A production control method and system for an alumina decomposition process

Abstract

This invention belongs to the field of industrial control technology and provides a production control method and system for an alumina decomposition process. The method includes: real-time process data acquisition, collaborative prediction and rolling optimization, speed adjustment, opening adjustment, process parameter optimization, parameter updating, linkage control, and feedback correction. Through collaborative prediction and rolling optimization, this invention achieves precise collaborative control of the temperatures of the first and last tanks, reducing temperature deviations and suppressing disturbances. Through closed-loop speed control, it achieves precise control of the additive dosage, eliminating the risk of overflow or empty tanks. Through incremental PID control of the air valve opening, it achieves stable following of material feeding and air pressure protection. Through BP neural network prediction and genetic algorithm optimization, it achieves prediction of the decomposition rate and multi-parameter collaborative optimization, which can stably improve the decomposition rate. Through bidirectional interaction and linkage models between upstream and downstream processes, it achieves material and energy matching between decomposition, roasting, and evaporation.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to a production control method and system for an alumina decomposition process. Background Technology

[0002] In the alumina production process, the decomposition process is the core link that determines the product yield, quality and production cost. Its process involves multiple key operations such as temperature control of the first and last tanks, addition of crystallization aids, detection of solid content in slurry, and adjustment of feed rate. Moreover, there are strong coupling relationships between the various operating parameters. Currently, the industry generally adopts a manual-led production mode.

[0003] However, this model has many problems. The temperatures of the initial and final tanks are affected by a combination of factors, including the mother liquor flow rate, the plate cooling water flow rate, the decomposition concentrate temperature, and the ambient temperature. Manual adjustment relies on the operator's experience, and the response time is usually over 30 minutes, with temperature deviations reaching significant levels. Furthermore, the inability to achieve multi-variable coordinated adjustment easily leads to uneven particle size distribution during crystallization. Existing single-loop PID control schemes can only adjust the cooling water flow rate individually, failing to anticipate pre-disturbances such as filtrate flow rate, resulting in limited control effectiveness. Current methods for adding crystallization aids often involve manual weighing and pouring, leading to measurement errors that can reach [amount missing]. Furthermore, the addition ratio cannot be dynamically adjusted in real time according to the feed flow rate of the decomposition tank, leading to waste of additives or poor crystallization effect. The liquid level in the crystallization aid storage tank relies solely on fixed-period manual inspection, which cannot be monitored in real time, posing a serious risk of tank overflow or empty tanks, potentially causing production interruptions. The slurry solids content is a key control parameter in the decomposition process. Current technology relies on manual sampling once a periodic time and sending it to the laboratory for testing. The test results are severely delayed and cannot guide timely production adjustments. Simultaneously, the sampling process is prone to leakage of highly alkaline slurry, posing a safety risk of alkali burns and environmental pollution. The feed rate is controlled manually by adjusting the air valve, resulting in a long response time and an inability to respond promptly to rapid changes in the decomposition tank level, posing a risk of tank overflow and empty tanks, easily leading to feed interruptions and affecting the continuous and stable production of subsequent roasting processes. Current methods for improving the decomposition rate rely entirely on the accumulated experience of operators. The operational differences between different operators are significant, and adjustments to process parameters are often arbitrary, failing to achieve stable and continuous improvement. The data between the decomposition process and the upstream evaporation process and the downstream roasting process are independent of each other. Information exchange relies on manual communication, and the data transmission delay is relatively long. It is impossible to dynamically adjust its own process parameters according to the changes in the operating conditions of the upstream and downstream processes, resulting in low production efficiency throughout the entire process. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the purpose of this invention is to provide a production control method and system for the alumina decomposition process, which solves the problems of low temperature control accuracy, large error in crystallization aid addition, lag in solid content detection, untimely adjustment of feed rate, lack of basis for decomposition rate optimization, and poor cross-process coordination in the existing alumina decomposition process.

[0005] To achieve the above objectives, the present invention provides the following solution: A production control method for an alumina decomposition process includes: Collect real-time operating parameters of decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data; Based on the real-time process data, the temperature of the first tank and the temperature of the last tank are jointly predicted and rolled optimized using a pre-built state-space model and a multivariate predictive control algorithm to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. The target amount of crystallization aid is set according to the feed flow rate, and the metering pump speed is adjusted using closed-loop control according to the target amount of crystallization aid. The opening of the lifting air valve is adaptively adjusted using incremental PID control based on the liquid level difference between the decomposition tank and the lifting tank. The concentration of the original decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor were all considered. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. The optimal control parameter sequence and the optimized process parameter combination are updated to the equipment at the decomposed process node; Based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination, the upstream evaporation process and the downstream roasting process are adjusted using the roasting linkage model and the evaporation linkage model, respectively. Based on the actual operating data and production indicators of each process, the predicted deviation is extracted and the feedback correction mechanism is used to update the processing model of each process, and the process returns to the step of "collecting the real-time operating parameters of the decomposed process nodes, preprocessing the real-time operating parameters, and obtaining real-time process data".

