Online monitoring and intelligent control methods and systems for green and low-consumption pellet production

Abstract

This invention discloses an online monitoring and intelligent control method and system for green and low-consumption pellet production, relating to the field of intelligent control technology. The method includes: collecting online operation data, basic production data, and product quality standard data for pellet production; calculating the usable value of waste heat recovery and the roasting heat replenishment requirement based on the basic production data and product quality standard data; establishing a roasting-cooling heat collaborative prediction model and outputting predicted values ​​for roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption; and generating a target control parameter set and executing control based on the output predicted values ​​for roasting section temperature, finished pellet compressive strength, return ore rate, unit gas consumption, usable waste heat recovery, and roasting heat replenishment requirement, thereby solving the problems of mismatch between waste heat utilization and gas replenishment during pellet production, and the difficulty in simultaneously achieving quality control and energy consumption control.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to an online monitoring and intelligent control method and system for green and low-consumption pellet production. Background Technology

[0002] Pellet production typically includes stages such as green pellet drying, preheating, roasting, and cooling. The roasting stage consumes fuel gas to maintain the required heat treatment temperature for the finished pellets, while the cooling stage generates exhaust air with relatively high temperatures. To reduce energy consumption in pellet production, existing production lines usually reuse the hot air exhausted from the cooling stage in the drying, preheating, or roasting stages to reduce the need for additional fuel gas. However, in actual production, the exhaust air temperature, reheat air flow rate, bed thickness, trolley speed, and the state of the material entering the furnace vary with each production batch. Adjusting the fuel gas flow rate or reheat air valve opening solely based on fixed experience can easily lead to a mismatch between the supply of waste heat from cooling and the reheat requirements for roasting, resulting in high fuel gas consumption or fluctuations in the quality of the finished pellets.

[0003] Chinese invention patent application CN115903713A, published on April 4, 2023, discloses an intelligent control method and system for thermal parameters of a chain grate machine and rotary kiln. Based on the coupling relationship of thermal parameters between pellet production equipment, it uses data mining and correlation analysis to analyze the correlation between the hot air circulation systems of the chain grate machine and rotary kiln, and constructs relevant mathematical models to visualize the adjustment of thermal parameters of the chain grate machine and rotary kiln, thereby adjusting various parameters in the thermal system of the chain grate machine and rotary kiln to adapt to changes in raw materials.

[0004] However, the aforementioned patent applications primarily establish mathematical models based on the correlation between thermal parameters to analyze and adjust the parameter relationships in the chain grate-rotary kiln thermal system. They do not address the real-time matching relationship between waste heat recovery in the cooling section and gas-fired supplementary heating in the roasting section, nor do they construct a waste heat recovery availability value representing the degree of utilization of cooling waste heat, nor do they construct a roasting supplementary heating demand value representing the current supplementary heating gap in the roasting section. Therefore, while this existing technology can improve the visualization and online optimization capabilities of thermal parameter adjustment, it still struggles to directly determine how much roasting supplementary heating the regenerating air in the cooling section can replace, and it is difficult to generate a low-consumption control parameter set based on the predicted results of the finished pellet compressive strength, return ore rate, and unit gas consumption.

[0005] Therefore, this invention provides an online monitoring and intelligent control method and system for green and low-consumption pellet production. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an online monitoring and intelligent control method and system for green and low-energy-consumption pellet production. By collecting production data for coordinated control of roasting and cooling, calculating the available value of waste heat recovery and the value of roasting supplementary heat demand, establishing a coordinated prediction model of roasting and cooling heat, and generating a low-energy-consumption control parameter set based on the model prediction results, it achieves coordinated control of gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed, thereby solving the problems of mismatch between waste heat utilization and gas supplementary heat, and difficulty in balancing quality control and energy consumption control in the pellet production process.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: an online monitoring and intelligent control method for green and low-consumption pellet production, characterized by comprising: S1. Collect online operation data, basic production data, and product quality standard data for pellet production; S2. Calculate the available value of waste heat recovery and the value of calcination heat replenishment based on production base data and product quality standard data; S3. Based on online operation data, production basic data and product quality standard data, train a physical information neural network to establish a roasting-cooling heat collaborative prediction model, and output the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption. S4. Based on the output predicted values ​​of roasting section temperature, finished pellet compressive strength, ore return rate, unit gas consumption, waste heat recovery availability, and roasting heat replenishment requirements, generate a target control parameter set and execute control.

[0008] Preferably, in S1, online operation data, basic production data, and product quality standard data of pellet production are collected as follows: Collect online operational data, basic production data, and product quality standard data for pellet production; the data collection cycle is the time interval between two consecutive data acquisitions from the industrial control system, and the data collection cycle duration is the time difference between two consecutive data acquisition moments. The online operating data includes the following values: roasting section temperature, gas flow rate, combustion air flow rate, roasting section oxygen content, roasting section furnace pressure, cooling section exhaust temperature, cooling section exhaust flow rate, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, bed thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The compressive strength value of the finished pellets is the value obtained by the quality testing equipment through compressive strength testing of the finished pellets of the current production batch; the return ore rate is the ratio of the return ore quality of the current production batch to the total quality of the finished pellets of the current production batch; the unit gas consumption is the ratio of the current gas consumption value to the current output value. The basic production data includes the current production batch number, ore type number, effective width of the trolley, bulk density of the pellet bed, lower heating value of the fuel gas, specific heat capacity of air at constant pressure, specific heat capacity of mineral powder, specific heat capacity of flux and binder, ambient air temperature, ambient pressure, mineral powder mass percentage, flux mass percentage and binder mass percentage; The product quality standard data includes a lower limit for compressive strength and an upper limit for return ore rate; the lower limit for compressive strength is the minimum compressive strength value specified in the product quality standard for the current production batch, and the upper limit for return ore rate is the maximum return ore rate specified in the product quality standard for the current production batch.

