution: A sinter internal return reduction method based on cold screening optimization, comprising the following steps: S1, a cold screening equipment with a double-layer vibrating screen structure is built, the cold screening equipment includes an upper layer screen and a lower layer screen, the screen hole size and the inclination angle of the upper layer screen and the lower layer screen are respectively set, wherein the screen hole size of the upper layer screen is larger than that of the lower layer screen, and an ultrasonic vibration device is installed on the surface of the upper layer screen and the lower layer screen; S2, a monitoring unit is installed on the cold screening equipment and the sinter production line, the monitoring unit transmits monitoring data to a data processing unit in real time through an industrial Ethernet or a wireless communication module; S3, data processing and model prediction, including data transmission, data preprocessing, constructing prediction model and model training and verification, wherein the data preprocessing includes moving average filtering, feature extraction and eigenvalue decomposition based on covariance matrix, the prediction model is a machine learning based internal return material prediction model and outputs the prediction value and its trend; S4, the regulation unit dynamically adjusts the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening equipment according to the model prediction result by using a proportional-integral-derivative controller; S5, the adjusted actual internal return material quantity data is fed back to the data processing unit to correct and optimize the sintering internal return material prediction model; Wherein, S3-S5 constitute a closed loop process of prediction, regulation, feedback correction, which can be executed cyclically and allows overlapping in time.

[0007] Preferably, the monitoring unit includes a weighing sensor and a particle size detector, a temperature sensor, a pressure sensor and a gas analyzer.

[0008] Preferably, the accuracy of the weighing sensor is ±0.1 kg, the resolution of the particle size detector is 0.1 mm, the accuracy of the temperature sensor is ±1℃, and the accuracy of the pressure sensor is ±0.1 kPa.

[0009] Preferably, the monitoring unit is installed on the cold screening equipment and the sintering production line, including: A weighing sensor and a particle size detector are installed at the inlet of the cold screening equipment, the outlet of the upper screen, the outlet of the lower screen and the internal return material outlet, respectively, for real-time monitoring of the weight and particle size distribution data of the materials at the corresponding positions; and A plurality of temperature sensors and pressure sensors are installed on the sintering production line to collect temperature and pressure process parameters in real time during sintering.

[0010] Preferably, the sintering internal return material prediction model is a neural network model, and forward propagation and back propagation are used to train and update the weights and biases.

[0011] Preferably, the model training is based on historical data and can be adjusted according to the data size and working conditions; and multi-algorithm fusion of neural network, support vector machine and random forest can be used.

[0012] Preferably, the inputs of the model include the material weight, particle size distribution, sintering temperature and pressure, and waste gas composition of each link of the cold screening, and the output of the model is the prediction value of the internal return material quantity and its change trend.

[0013] Preferably, the regulation unit uses a proportional-integral-derivative controller to adjust the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening.

[0014] Preferably, the ultrasonic vibration device has a vibration frequency of 20-40 kHz and a power of 50-200 W.

[0015] A sintering internal return system based on cold screening optimization comprises a cold screening device, a monitoring unit, a data processing unit and a regulation unit. The cold screening device comprises an upper layer screen and a lower layer screen, the screen hole size and the inclination angle of the upper layer screen and the lower layer screen are set respectively, the screen hole size of the upper layer screen is larger than that of the lower layer screen, and the ultrasonic vibration device is arranged on the surface of the upper layer screen and the lower layer screen. The monitoring unit is used for collecting the weight, particle size distribution of the material in the cold screening device, the sintering temperature, pressure and exhaust composition in real time and transmitting the data to the data processing unit. The data processing unit is configured to perform data preprocessing, model construction, training and / or verification in the above method and output prediction results. The regulation unit is configured to dynamically link and adjust the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening device according to the model prediction results by using a proportional-integral-derivative controller.

[0016] Compared with the prior art, the present application has the following advantages: Compared with the prior art, the present application has the following advantages:

[0017] The present application is configured with multi-source real-time monitoring and industrial network transmission of weighing, particle size, temperature, pressure and gas composition at key parts of the cold screening and the sintering line, and builds a comprehensive data base, which is beneficial to timely identification of abnormalities and guarantee of production stability and product quality.

[0018] Through the preprocessing of moving average, covariance and characteristic decomposition and machine learning modeling (mainly neural network, combined with support vector machine, random forest, etc.), real-time prediction and trend warning of the internal return amount are realized, and the dependence on artificial experience is reduced.