[0009] Preferably, a historical production database is acquired. This historical production database is a data set formed by the pellet production line prior to the current collection cycle, including multiple historical production records. These historical production records include the mineral type number, product quality standard data, online operation data, unit gas consumption value, finished pellet compressive strength value, and return ore rate value for each historical production batch. Based on the current production batch number, mineral type number, and product quality standard data, a matching historical production record is selected from the historical production database. This matching historical production record refers to a historical production record that has the same mineral type number, the same lower limit for compressive strength, and the same upper limit for return ore rate as the current production batch, and whose finished pellet compressive strength value is not lower than the lower limit for compressive strength and whose return ore rate value is not higher than the upper limit for return ore rate. The roasting section temperature values ​​in the matching historical production records are sorted by numerical value, and the median of the sorted values ​​is taken as the target roasting temperature value. The usable waste heat recovery value is calculated based on the cooling section exhaust temperature value, cooling section exhaust flow rate value, reheat air temperature value, reheat air flow rate value, ambient air temperature value, ambient pressure value, and collection cycle duration. Calculate the air density value based on the ambient wind temperature and ambient pressure values; calculate the exhaust heat value and regenerative heat value of the cooling section; take the difference between the exhaust temperature value of the cooling section and the ambient wind temperature value as the exhaust temperature rise value of the cooling section; if the difference is less than zero, the exhaust temperature rise value of the cooling section is recorded as zero; take the difference between the regenerative air temperature value and the ambient wind temperature value as the regenerative air temperature rise value; if the difference is less than zero, the regenerative air temperature rise value is recorded as zero. The smaller of the exhaust heat value of the cooling section and the regenerated heat value is taken as the usable value for waste heat recovery. Calculate the equivalent specific heat capacity of the pellet bed and the material throughput during the collection cycle. Based on the target roasting temperature, roasting section temperature, bed thickness, trolley speed, trolley effective width, pellet bed bulk density, equivalent specific heat capacity of the pellet bed, and collection cycle duration, calculate the roasting heat replenishment requirement.

[0010] Preferably, historical matching production records are extracted from the historical production database and training samples are formed in the order of collection time; each training sample includes model input data for one collection cycle and model validation data for the next collection cycle; The model input data includes the following values: calcination section temperature, gas flow rate, combustion air flow rate, oxygen content in the calcination section, furnace pressure in the calcination section, exhaust air temperature in the cooling section, exhaust air flow rate in the cooling section, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, material layer thickness, unit gas consumption, available waste heat recovery, and calcination heat replenishment requirement. The model validation data includes the roasting section temperature, finished pellet compressive strength, ore return rate, and unit gas consumption in the next acquisition cycle. A calcination-cooling heat co-prediction model is established using a physical information neural network. This model is used to output the predicted values ​​of the calcination section temperature, the compressive strength of the finished pellets, the return rate, and the unit gas consumption for the next acquisition cycle, based on the model input data of the current acquisition cycle.

[0011] Preferably, when training the roasting-cooling heat co-prediction model, the model input data in each training sample is input into the physical information neural network to obtain sample prediction output data; the sample prediction output data includes the predicted value of roasting section temperature, the predicted value of finished pellet compressive strength, the predicted value of return ore rate, and the predicted value of unit gas consumption. The predicted output data of the sample is compared with the model validation data in the same training sample. The differences between the predicted roasting section temperature and the roasting section temperature in the next collection cycle, the predicted compressive strength of the finished pellets and the compressive strength of the finished pellets in the next collection cycle, the predicted return rate and the return rate in the next collection cycle, and the predicted unit gas consumption and the unit gas consumption in the next collection cycle are calculated respectively. The above four differences are squared and added together to obtain the sum of squares of the data prediction error. Calculate the heat input of the gas, the heat carried out by the flue gas, and the heat required for the material bed to heat up; calculate the heat balance error. The allowable error variation is determined based on the minimum detection resolution of each data item in the model validation data, and is the sum of the squares of the minimum detection resolution corresponding to the roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption. The model parameters of the physical information neural network include the weight parameters and bias parameters of each network layer. During training, the gradient of the model training error with respect to the weight parameters and bias parameters is calculated using the backpropagation algorithm, and the weight parameters and bias parameters are updated using the gradient descent algorithm. When the decrease in model training error after two consecutive parameter updates is not greater than the allowable error variation, the updating of weight parameters and bias parameters is stopped, and the roasting-cooling heat co-prediction model is obtained. When using the roasting-cooling heat co-prediction model, input the model input data of the current collection period into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption for the next collection period.

[0012] Preferably, a control parameter correlation matrix is ​​generated based on the synchronous change direction of the gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed values ​​in the historical matching production records; the control parameter correlation matrix is ​​used to represent the relationship between the change direction of the gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed values. The control parameters include the gas flow rate, the combustion air flow rate, the regenerative air flow rate, and the trolley speed. The matrix elements in the control parameter correlation matrix are obtained as follows: For any two control parameters, read the values ​​of two adjacent acquisition cycles from the historical matching production records; if both control parameters increase or decrease, it is recorded as a change in the same direction; if one control parameter increases and the other decreases, it is recorded as a change in the opposite direction; if either control parameter remains unchanged, the number of changes in the same direction and the number of changes in the opposite direction are not counted in this adjacent acquisition cycle. Divide the difference between the number of changes in the same direction and the number of changes in the opposite direction by the sum of the number of changes in the same direction and the number of changes in the opposite direction to obtain the corresponding matrix element; when the sum of the number of changes in the same direction and the number of changes in the opposite direction is zero, the corresponding matrix element is recorded as zero. The current gas flow rate, current combustion air flow rate, current regenerating air flow rate, and current trolley speed are combined to form the current control parameter group. Based on the minimum adjustment resolution of the gas flow controller, combustion air controller, regenerating air valve controller, and trolley speed controller, three values ​​are generated for each control parameter: decreasing by N adjustment steps, keeping it unchanged, and increasing by N adjustment steps. These four control parameter values ​​are then combined to obtain a candidate control parameter group; N is a positive integer. Candidate control parameter groups exceeding the allowable output range of the corresponding controller are deleted. The allowable output range of the controller is from the minimum control value to the maximum control value that the corresponding controller can output. The gas flow controller... The gas flow controller receives the target gas flow rate and adjusts the opening of the gas regulating valve to make the actual gas flow rate change towards the target gas flow rate; the combustion air controller receives the target combustion air flow rate and adjusts the combustion air fan speed or combustion air valve opening to make the actual combustion air flow rate change towards the target combustion air flow rate; the regenerative air valve controller receives the target regenerative air flow rate and adjusts the regenerative air pipeline valve opening to make the actual regenerative air flow rate change towards the target regenerative air flow rate; the trolley speed controller receives the target trolley speed and adjusts the operating frequency or speed of the trolley drive motor to make the actual trolley speed change towards the target trolley speed.

[0013] Preferably, candidate control parameter groups are screened based on the control parameter correlation matrix: when the matrix elements corresponding to any two control parameters in the control parameter correlation matrix are positive, candidate control parameter groups in which the two control parameters change in the same direction relative to the current control parameter group are retained; when the corresponding matrix elements are negative, candidate control parameter groups in which the two control parameters change in opposite directions relative to the current control parameter group are retained; when the corresponding matrix elements are zero, candidate control parameter groups are not deleted based on the direction of change of the two control parameters. Replace the gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed values ​​in the current acquisition cycle model input data with each set of candidate control parameters, and input them into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption corresponding to each set of candidate control parameters. Candidate control parameter groups whose predicted compressive strength of finished pellets is lower than the lower limit of the qualified compressive strength are deleted, and candidate control parameter groups whose predicted return rate is higher than the upper limit of the qualified return rate are deleted. The remaining candidate control parameter groups are used as available control parameter groups. When there are no available control parameter groups, the current control parameter group is used as the target control parameter group, and the next collection cycle begins. When the available waste heat recovery value is greater than or equal to the calcination heat replenishment demand value, the group with the smallest predicted unit gas consumption value, and the candidate gas flow rate value is not higher than the current gas flow rate value and the candidate regenerating air flow rate value is not lower than the current regenerating air flow rate value is selected from the available control parameter group as the target control parameter group; when there is no available control parameter group that meets the above conditions at the same time, the group with the smallest predicted unit gas consumption value is selected from the available control parameter group as the target control parameter group. When the available waste heat recovery value is less than the calcination heat replenishment requirement value, first retain the candidate control parameter group from the available control parameter group whose predicted calcination temperature is not lower than the target calcination temperature value, and then select the group with the smallest predicted unit gas consumption value from the retained candidate control parameter group as the target control parameter group; when there is no candidate control parameter group whose predicted calcination temperature is not lower than the target calcination temperature value, select the candidate control parameter group with the largest predicted value of finished pellet compressive strength from the available control parameter group as the target control parameter group; The target control parameter group includes the target gas flow rate, the target combustion air flow rate, the target regenerative air flow rate, and the target trolley speed; the target gas flow rate is sent to the gas flow controller, the target combustion air flow rate is sent to the combustion air controller, the target regenerative air flow rate is sent to the regenerative air valve controller, and the target trolley speed is sent to the trolley speed controller. After the control is executed, online operating data is collected again in the next acquisition cycle, the available value of waste heat recovery and the value of calcination heat replenishment are calculated, and the target control parameter set is regenerated to form continuous closed-loop control.

[0014] An online monitoring and intelligent control system for green and low-consumption pellet production includes a data acquisition unit, a heat calculation unit, a predictive analysis unit, a candidate parameter generation unit, a target parameter determination unit, and a control execution unit. The data acquisition unit is used to collect online operation data, production basic data, and product quality standard data of pellet production. The online operation data includes roasting section temperature, gas flow rate, combustion air flow rate, cooling section exhaust temperature, cooling section exhaust air flow rate, reheat air temperature, reheat air flow rate, trolley speed, material layer thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The heat calculation unit is used to calculate the usable value of waste heat recovery based on the exhaust air temperature of the cooling section, the exhaust air flow rate of the cooling section, the temperature of the regenerated air, the flow rate of the regenerated air, the ambient air temperature, the ambient pressure, and the duration of the data collection cycle. It also calculates the roasting heat replenishment requirement based on the target roasting temperature, the roasting section temperature, the material layer thickness, the trolley speed, the effective width of the trolley, the bulk density of the pellet material layer, and the equivalent specific heat capacity of the pellet material layer. The predictive analysis unit is used to input the model input data into the roasting-cooling heat co-prediction model and output the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption. The candidate parameter generation unit is used to generate a control parameter correlation matrix based on the synchronous change direction between the gas flow rate, combustion air flow rate, regenerative air flow rate and trolley speed value in the historical matching production records, and to generate a candidate control parameter group based on the control parameter correlation matrix. The target parameter determination unit is used to select the target control parameter group from the candidate control parameter group based on the predicted values ​​of the compressive strength of the finished pellets, the predicted value of the return rate, and the predicted value of the unit gas consumption. The control execution unit is used to send the target gas flow rate value to the gas flow controller, the target combustion air flow rate value to the combustion air controller, the target regenerative air flow rate value to the regenerative air valve controller, and the target trolley speed value to the trolley speed controller.

[0015] (III) Beneficial Effects This invention provides an online monitoring and intelligent control method and system for green and low-consumption pellet production, which has the following beneficial effects: 1. By collecting operating data of the roasting section, waste heat data of the cooling section, energy consumption data and quality inspection data, and calculating the available value of waste heat recovery and the value of roasting heat supplementation, the waste heat utilization status of the cooling section and the heat supplementation demand of the roasting section can be quantitatively represented, providing clear data basis for the coordinated control of subsequent gas flow, combustion air flow, regenerating air flow and trolley speed.

[0016] 2. By using the available value of waste heat recovery and the value of roasting heat replenishment as model input data, a roasting-cooling heat collaborative prediction model is established. This model can predict the roasting section temperature, finished pellet compressive strength, return rate and unit gas consumption corresponding to different control parameter groups, thus avoiding control lag caused by adjusting the gas flow rate only based on the current temperature of the roasting section.