[0019] Based on the prediction results, the proportional-integral-derivative controller is used to dynamically link and adjust the cold screening vibration frequency, amplitude and ultrasonic power, and the actual internal return amount is used as feedback to online correct the model, forming a closed-loop control of "prediction - parameter adjustment - feedback correction", so as to improve the screening effect, reduce energy consumption and enhance the long-term operation stability.

[0020] The application has good scalability and easy deployment: parameters can be flexibly set according to raw materials and scale, small enterprises can use cloud data processing to reduce hardware investment and quickly go online. In a typical production line, the amount of internal return material is reduced from an average of 15% to 12% (a decrease of 20%), and the finished sinter drum index is increased by about 3%; under fluctuating conditions, the advance control reduces the internal return material by 30% to 9% (a decrease of about 70%); in a small enterprise scenario, the amount of internal return material is reduced from 18% to 10%, the energy consumption per ton of ore is reduced by about 8%, and the production efficiency is increased by about 8%, which reflects significant comprehensive benefits. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A method flowchart of a preferred embodiment of the application; Figure 2 A system structure diagram in a preferred embodiment of the application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0023] As shown in Figure 1 and Figure 2 : First preferred embodiment: The sinter internal return reduction method based on cold screening optimization comprises the following steps: S1, a cold screening device with a double-layer vibrating screen structure is built, the cold screening device comprises an upper layer screen and a lower layer screen, and the sizes and inclination angles of the screen holes of the upper layer screen and the lower layer screen are respectively set, wherein the size of the screen hole of the upper layer screen is greater than that of the lower layer screen, the sizes, inclination angles, vibration frequencies and amplitudes of the screen holes of the upper layer screen and the lower layer screen can be independently adjusted, and a plurality of groups of ultrasonic vibration devices are respectively installed on the surfaces of the upper layer screen and the lower layer screen; Further, the inclination angles of the upper layer screen and the lower layer screen are different, and different inclination angles can change the flow path of the material on the screen, which helps to improve the uniformity and accuracy of screening.

[0024] Further, the size of the screen hole of the upper layer screen can be selected as 18mm*18mm, 20mm*20mm or 22mm*22mm, the upper layer screen adopts high wear resistance alloy material or surface hardening treated manganese steel material, which is used for preliminary screening of large sinter and sundries, and can effectively resist material impact wear.

[0025] Further, the size of the screen holes of the lower screen can be selected as 8mm*8mm, 10mm*10mm or 12mm*12mm, and the lower screen is selected as a stainless steel woven mesh with high screening precision, which is used for fine screening of qualified sinter and ensures the smooth passing of qualified sinter.

[0026] Specifically, the vibration frequency of the ultrasonic vibration device can be adjusted in the range of 20kHz-40kHz, and the power can be adjusted in the range of 50W-200W, so as to effectively prevent the screen hole from being blocked by high-frequency vibration.

[0027] The size of the screen holes, the inclination angle, the vibration frequency and the amplitude of the upper screen and the lower screen can be independently adjusted, which can adapt to different raw material characteristics and production scales, so as to meet the production needs of large and medium-sized sintering enterprises and small sintering enterprises.

[0028] S2, a monitoring unit is installed on the cold screening equipment and the sintering production line, and the monitoring unit transmits monitoring data to a data processing unit in real time through an industrial Ethernet or a wireless communication module; The monitoring unit comprises a weighing sensor and a particle size detector, a temperature sensor, a pressure sensor and a gas analyzer. Specifically, the weighing sensor and the particle size detector are installed at the feeding port of the cold screening equipment, the discharge port of the upper screen, the discharge port of the lower screen and the internal return material outlet, respectively, for real-time monitoring of the weight and particle size distribution data of the materials at the corresponding positions. A plurality of temperature sensors and pressure sensors are installed on the sintering production line for real-time acquisition of temperature and pressure process parameters in the sintering process.

[0029] Specifically, the weighing sensor (with an accuracy of ±0.1kg) and the particle size detector (with a resolution of 0.1mm) can real-time and accurately monitor the weight and particle size distribution of the materials at the corresponding positions; the temperature sensor (with an accuracy of ±1℃) and the pressure sensor (with an accuracy of ±0.1kPa) can comprehensively collect temperature and pressure parameters in the sintering process; and the gas analyzer can real-time monitor the contents of key components such as oxygen, carbon monoxide and carbon dioxide in the sintering exhaust gas, so as to optimize the input data of the model.

[0030] S3, data processing and model prediction, including data transmission, data preprocessing, construction of a prediction model and model training and verification, wherein the data preprocessing comprises sliding average filtering, feature extraction based on a covariance matrix and eigenvalue decomposition, the prediction model is an internal return material prediction model based on machine learning and outputs a prediction value and a trend thereof; comprising: S31, data transmission: the weight and particle size distribution data of the materials and the sintering process parameters collected by the monitoring unit are transmitted to the data processing unit in real time; S32, data preprocessing is performed on the data from the monitoring unit, S321 uses a moving average filtering formula to smooth data, reduce the impact of random noise, and improve data quality. The moving average filtering formula is as follows: (1) In the formula: : The moving average over time t; : The original data value at time i; n: Size of the sliding window; S322 extracts the main features from the preprocessed data by calculating the covariance matrix C, reducing the data dimensionality and extracting the features that have the most impact on internal return material prediction, thereby reducing computational complexity. The specific formula for calculating the covariance matrix C is as follows: (2) In the formula: N: Sample size, reflecting the scale of the data; T: Transpose symbol, used in matrix operations; S323, perform eigenvalue decomposition on the covariance matrix to obtain eigenvectors and eigenvalues, and select the eigenvectors corresponding to the k largest eigenvalues ​​as the principal features; S33, Construct a prediction model, which outputs a numerical prediction of the internal return material quantity and its changing trend, used to trigger subsequent linkage parameter adjustment; S331, a neural network algorithm is used to establish a prediction model for internal material return during sintering. The prediction model formula is the forward propagation formula, specifically: (3) (4) In the formula: : No. The activation value of the layer; : No. Weighted input of the layer; : No. Layer weight matrix; : No. Layer bias vector; Activation function.

[0031] S332, the parameters of the sintering internal return prediction model are optimized, and the weights and biases are updated using the backpropagation formula, specifically: (5) In the formula: The error term of the l-th layer represents the contribution of the neurons in that layer to the final output error. The transpose of the weight matrix of layer l is used to backpropagate the error from layer l+1 to layer l. δ(l+1): Error term of layer l+1; Activation function Weighted input of layer l The derivative (calculated element by element); Element-wise multiplication ensures that the error is correctly combined with the gradient of the activation function; (6) In the formula: : The weight matrix of the l-th layer; : Learning rate, controls the magnitude of parameter updates; m is the number of samples in the batch. δi(l): The error term of the i-th sample in the l-th layer; : The transpose of the activation value of the i-th sample in layer l-1; (7) In the formula: : The bias vector of the l-th layer; : The error term of the i-th sample in the l-th layer.

[0032] S34, Model Training and Validation: The model was trained based on three months of historical data from the workshop (raw material composition, process parameters, internal return material volume) and validated after multiple optimizations. Furthermore, during model training, the data processing unit employs a combination of various machine learning algorithms, including support vector machine, random forest, and neural network algorithms. By integrating the advantages of different algorithms, the accuracy and generalization ability of the prediction model are improved.

[0033] The input of the sintering internal return material prediction model is all the collected data (after sliding average filtering and covariance matrix feature extraction), including: cold screening each link material weight (such as total weight of the feed, weight of the large block material screened out by the upper layer screen, weight of the qualified material screened out by the lower layer screen), particle size distribution (such as the proportion of <10mm particles, 20-50mm particles), sintering temperature (such as ignition temperature, high temperature section peak temperature), sintering pressure (such as air permeability pressure difference of the material layer), etc.; the output of the model is the predicted value (unit: tons / hour) of the internal return material and the change trend (such as the internal return material will increase by 0.5 tons / hour in the next 30 minutes).

[0034] S4, after the control unit receives the prediction result, the vibration frequency, amplitude and ultrasonic vibration device power of the cold screen are adjusted in linkage, and a proportional-integral-derivative (PID) controller is used to realize dynamic optimization of the parameters; specifically: A proportional-integral-derivative (PID) controller is used, and the algorithm formula is: (8) In the formula: : controller output value (corresponding to the adjustment amount of vibration frequency / amplitude / power) : difference between set value and real-time feedback value; Kp, Ki and Kd: proportional, integral and differential coefficients (determined by pre-calibration).