[0017] 3. By generating candidate control parameter groups based on the control parameter correlation matrix and screening target control parameter groups in conjunction with the calcination-cooling heat co-prediction model, the gas flow rate, combustion air flow rate, regenerating air flow rate and trolley speed are adjusted according to the linkage direction in the historical qualified production records, reducing parameter conflicts caused by independent adjustment of individual control parameters.

[0018] 4. By selecting a target control parameter group with a lower predicted unit gas consumption value under the premise that the predicted value of the compressive strength of the finished pellets is not lower than the lower limit of the qualified compressive strength and the predicted value of the return rate is not higher than the upper limit of the qualified return rate, it is beneficial to reduce the unit gas consumption while ensuring the quality of the finished pellets, and achieve green and low-consumption operation of pellet production. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the online monitoring and intelligent control method for green and low-consumption pellet production according to the present invention; Figure 2 This is a schematic diagram of the online monitoring and intelligent control system for green and low-consumption pellet production according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 This invention provides an online monitoring and intelligent control method and system for green and low-consumption pellet production, comprising the following steps: S1. Collect online operation data, production basic data, and product quality standard data for pellet production.

[0022] Collect online operational data, basic production data, and product quality standard data for pellet production; the data collection cycle is the time interval between two consecutive data acquisitions from the industrial control system, and the data collection cycle duration is the time difference between two consecutive data acquisition moments. The online operating data includes the following values: roasting section temperature, gas flow rate, combustion air flow rate, roasting section oxygen content, roasting section furnace pressure, cooling section exhaust temperature, cooling section exhaust flow rate, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, bed thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The compressive strength value of the finished pellets is the value obtained by the quality testing equipment through compressive strength testing of the finished pellets of the current production batch; the return ore rate is the ratio of the return ore quality of the current production batch to the total quality of the finished pellets of the current production batch; the unit gas consumption is the ratio of the current gas consumption value to the current output value. The basic production data includes the current production batch number, ore type number, effective width of the trolley, bulk density of the pellet bed, lower heating value of the fuel gas, specific heat capacity of air at constant pressure, specific heat capacity of mineral powder, specific heat capacity of flux and binder, ambient air temperature, ambient pressure, mineral powder mass percentage, flux mass percentage and binder mass percentage; The product quality standard data includes a lower limit for compressive strength and an upper limit for return ore rate; the lower limit for compressive strength is the minimum compressive strength value specified in the product quality standard for the current production batch, and the upper limit for return ore rate is the maximum return ore rate specified in the product quality standard for the current production batch.

[0023] S2. Calculate the available value of waste heat recovery and the value of calcination heat replenishment based on production base data and product quality standard data.

[0024] Obtain the historical production database, which is a data set formed by the pellet production line before the current collection cycle. It includes multiple historical production records, including the mineral type number of the historical production batch, product quality standard data, online operation data, unit gas consumption value, finished pellet compressive strength value and return ore rate value. Based on the current production batch number, mineral type number, and product quality standard data, historical matching production records are selected from the historical production database. The historical matching production records refer to historical production records that have the same mineral type number, the same lower limit of compressive strength, and the same upper limit of return rate as the current production batch, and whose finished pellet compressive strength value is not lower than the lower limit of compressive strength and whose return rate value is not higher than the upper limit of return rate. The roasting temperature values ​​in the historical matching production records are sorted by numerical value, and the median of the sorted values ​​is taken as the target roasting temperature value; the target roasting temperature value is used to represent the roasting temperature benchmark value corresponding to the current ore type number and the current product quality standard data. The usable value of waste heat recovery is calculated based on the exhaust air temperature of the cooling section, the exhaust air flow rate of the cooling section, the temperature and flow rate of the regenerated air, the ambient air temperature, the ambient pressure, and the duration of the data collection cycle. The usable value of waste heat recovery is used to represent the amount of usable heat that the regenerated air can provide to the roasting section or the preheating section within one data collection cycle. Specifically, the air density value is calculated based on the ambient wind temperature and ambient pressure values: air density value = ambient pressure value ÷ [287 × (ambient wind temperature value + 273.15)]; the difference between the cooling section exhaust air temperature value and the ambient wind temperature value is used as the cooling section exhaust air temperature rise value; if the difference is less than zero, the cooling section exhaust air temperature rise value is recorded as zero; the difference between the regenerated air temperature value and the ambient wind temperature value is used as the regenerated air temperature rise value; if the difference is less than zero, the regenerated air temperature rise value is recorded as zero. Among them, the exhaust heat value of the cooling section = air density value × air specific heat capacity at constant pressure × cooling section exhaust flow rate value × cooling section exhaust temperature rise value × data collection period duration; Regenerated air heat value = air density value × air specific heat capacity at constant pressure × regenerated air flow rate value × regenerated air temperature rise value × data collection cycle duration; The smaller of the exhaust heat value of the cooling section and the regenerated heat value is taken as the usable value for waste heat recovery. The roasting heat replenishment requirement is calculated based on the target roasting temperature, roasting section temperature, material layer thickness, trolley speed, trolley effective width, pellet material layer bulk density, pellet material layer equivalent specific heat capacity, and collection cycle duration. The roasting heat replenishment requirement represents the amount of heat required to reach the target roasting temperature within the current collection cycle. Wherein, the equivalent specific heat capacity of the pellet bed = mineral powder mass percentage × mineral powder specific heat capacity + flux mass percentage × flux specific heat capacity + binder mass percentage × binder specific heat capacity; When the temperature of the roasting section is lower than the target roasting temperature, the difference between the target roasting temperature and the temperature of the roasting section is taken as the roasting temperature difference; when the temperature of the roasting section is not lower than the target roasting temperature, the roasting temperature difference is recorded as zero. Among them, the material throughput value during the collection cycle = material layer thickness value × trolley speed value × trolley effective width value × pellet material layer bulk density value × collection cycle duration; The required heating value for roasting = the amount of material passing through during the collection cycle × the equivalent specific heat capacity of the pellet bed × the roasting temperature difference.

[0025] S3. Establish a synergistic prediction model for roasting-cooling heat.

[0026] Historical matching production records are extracted from the historical production database and training samples are formed in the order of collection time. Each training sample includes model input data for one collection period and model validation data for the next collection period. The model input data includes the following values: calcination section temperature, gas flow rate, combustion air flow rate, oxygen content in the calcination section, furnace pressure in the calcination section, exhaust air temperature in the cooling section, exhaust air flow rate in the cooling section, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, material layer thickness, unit gas consumption, available waste heat recovery, and calcination heat replenishment requirement. The model validation data includes the roasting section temperature, finished pellet compressive strength, ore return rate, and unit gas consumption in the next acquisition cycle. A calcination-cooling heat co-prediction model is established using a physical information neural network. The calcination-cooling heat co-prediction model is used to output the predicted values ​​of calcination section temperature, finished pellet compressive strength, return rate and unit gas consumption for the next acquisition cycle based on the model input data of the current acquisition cycle. When training the roasting-cooling heat co-prediction model, the model input data in each training sample is input into the physical information neural network to obtain sample prediction output data; the sample prediction output data includes the predicted value of roasting section temperature, the predicted value of finished pellet compressive strength, the predicted value of return ore rate and the predicted value of unit gas consumption. The predicted output data of the sample is compared with the model validation data in the same training sample. The differences between the predicted roasting section temperature and the roasting section temperature in the next collection cycle, the predicted compressive strength of the finished pellets and the compressive strength of the finished pellets in the next collection cycle, the predicted return rate and the return rate in the next collection cycle, and the predicted unit gas consumption and the unit gas consumption in the next collection cycle are calculated respectively. The above four differences are squared and added together to obtain the sum of squares of the data prediction error. Calculate the heat balance error. Heat balance error = heat input from gas + heat input from regenerated air - heat carried out from flue gas - heat required for material bed heating. Model training error is calculated as = sum of squares of data prediction errors + square of heat balance error. Wherein, the input heat of gas = gas flow rate × gas lower heating value × collection cycle duration; the input heat of regenerated air is the heat value of regenerated air; the heat carried out by flue gas = air density × air specific heat capacity at constant pressure × flue gas flow rate × (flue gas temperature - ambient air temperature) × collection cycle duration, when the flue gas temperature is lower than the ambient air temperature, the heat carried out by flue gas is recorded as zero; the heat required for heating the material bed = material throughput during the collection cycle × equivalent specific heat capacity of the pellet material bed × roasting temperature difference; The allowable error variation is determined based on the minimum detection resolution of each data item in the model validation data, specifically the sum of the squares of the minimum detection resolution corresponding to the roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption. The model parameters of the physical information neural network include the weight parameters and bias parameters of each network layer. During training, the gradient of the model training error with respect to the weight parameters and bias parameters is calculated using the backpropagation algorithm, and the weight parameters and bias parameters are updated using the gradient descent algorithm. When the decrease in model training error after two consecutive parameter updates is not greater than the allowable error variation, the updating of the weight parameters and bias parameters is stopped, resulting in the roasting-cooling heat co-prediction model. When using the roasting-cooling heat co-prediction model, input the model input data of the current collection period into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption for the next collection period.

[0027] S4. Generate the target control parameter set and execute the control.

[0028] Based on the synchronous change direction of gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed in historical matching production records, a control parameter correlation matrix is ​​generated; the control parameter correlation matrix is ​​used to represent the relationship between the change direction of gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed. The control parameters include the gas flow rate, the combustion air flow rate, the regenerative air flow rate, and the trolley speed. The matrix elements in the control parameter correlation matrix are obtained as follows: For any two control parameters, read the values ​​of two adjacent acquisition cycles from the historical matching production records; if both control parameters increase or decrease, it is recorded as a change in the same direction; if one control parameter increases and the other decreases, it is recorded as a change in the opposite direction; if either control parameter remains unchanged, the number of changes in the same direction and the number of changes in the opposite direction are not counted in this adjacent acquisition cycle. Divide the difference between the number of changes in the same direction and the number of changes in the opposite direction by the sum of the number of changes in the same direction and the number of changes in the opposite direction to obtain the corresponding matrix element; when the sum of the number of changes in the same direction and the number of changes in the opposite direction is zero, the corresponding matrix element is recorded as zero. The current gas flow rate, current combustion air flow rate, current regenerating air flow rate, and current trolley speed are combined to form the current control parameter group. Based on the minimum adjustment resolution of the gas flow controller, combustion air controller, regenerating air valve controller, and trolley speed controller, three values ​​are generated for each control parameter: decreasing by N adjustment steps, keeping it unchanged, and increasing by N adjustment steps. These four control parameter values ​​are then combined to obtain a candidate control parameter group; N is a positive integer. Candidate control parameter groups exceeding the allowable output range of the corresponding controller are deleted. The allowable output range of the controller is from the minimum control value to the maximum control value that the corresponding controller can output. The gas flow controller... The gas flow controller receives the target gas flow rate and adjusts the opening of the gas regulating valve to make the actual gas flow rate change towards the target gas flow rate; the combustion air controller receives the target combustion air flow rate and adjusts the combustion air fan speed or combustion air valve opening to make the actual combustion air flow rate change towards the target combustion air flow rate; the regenerative air valve controller receives the target regenerative air flow rate and adjusts the regenerative air pipeline valve opening to make the actual regenerative air flow rate change towards the target regenerative air flow rate; the trolley speed controller receives the target trolley speed and adjusts the operating frequency or speed of the trolley drive motor to make the actual trolley speed change towards the target trolley speed. Candidate control parameter groups are filtered based on the control parameter correlation matrix: when the matrix elements corresponding to any two control parameters in the control parameter correlation matrix are positive, candidate control parameter groups in which the two control parameters change in the same direction relative to the current control parameter group are retained; when the corresponding matrix elements are negative, candidate control parameter groups in which the two control parameters change in opposite directions relative to the current control parameter group are retained; when the corresponding matrix elements are zero, candidate control parameter groups are not deleted based on the direction of change of the two control parameters. Replace the gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed values ​​in the current acquisition cycle model input data with each set of candidate control parameters, and input them into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption corresponding to each set of candidate control parameters. Candidate control parameter groups whose predicted compressive strength of finished pellets is lower than the lower limit of the qualified compressive strength are deleted, and candidate control parameter groups whose predicted return rate is higher than the upper limit of the qualified return rate are deleted. The remaining candidate control parameter groups are used as available control parameter groups. When there are no available control parameter groups, the current control parameter group is used as the target control parameter group, and the next collection cycle begins. When the available waste heat recovery value is greater than or equal to the calcination heat replenishment demand value, the group with the smallest predicted unit gas consumption value, and the candidate gas flow rate value is not higher than the current gas flow rate value and the candidate regenerating air flow rate value is not lower than the current regenerating air flow rate value is selected from the available control parameter group as the target control parameter group; when there is no available control parameter group that meets the above conditions at the same time, the group with the smallest predicted unit gas consumption value is selected from the available control parameter group as the target control parameter group. When the available waste heat recovery value is less than the calcination heat replenishment requirement value, first retain the candidate control parameter group from the available control parameter group whose predicted calcination temperature is not lower than the target calcination temperature value, and then select the group with the smallest predicted unit gas consumption value from the retained candidate control parameter group as the target control parameter group; when there is no candidate control parameter group whose predicted calcination temperature is not lower than the target calcination temperature value, select the candidate control parameter group with the largest predicted value of finished pellet compressive strength from the available control parameter group as the target control parameter group; The target control parameter group includes the target gas flow rate, the target combustion air flow rate, the target regenerative air flow rate, and the target trolley speed; the target gas flow rate is sent to the gas flow controller, the target combustion air flow rate is sent to the combustion air controller, the target regenerative air flow rate is sent to the regenerative air valve controller, and the target trolley speed is sent to the trolley speed controller. After control is executed, online operating data is collected again in the next acquisition cycle, the available value of waste heat recovery and the value of calcination heat replenishment are calculated, and the target control parameter set is regenerated to form continuous closed-loop control.