[0035] S5, the adjusted actual internal return material data is fed back to the data processing unit, and the data processing unit corrects and optimizes the sintering internal return material prediction model according to the difference between the feedback data and the prediction data, so as to continuously improve the prediction accuracy of the model and the effectiveness of the control.

[0036] Among them, S3-S5 constitute a closed loop process of prediction, control and feedback correction, which can be executed in a loop and allow overlapping execution in time.

[0037] Second preferred embodiment: The sintering internal return material reduction method based on cold screen optimization based on the first preferred embodiment, comprising: S1, a cold screening equipment with a double-layer vibrating screen structure is built, including an upper layer screen and a lower layer screen, the size of the screen hole of the upper layer screen is accurately set to 20mm×20mm, and a high wear-resistant alloy material is used, which can effectively resist the impact and wear of large sinter and sundries; the size of the screen hole of the lower layer screen is 10mm×10mm, and a stainless steel woven mesh with high screening precision is selected to ensure that the qualified sinter passes smoothly.

[0038] The vibration frequency and amplitude of the upper screen and the lower screen can be independently adjusted; and a plurality of ultrasonic vibration devices are uniformly installed on the surface of the screen, the vibration frequency of each device can be adjusted in the range of 20 kHz to 40 kHz, and the power adjustment range is 50 W to 200 W, so that the screen hole is effectively prevented from being blocked by high-frequency vibration. S2, installing a monitoring unit, comprising: At the feed inlet of the cold screening device, the upper screen discharge outlet, the lower screen discharge outlet and the internal return material outlet, a weighing sensor and a particle size detector are respectively installed for real-time monitoring of the weight and particle size distribution data of the materials at the corresponding positions; A plurality of temperature sensors and pressure sensors are installed on the sintering production line to collect temperature and pressure process parameters in real time during sintering.

[0039] Specifically, the weighing sensor (with a precision of ±0.1 kg) and the particle size detector (with a resolution of 0.1 mm) can real-time and accurately monitor the weight and particle size distribution of the materials at the corresponding positions; the temperature sensor (with a precision of ±1℃) and the pressure sensor (with a precision of ±0.1 kPa) can comprehensively collect temperature and pressure parameters during sintering.

[0040] S3, data processing and model prediction, comprising: S31, data transmission: the material weight, particle size distribution data and sintering process parameters collected by the monitoring unit are transmitted to the data processing unit in real time; S32, data preprocessing of data from the monitoring unit, S321, smoothing data by a moving average filtering formula to reduce the influence of random noise, improve data quality, and the moving average filtering formula is specifically: (1) In the formula: : the moving average value at time t; : the original data value at time i; n: the size of the sliding window; S322, extracting main features by calculating the covariance matrix C of the preprocessed data to reduce the data dimension, extracting the features that have the greatest influence on the internal return material prediction, and reducing the computational complexity; The specific formula for calculating the covariance matrix C is: (2) In the formula: N: sample size, reflecting the data size; T: transpose symbol, used for matrix operation; S323, eigenvalue decomposition is performed on the covariance matrix to obtain eigenvectors and eigenvalues, and eigenvectors corresponding to the first k largest eigenvalues are selected as principal characteristics; S33, a sinter internal return material prediction model is constructed: S331, a neural network algorithm is used to establish a sinter internal return material prediction model, and the prediction model formula is a forward propagation formula, which is specifically: , (3) (4) In the formula: : the activation value of the i-th layer; : the weighted input of the i-th layer; : the weight matrix of the i-th layer; : the bias vector of the i-th layer; : the activation function. S332, the parameters of the sinter internal return material prediction model are optimized, and the weights and biases are updated using the back propagation formula, which is specifically: (5) In the formula:

[0041] : the error term of the i-th layer, indicating the contribution of the neurons of the layer to the final output error; : the transpose of the weight matrix of the i-th layer, used to propagate the error from the i+1 layer to the i layer; δ(i+1): the error term of the i+1 layer; : the activation function the derivative (element-wise calculation) of the weighted input of the i-th layer : element-wise multiplication, ensuring that the error and the gradient of the activation function are correctly combined; (6) In the formula: : the weight matrix of the i-th layer; : learning rate, controlling the amplitude of parameter update, m batch sample quantity; δi(i): the error term of the i-th sample in the i-th layer; : the bias vector of the i-th layer; : the activation function the derivative (element-wise calculation) of the weighted input of the i-th layer : element-wise multiplication, ensuring that the error and the gradient of the activation function are correctly combined;​ : transpose of the (i-1)-th layer activation value of the i-th sample; (7) wherein: : bias vector of the i-th layer; : error term of the i-th sample at the i-th layer.