[0029] Please see Figure 2 This invention provides an online monitoring and intelligent control system for green and low-consumption pellet production. The system includes a data acquisition unit, a heat calculation unit, a predictive analysis unit, a candidate parameter generation unit, a target parameter determination unit, and a control execution unit. The data acquisition unit is used to collect online operation data, production basic data, and product quality standard data of pellet production. The online operation data includes roasting section temperature, gas flow rate, combustion air flow rate, cooling section exhaust temperature, cooling section exhaust air flow rate, reheat air temperature, reheat air flow rate, trolley speed, material layer thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The heat calculation unit is used to calculate the usable value of waste heat recovery based on the exhaust air temperature of the cooling section, the exhaust air flow rate of the cooling section, the temperature of the regenerated air, the flow rate of the regenerated air, the ambient air temperature, the ambient pressure, and the duration of the data collection cycle. It also calculates the roasting heat replenishment requirement based on the target roasting temperature, the roasting section temperature, the material layer thickness, the trolley speed, the effective width of the trolley, the bulk density of the pellet material layer, and the equivalent specific heat capacity of the pellet material layer. The predictive analysis unit is used to input the model input data into the roasting-cooling heat co-prediction model and output the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption. The candidate parameter generation unit is used to generate a control parameter correlation matrix based on the synchronous change direction between the gas flow rate, combustion air flow rate, regenerative air flow rate and trolley speed value in the historical matching production records, and to generate a candidate control parameter group based on the control parameter correlation matrix. The target parameter determination unit is used to select the target control parameter group from the candidate control parameter group based on the predicted values ​​of the compressive strength of the finished pellets, the predicted value of the return rate, and the predicted value of the unit gas consumption. The control execution unit is used to send the target gas flow rate value to the gas flow controller, the target combustion air flow rate value to the combustion air controller, the target regenerative air flow rate value to the regenerative air valve controller, and the target trolley speed value to the trolley speed controller.

[0030] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0031] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0032] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An online monitoring and intelligent control method for green and low-consumption pellet production, characterized in that, include: S1. Collect online operation data, basic production data, and product quality standard data for pellet production; S2. Calculate the available value of waste heat recovery and the value of calcination heat replenishment based on production base data and product quality standard data; S3. Based on online operation data, production basic data and product quality standard data, train a physical information neural network to establish a roasting-cooling heat collaborative prediction model, and output the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption. S4. Based on the output predicted values ​​of roasting section temperature, finished pellet compressive strength, ore return rate, unit gas consumption, waste heat recovery availability, and roasting heat replenishment requirements, generate a target control parameter set and execute control.

2. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S1, online operational data, basic production data, and product quality standard data for pellet production are collected, as follows: Collect online operational data, basic production data, and product quality standard data for pellet production; the collection cycle is the time interval between two consecutive data collections, and the collection cycle duration is the time difference between two consecutive data collection moments. The online operating data includes the following values: roasting section temperature, gas flow rate, combustion air flow rate, roasting section oxygen content, roasting section furnace pressure, cooling section exhaust temperature, cooling section exhaust flow rate, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, bed thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The compressive strength value of the finished pellets is the value obtained by the quality testing equipment through compressive strength testing of the finished pellets of the current production batch; the return ore rate is the ratio of the return ore quality of the current production batch to the total quality of the finished pellets of the current production batch; the unit gas consumption is the ratio of the current gas consumption value to the current output value. The basic production data includes the current production batch number, ore type number, effective width of the trolley, bulk density of the pellet bed, lower heating value of the fuel gas, specific heat capacity of air at constant pressure, specific heat capacity of mineral powder, specific heat capacity of flux and binder, ambient air temperature, ambient pressure, mineral powder mass percentage, flux mass percentage and binder mass percentage; The product quality standard data includes a lower limit for compressive strength and an upper limit for return ore rate; the lower limit for compressive strength is the minimum compressive strength value specified in the product quality standard for the current production batch, and the upper limit for return ore rate is the maximum return ore rate specified in the product quality standard for the current production batch.

3. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S2, the available value of waste heat recovery and the value of calcination heat replenishment are calculated based on production base data and product quality standard data, as follows: A historical production database is acquired. This database is a collection of data recorded by the pellet production line prior to the current collection period, including multiple historical production records. These records include the mineral type number, product quality standard data, online operation data, unit gas consumption, finished pellet compressive strength, and return ore rate for each historical production batch. Based on the current production batch number, mineral type number, and product quality standard data, a matching historical production record is selected from the historical production database. This matching record refers to a historical production record that shares the same mineral type number, the same lower limit for compressive strength, and the same upper limit for return ore rate as the current production batch, and whose finished pellet compressive strength is not lower than the lower limit and whose return ore rate is not higher than the upper limit. The roasting temperature values ​​in the matching historical production records are sorted by numerical value, and the median of the sorted values ​​is taken as the target roasting temperature value. The usable value of waste heat recovery is calculated based on the exhaust air temperature of the cooling section, the exhaust air flow rate of the cooling section, the temperature of the regenerated air, the flow rate of the regenerated air, the ambient air temperature, the ambient pressure, and the duration of the data collection period. Calculate the air density value based on the ambient wind temperature and ambient pressure values; calculate the exhaust heat value and regenerative heat value of the cooling section; take the difference between the exhaust temperature value of the cooling section and the ambient wind temperature value as the exhaust temperature rise value of the cooling section; if the difference is less than zero, the exhaust temperature rise value of the cooling section is recorded as zero; take the difference between the regenerative air temperature value and the ambient wind temperature value as the regenerative air temperature rise value; if the difference is less than zero, the regenerative air temperature rise value is recorded as zero. The smaller of the exhaust heat value of the cooling section and the regenerated heat value is taken as the usable value for waste heat recovery. Calculate the equivalent specific heat capacity of the pellet bed and the material throughput during the collection cycle. Based on the target roasting temperature, roasting section temperature, bed thickness, trolley speed, trolley effective width, pellet bed bulk density, equivalent specific heat capacity of the pellet bed, and collection cycle duration, calculate the roasting heat replenishment requirement.

4. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S3, a synergistic prediction model for calcination-cooling heat is established as follows: Historical matching production records are extracted from the historical production database and training samples are formed in the order of collection time. Each training sample includes model input data for one collection period and model validation data for the next collection period. The model input data includes the following values: calcination section temperature, gas flow rate, combustion air flow rate, oxygen content in the calcination section, furnace pressure in the calcination section, exhaust air temperature in the cooling section, exhaust air flow rate in the cooling section, regenerating air temperature, regenerating air flow rate, flue gas temperature, flue gas flow rate, trolley speed, material layer thickness, unit gas consumption, available waste heat recovery, and calcination heat replenishment requirement. The model validation data includes the roasting section temperature, finished pellet compressive strength, ore return rate, and unit gas consumption in the next acquisition cycle. A calcination-cooling heat co-prediction model is established using a physical information neural network. This model is used to output the predicted values ​​of the calcination section temperature, the compressive strength of the finished pellets, the return rate, and the unit gas consumption for the next acquisition cycle, based on the model input data of the current acquisition cycle.

5. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S3, a synergistic prediction model for calcination-cooling heat is established as follows: When training the roasting-cooling heat co-prediction model, the model input data in each training sample is input into the physical information neural network to obtain sample prediction output data; the sample prediction output data includes the predicted value of roasting section temperature, the predicted value of finished pellet compressive strength, the predicted value of return ore rate and the predicted value of unit gas consumption. The predicted output data of the sample is compared with the model validation data in the same training sample. The differences between the predicted roasting section temperature and the roasting section temperature in the next collection cycle, the predicted compressive strength of the finished pellets and the compressive strength of the finished pellets in the next collection cycle, the predicted return rate and the return rate in the next collection cycle, and the predicted unit gas consumption and the unit gas consumption in the next collection cycle are calculated respectively. The above four differences are squared and added together to obtain the sum of squares of data prediction errors. Calculate the heat input of the gas, the heat carried out by the flue gas, and the heat required for the material bed to heat up; calculate the heat balance error. The allowable error variation is determined based on the minimum detection resolution of each data item in the model validation data, which is the sum of the squares of the minimum detection resolution corresponding to the roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption; the model parameters of the physical information neural network include the weight parameters and bias parameters of each network layer; During training, the gradient of the model training error with respect to the weight parameters and bias parameters is calculated using the backpropagation algorithm, and the weight parameters and bias parameters are updated using the gradient descent algorithm. When the decrease in model training error after two consecutive parameter updates is not greater than the allowable error change value, the updating of weight parameters and bias parameters is stopped, and the calcination-cooling heat co-prediction model is obtained. When using the roasting-cooling heat co-prediction model, input the model input data of the current collection period into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption for the next collection period.

6. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S4, the target control parameter set is generated and control is executed as follows: Based on the synchronous change direction of gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed in historical matching production records, a control parameter correlation matrix is ​​generated; the control parameter correlation matrix is ​​used to represent the relationship between the change direction of gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed. The control parameters include the gas flow rate, the combustion air flow rate, the regenerative air flow rate, and the trolley speed. The matrix elements in the control parameter correlation matrix are obtained as follows: For any two control parameters, read the values ​​of two adjacent acquisition cycles from the historical matching production records; if both control parameters increase or decrease, it is recorded as a change in the same direction; if one control parameter increases and the other decreases, it is recorded as a change in the opposite direction; if either control parameter remains unchanged, the number of changes in the same direction and the number of changes in the opposite direction are not counted in this adjacent acquisition cycle. Divide the difference between the number of changes in the same direction and the number of changes in the opposite direction by the sum of the number of changes in the same direction and the number of changes in the opposite direction to obtain the corresponding matrix element; when the sum of the number of changes in the same direction and the number of changes in the opposite direction is zero, the corresponding matrix element is recorded as zero. The current gas flow rate, current combustion air flow rate, current regenerating air flow rate, and current trolley speed are combined to form the current control parameter group. Based on the minimum adjustment resolution of the gas flow controller, combustion air controller, regenerating air valve controller, and trolley speed controller, three values ​​are generated for each control parameter: decrease by N adjustment steps, remain unchanged, and increase by N adjustment steps. The values ​​of the four control parameters are then combined to obtain the candidate control parameter group; N is a positive integer. The following steps are taken to remove candidate control parameter groups that exceed the allowable output range of the corresponding controller; the allowable output range of the controller is the minimum to maximum control value that the corresponding controller can output; the gas flow controller is used to receive the target gas flow value and adjust the opening of the gas regulating valve to make the actual gas flow value change towards the target gas flow value; the combustion air controller is used to receive the target combustion air flow value and adjust the combustion air fan speed or combustion air valve opening to make the actual combustion air flow value change towards the target combustion air flow value; the regenerative air valve controller is used to receive the target regenerative air flow value and adjust the regenerative air pipeline valve opening to make the actual regenerative air flow value change towards the target regenerative air flow value; the trolley speed controller is used to receive the target trolley speed value and adjust the operating frequency or speed of the trolley drive motor to make the actual trolley speed value change towards the target trolley speed value.

7. The online monitoring and intelligent control method for green and low-consumption pellet production according to claim 1, characterized in that: In S4, the target control parameter set is generated and control is executed as follows: Candidate control parameter groups are filtered based on the control parameter correlation matrix: when the matrix elements corresponding to any two control parameters in the control parameter correlation matrix are positive, candidate control parameter groups in which the two control parameters change in the same direction relative to the current control parameter group are retained; when the corresponding matrix elements are negative, candidate control parameter groups in which the two control parameters change in opposite directions relative to the current control parameter group are retained; when the corresponding matrix elements are zero, candidate control parameter groups are not deleted based on the direction of change of the two control parameters. Replace the gas flow rate, combustion air flow rate, regenerating air flow rate, and trolley speed values ​​in the current acquisition cycle model input data with each set of candidate control parameters, and input them into the roasting-cooling heat co-prediction model to obtain the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate, and unit gas consumption corresponding to each set of candidate control parameters. Candidate control parameter groups whose predicted compressive strength of finished pellets is lower than the lower limit of the qualified compressive strength are deleted, and candidate control parameter groups whose predicted return rate is higher than the upper limit of the qualified return rate are deleted. The remaining candidate control parameter groups are used as the available control parameter groups. If no available control parameter set exists, the current control parameter set will be used as the target control parameter set, and the next acquisition cycle will begin. When the available waste heat recovery value is greater than or equal to the calcination heat replenishment demand value, the group with the smallest predicted unit gas consumption value, and the candidate gas flow rate value is not higher than the current gas flow rate value and the candidate regenerating air flow rate value is not lower than the current regenerating air flow rate value is selected from the available control parameter group as the target control parameter group; when there is no available control parameter group that meets the above conditions at the same time, the group with the smallest predicted unit gas consumption value is selected from the available control parameter group as the target control parameter group. When the available value of waste heat recovery is less than the value of calcination heat replenishment, first retain the candidate control parameter group from the available control parameter group whose predicted calcination temperature is not lower than the target calcination temperature value, and then select the group with the smallest predicted unit gas consumption from the retained candidate control parameter group as the target control parameter group. When there is no candidate control parameter group whose predicted calcination temperature is not lower than the target calcination temperature, the candidate control parameter group with the largest predicted compressive strength of the finished pellets is selected from the available control parameter groups as the target control parameter group. The target control parameter group includes the target gas flow rate, the target combustion air flow rate, the target regenerative air flow rate, and the target trolley speed; the target gas flow rate is sent to the gas flow controller, the target combustion air flow rate is sent to the combustion air controller, the target regenerative air flow rate is sent to the regenerative air valve controller, and the target trolley speed is sent to the trolley speed controller. After the control is executed, online operating data is collected again in the next acquisition cycle, the available value of waste heat recovery and the value of calcination heat replenishment are calculated, and the target control parameter set is regenerated to form continuous closed-loop control.

8. An online monitoring and intelligent control system for green and low-consumption pellet production, comprising a data acquisition unit, a heat calculation unit, a predictive analysis unit, a candidate parameter generation unit, a target parameter determination unit, and a control execution unit: The data acquisition unit is used to collect online operation data, production basic data, and product quality standard data of pellet production. The online operation data includes roasting section temperature, gas flow rate, combustion air flow rate, cooling section exhaust temperature, cooling section exhaust air flow rate, reheat air temperature, reheat air flow rate, trolley speed, material layer thickness, current output, current gas consumption, finished pellet compressive strength, and return ore rate. The heat calculation unit is used to calculate the usable value of waste heat recovery based on the exhaust air temperature of the cooling section, the exhaust air flow rate of the cooling section, the temperature of the regenerated air, the flow rate of the regenerated air, the ambient air temperature, the ambient pressure, and the duration of the data collection cycle. It also calculates the roasting heat replenishment requirement based on the target roasting temperature, the roasting section temperature, the material layer thickness, the trolley speed, the effective width of the trolley, the bulk density of the pellet material layer, and the equivalent specific heat capacity of the pellet material layer. The predictive analysis unit is used to input the model input data into the roasting-cooling heat co-prediction model and output the predicted values ​​of roasting section temperature, finished pellet compressive strength, return ore rate and unit gas consumption. The candidate parameter generation unit is used to generate a control parameter correlation matrix based on the synchronous change direction between the gas flow rate, combustion air flow rate, regenerative air flow rate and trolley speed value in the historical matching production records, and to generate a candidate control parameter group based on the control parameter correlation matrix. The target parameter determination unit is used to select the target control parameter group from the candidate control parameter group based on the predicted values ​​of the compressive strength of the finished pellets, the predicted value of the return rate, and the predicted value of the unit gas consumption. The control execution unit is used to send the target gas flow rate value to the gas flow controller, the target combustion air flow rate value to the combustion air controller, the target regenerative air flow rate value to the regenerative air valve controller, and the target trolley speed value to the trolley speed controller.

Images