[0042] S34, model training and verification: based on the historical data of the workshop for three months (raw material composition, process parameters, and internal return material quantity), the model is trained, and the verification is optimized for multiple times; Further, in the model training process, the data processing unit also adopts a combination of multiple machine learning algorithms, including the combination of support vector machine algorithm, random forest algorithm, and neural network algorithm, to improve the accuracy and generalization ability of the prediction model by integrating the advantages of different algorithms.

[0043] The input of the sintering internal return material prediction model is all the collected data (after sliding average filtering and covariance matrix feature extraction), including: the weight of the material at each link of the cold screen (such as the total weight of the incoming material, the weight of the large material screened out by the upper layer, and the weight of the qualified material screened out by the lower layer), the particle size distribution (such as the proportion of <10mm particles and the proportion of 20-50mm particles), the sintering temperature (such as the ignition temperature and the peak temperature in the high temperature section), the sintering pressure (such as the pressure difference of the material layer permeability), etc.; the output of the model is the predicted value of the internal return material quantity (unit: ton / hour) and the change trend (such as the internal return material quantity will increase by 0.5 tons / hour in the next 30 minutes).

[0044] The model learns the historical data of the workshop for the past three months (including the above input parameters and the actual internal return material quantity in the corresponding time period) to establish the data correlation law: for example, when the proportion of <10mm particles at the discharge port of the lower screen is less than 85% and the temperature in the high temperature section of the sintering is less than 1200℃, the internal return material quantity will significantly increase (because the qualified particle size material is not enough to be screened out, and the fine particles are included in the internal return material).

[0045] S4, the regulation unit dynamically regulates the cold screen parameters according to the model results; The regulation unit receives the output results of the sintering internal return material prediction model and dynamically adjusts the vibration frequency and amplitude of the cold screen by adjusting the power of the ultrasonic vibration device; specifically, a proportional-integral-derivative (PID) controller is used, and the algorithm formula is: (8) wherein: : controller output value (corresponding to the adjustment amount of vibration frequency / amplitude / power) : difference between the set value and the real-time feedback value; Kp, Ki and Kd: proportional, integral, derivative coefficients (determined by pre-calibration).

[0046] In the actual production process, the monitoring unit collects data in real time and transmits them to the data processing unit at a speed of milliseconds. When the prediction model shows that the amount of internal return material has an upward trend through data analysis, the control unit quickly responds by increasing the cold screening vibration frequency from the initial 15 Hz to 18 Hz, increasing the amplitude from 3 mm to 4 mm, and simultaneously increasing the power of the ultrasonic vibration device from 100 W to 120 W. After a month of stable operation, the amount of internal return material in the sintering process of this workshop was reduced from the original average of 15% to 12%, with a reduction of 20%; at the same time, due to the improvement of screening effect, the particle size of sintered ore is more uniform, and the stability of quality is significantly improved, the drum index of finished sintered ore is increased by 3%, which meets the higher production quality requirements.

[0047] Step five: feedback the adjusted actual internal return material data to the data processing unit, and the data processing unit corrects and optimizes the sintering internal return material prediction model according to the difference between the feedback data and the prediction data, and continuously improves the prediction accuracy of the model and the effectiveness of the control.

[0048] Third preferred embodiment: On the basis of the second preferred embodiment, considering the differences in the properties of raw materials, especially the particle size distribution and gangue composition of iron ore, targeted adjustments are made when implementing the system and method of the present application. The size of the upper screen mesh is adjusted to 22 mm x 22 mm, and the size of the lower screen mesh is adjusted to 12 mm x 12 mm to adapt to the particle size characteristics of the raw materials. At the same time, the screen material is optimized, the upper screen is made of manganese steel with surface hardening treatment to enhance wear resistance, and the lower screen is made of stainless steel mesh with special weaving process to further improve the screening accuracy.

[0049] During the model training process, the data processing unit uses a combination of multiple machine learning algorithms (mainly neural network algorithm, combined with support vector machine and random forest algorithm) to build a sinter internal return material prediction model. The input data adds sinter exhaust composition data (such as oxygen content, carbon monoxide concentration) on the basis of embodiment 1, which is collected in real time by a gas analyzer; the model output is still the internal return material prediction value and trend, and the learning of the correlation between the particle size fluctuation of the raw materials and the internal return material is strengthened through the fusion of multiple algorithms (such as when the <5 mm powder ore in the iron ore accounts for more than 30% and the oxygen content in the exhaust gas is less than 18%, the probability of internal return material increase is increased by 40%).

[0050] When the monitoring system detects abnormal fluctuations in sintering temperature (such as a sudden drop of 50°C in high-temperature section temperature), combined with particle size data (the proportion of particles less than 12mm on the lower screen decreases to 70%), the prediction model determines that the amount of internal return material will increase by 2 tons / hour within 1 hour, and immediately issues a warning. After receiving the warning information, the control unit increases the cold screen amplitude from the initial 2.5mm to 3.2mm, and increases the ultrasonic vibration device power from 80W to 100W. In a production anomaly caused by fluctuations in raw material ratio, through this advance control measure, the factory effectively controls the increase of internal return material. Compared with not using the invention, under similar raw material fluctuation conditions, the internal return material quantity increase rate is reduced from 30% to 9%, a reduction of 70%, greatly reducing the impact of production fluctuations on sintering indicators, and ensuring the stability and continuity of production.

[0051] Fourth preferred embodiment: Based on the second preferred embodiment, for a small sintering enterprise, due to its small production scale, limited funds and technology, the system is simplified and adapted when implementing the invention. The cold screen uses a small double-layer vibrating screen with compact structure and low cost, the upper screen mesh size is 18mm x 18mm, and the lower screen mesh size is 8mm x 8mm, which meets the basic screening needs of the enterprise. To reduce hardware investment costs, the data processing unit uses mature cloud computing resources to train models and process data by renting high-performance computing services, avoiding the need for the enterprise to purchase expensive data processing equipment. The cloud data processing unit builds a simplified sinter internal return material prediction model, focusing on core parameters: cold screen feed weight, lower screen qualified material weight, and sintering machine outlet temperature. The model output is the hourly average prediction of internal return material quantity (for example, under the current working condition, the average internal return material quantity is 5 tons / hour, which is 1 ton higher than the baseline value). By learning the production data of the enterprise in the past month, the model clearly associates the rule: when the proportion of lower screen qualified material weight to feed weight is less than 60%, the internal return material quantity will increase synchronously (every 5% decrease, the internal return material quantity increases by 0.8 tons / hour).

[0052] In daily production, the system adopts a control strategy with high automation degree. The monitoring unit collects key data in real time and uploads them to the cloud data processing unit. After model analysis, the control unit automatically adjusts the cold screening parameters according to the prediction results. To improve the usability of the system, a simple and intuitive operation interface is also provided to facilitate enterprise operators to view parameters and manually intervene. After three months of actual operation, the sinter internal return material quantity of the small enterprise is significantly reduced from the original average of 18% to 10%, with a reduction of 44%; at the same time, due to the reduction of internal return material quantity, the energy consumption in the sintering process is significantly reduced, the energy consumption per ton of sinter is reduced by 8%, and the production efficiency is improved by 8%, effectively improving the economic benefits and market competitiveness of the enterprise, providing a feasible solution for the technical upgrading and sustainable development of small sintering enterprises.

[0053] Comparative example: 1. In the raw material processing link, the existing technology has a relatively basic control over raw materials. In terms of raw material quality testing, only the main chemical components of iron ore, flux and other raw materials are subjected to routine testing, such as testing the iron grade, calcium oxide and other indicators of iron ore, but there is a lack of precise monitoring of factors closely related to internal return material, such as raw material particle size composition and moisture fluctuation. In the crushing and screening process, single-layer fixed sieves are commonly used for raw material particle size screening, and the sieve size cannot be adjusted, which cannot be flexibly changed according to the characteristics of the raw materials, resulting in low screening efficiency, a large amount of qualified particle size materials being mistakenly screened back to the furnace, and an increase in internal return material. In addition, the raw material blending process relies on manual experience for proportioning and adjustment, lacks real-time data support and automatic adjustment mechanism, and is difficult to adapt to frequent changes in raw material properties, thereby affecting the stability of the sintering process and indirectly leading to an increase in internal return material. 2. In the sintering process control link, the existing technology has a relatively backward control means. In terms of temperature control, a fixed temperature set value is commonly used, and the sintering temperature is adjusted by operators through empirical judgment by observing the flame color, sinter cross-section state, etc., which cannot accurately control the real-time changes in raw material composition, permeability, etc., and is prone to over-high or over-low temperature, affecting the sintering effect of the sinter, leading to a decrease in the yield of finished products and an increase in internal return material. The running speed and layer thickness of the sintering machine are also relatively lagging, usually set according to the fixed production plan, and cannot be optimized in time according to the actual situation in the sintering process, making it difficult for the sintering process to reach the best state, further exacerbating the internal return material problem. At the same time, the existing technology lacks control over the sintering atmosphere, only roughly controls the negative pressure by adjusting the air volume of the exhaust fan, and cannot accurately monitor and adjust the gas composition such as oxygen and carbon monoxide in the sintering process, affecting the full performance of the sintering chemical reaction, reducing the quality of the sinter, and causing an increase in internal return material. 3. Cold screening and subsequent processing are the main shortcomings of existing technologies. Currently, most cold screening equipment adopts a single-layer screen structure with ordinary screen material, poor wear resistance, and the screen holes are easily blocked by materials, reducing screening efficiency. The vibration frequency and amplitude of the cold screen are fixed and cannot be dynamically adjusted according to the properties of the material, output, and other factors, resulting in unstable screening effects. In actual production, due to incomplete screening by cold screens, a large amount of qualified sintered ore is returned to the sintering batching system as internal return material, increasing unnecessary circulation burden. In addition, existing technologies lack effective monitoring and analysis methods for internal return material, making it impossible to accurately grasp key information such as particle size distribution and composition changes, and making it difficult to take targeted optimization measures. In terms of equipment maintenance, a regular inspection mode is usually adopted, which cannot promptly detect and deal with potential faults in the cold screen operation process. Sudden equipment failures can lead to production interruptions, further affecting the stable control of internal return material.

[0054] The effects of the sintering internal reversion reduction methods based on cold sieve optimization in Examples 1 to 4 were compared with those in the control example, and the results are shown in the table below: Table 1. Comparison of the effects of the examples:

[0055] As shown in Table 1, Examples 1 to 4 have many beneficial effects compared to the control example: By combining a double-layer vibrating screen with an ultrasonic anti-clogging device, the low screening efficiency and easy clogging of the single-layer fixed screen in the control example were improved, achieving precise grading and screening. Multiple types of high-precision sensors and a comprehensive data monitoring system replaced the single and lagging monitoring method of the control example, enabling the acquisition of rich production data in real time. The use of machine learning algorithms to build predictive models and process data changed the control example's reliance on manual experience-based decision-making, achieving dynamic prediction and precise control of internal return material volume. Simultaneously, real-time equipment status monitoring and health assessment mechanisms compensated for the shortcomings of the control example's periodic maintenance, effectively ensuring stable production. These technological improvements significantly reduced the amount of internal return material in sintering compared to the control example, improved sinter quality, reduced energy consumption, and increased production efficiency and economic benefits, demonstrating significant advantages in multiple key aspects of sintering production, including equipment, monitoring, data processing, and maintenance.

[0056] Fifth preferred embodiment: A sintering internal return system based on cold screening optimization includes: cold screening equipment, monitoring unit, data processing unit and control unit; The cold screening equipment includes an upper screen and a lower screen. The screen aperture size and inclination angle of the upper screen and the lower screen are set respectively. The screen aperture size of the upper screen is larger than that of the lower screen. An ultrasonic vibration device is provided on the surface of the upper screen and the lower screen. The monitoring unit is used to collect the material weight, particle size distribution, sintering temperature, pressure and exhaust composition in the cold screening equipment in real time and transmit the data to the data processing unit; The data processing unit is configured to perform the data preprocessing, model construction, training and / or verification and output the prediction results of the first preferred embodiment; The regulation unit is configured to dynamically link and adjust the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening equipment according to the model prediction results by using a proportional-integral-derivative (PID) controller.

[0057] In the above embodiments, all or part of them can be realized by software, hardware, firmware or any combination thereof. When all or part of them are realized in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (such as infrared, wireless, microwave, etc.)) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (such as floppy disk, hard disk, magnetic tape), optical media (such as DVD), or semiconductor media (such as solid state disk (SSD)) and the like.

[0058] The above described preferred embodiments of the present application are only used to limit the present application, and any modification, equivalent replacement and improvement within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for reducing sinter internal returns based on cold screening optimization, characterized by, Comprise the following steps: S1, build a double-layer vibrating screen structure of cold screening equipment, the cold screening equipment includes upper layer screen and lower layer screen, respectively set the size and the inclination angle of the screen hole of the upper layer screen and the lower layer screen, wherein the size of the screen hole of the upper layer screen is larger than that of the lower layer screen, the ultrasonic vibration device is installed on the surface of the upper layer screen and the lower layer screen respectively; S2, install monitoring unit on the cold screening equipment and the sintering production line, the monitoring unit transmits the monitoring data to the data processing unit in real time through industrial Ethernet or wireless communication module; S3, data processing and model prediction, including data transmission, data preprocessing, constructing prediction model and model training and verification, wherein the data preprocessing includes sliding average filtering, feature extraction and eigenvalue decomposition based on covariance matrix, the prediction model is a machine learning based internal return material prediction model and outputs the prediction value and trend; S4, the regulation unit adjusts the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening equipment dynamically according to the model prediction result; S5, the adjusted actual internal return material quantity data is fed back to the data processing unit to correct and optimize the sintering internal return material prediction model; Wherein, S3~S5 constitute a closed loop process of prediction, regulation, feedback correction, which can be executed cyclically and allows overlapping execution in time.

2. The cold screen optimization based sinter internal return reduction method according to claim 1, characterized in that, The monitoring unit comprises: weighing sensor and particle size detector, temperature sensor, pressure sensor and gas analyzer.

3. The cold screen optimization based sinter internal return reduction method according to claim 2, wherein, The accuracy of the weighing sensor is ±0.1 kg, the resolution of the particle size detector is 0.1 mm, the accuracy of the temperature sensor is ±1℃, and the accuracy of the pressure sensor is ±0.1 kPa.

4. The cold screening optimization based sinter internal return reduction method according to claim 1, characterized in that, The monitoring unit is installed on the cold screening equipment and the sintering production line, comprising: At the inlet of the cold screening equipment, the outlet of the upper layer screen, the outlet of the lower layer screen and the internal return material outlet, respectively install weighing sensor and particle size detector to monitor the weight and particle size distribution data of the materials at the corresponding positions in real time; and Install several temperature sensors and pressure sensors on the sintering production line to collect temperature and pressure process parameters in real time during sintering.

5. The cold-sieve optimization based sinter internal return reduction method according to claim 1, wherein, The sintering internal return material prediction model is a neural network model, and forward propagation and back propagation are used to train and update weights and biases.

6. The cold-sieve optimization based sinter internal return reduction method according to claim 1, wherein, The model training is based on historical data and can be adjusted according to data size and working conditions; and multi-algorithm fusion of neural network, support vector machine and random forest can be used.

7. The cold-sieve optimization based sinter internal return reduction method according to claim 1, wherein, The input of the model includes: material weight, particle size distribution, sintering temperature and pressure, and waste gas composition of each link of the cold screening, and the output of the model is the prediction value and trend of the internal return material quantity.

8. The cold-sieve optimization based sinter internal return reduction method according to claim 1, wherein, The regulation unit uses a proportional-integral-derivative controller to adjust the vibration frequency, amplitude and ultrasonic vibration device power of the cold screening.

9. The cold-sieve optimization based sinter internal return reduction method of claim 1, wherein, The vibration frequency of the ultrasonic vibration device is 20 kHz~40 kHz, and the power is 50 W~200 W.

10. A sintering internal return system based on cold screening optimization, characterized by, Comprise: Cold screening equipment, monitoring unit, data processing unit and regulation unit; The cold screening device comprises an upper layer screen and a lower layer screen, the sizes and angles of inclination of the upper layer screen and the lower layer screen are respectively set, the size of the upper layer screen is larger than that of the lower layer screen, and the surfaces of the upper layer screen and the lower layer screen are provided with ultrasonic vibration devices; The monitoring unit is used for collecting the weight, particle size distribution, sintering temperature, pressure and waste gas composition of the material in the cold screening device in real time and transmitting the data to the data processing unit; The data processing unit is configured to perform the data preprocessing, model construction, training and / or verification of any one of claims 1 to 9 and output a prediction result; The regulation unit is configured to dynamically link and adjust the vibration frequency, amplitude and power of the ultrasonic vibration device of the cold screening device according to the model prediction result.

Images