[0006] Preferably, based on the real-time process data, a pre-built state-space model and a multivariate predictive control algorithm are used to collaboratively predict and roll-optimize the temperatures of the first and last tanks, obtaining an optimal control parameter sequence that satisfies temperature difference constraints and temperature stability requirements, including: Historical production data were analyzed using Pearson correlation analysis to identify key perturbation variables and key manipulated variables. The key perturbation variables included: filtrate flow rate, decomposition concentrate temperature, and ambient temperature. The key manipulated variables included: mother liquor flow rate and plate cooling water flow rate. A second-order discrete state-space prediction model with the first tank temperature and the last tank temperature as state variables is constructed, and the second-order discrete state-space prediction model is identified offline using a subspace identification algorithm. A minimization objective is set, and the second-order discrete state-space prediction model is rolled over based on the minimization objective to obtain the optimal control parameter sequence corresponding to the key manipulated variable.

[0007] Preferably, the minimization objective is: ;in, The objective function is... For prediction in the time domain; This is the output of the i-th future step predicted at time k; Set a value for the output of the i-th step in the future; This represents the output error weight matrix; Indicates control over the time domain; This represents the control increment predicted at time k for the j-th future step; To control the incremental weight matrix.

[0008] Preferably, the target addition amount of the crystallization aid is set according to the feed flow rate, and the metering pump speed is adjusted using closed-loop control according to the target addition amount of the crystallization aid, including: The feed flow rate is input into a proportional control model for calculation to obtain the target addition amount of the crystallization aid; the expression of the proportional control model is: ;in, The target addition amount of the crystallization aid; To add a scaling factor; The feed flow rate is [value missing]. The target addition amount of the crystallization aid is converted into the target rotational speed of the metering pump based on the calibration curve of the metering pump. The actual speed of the metering pump is adjusted in a closed loop using a PI control algorithm until the actual speed of the metering pump reaches the target speed and then the speed is maintained. Monitor the liquid level in the crystallization aid storage tank. When the liquid level in the storage tank reaches the first-level threshold, trigger an alarm. When the liquid level in the storage tank reaches the second-level threshold, cut off the power supply to the metering pump.

[0009] Preferably, the expression for the incremental PID control is: ;in, This represents the current adjustment amount of the damper opening. This is the proportionality coefficient; This is the normalized liquid level difference deviation signal at the current moment; This is the liquid level difference deviation signal from the previous moment; The integral coefficient; These are the differential coefficients; This represents the liquid level difference deviation signal between the first two time points.

[0010] Preferably, the concentration of the decomposition stock solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the amount of mother liquor α are included. k The input values ​​are used to a BP neural network to predict the decomposition rate, and the process parameters are optimized with the goal of maximizing the decomposition rate, resulting in an optimal combination of process parameters, including: The normalized concentration of the decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the mother liquor α are used to determine the optimal concentration of the decomposition solution. k The input values ​​are used to train a three-layer BP neural network to obtain the predicted decomposition rate. The objective function is to maximize the predicted decomposition rate. With the allowable fluctuation range of each input parameter as a constraint, a genetic algorithm is used to perform global optimization based on the objective function to obtain the optimized combination of process parameters.

[0011] Preferably, the expression for the roasting linkage model is: The expression for the evaporation linkage model is: ;in, The adjusted temperature setpoint for the roasting furnace; The reference calcination temperature; This is the temperature regulation coefficient; Real-time slurry moisture content; Based on the moisture content; The amount of evaporation that needs to be adjusted for the evaporation process; This represents the amount of mother liquor required for the decomposition process; This represents the amount of mother liquor lost.

[0012] Preferably, the upstream evaporation process and the downstream roasting process are adjusted respectively using a roasting linkage model and an evaporation linkage model based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination, including: The moisture content of the slurry and the particle size of the decomposed slurry are input into the roasting linkage model for calculation to obtain the roasting furnace adjustment temperature. The feeding capacity of the roasting furnace is determined based on the real-time thermal state and equipment constraints of the roasting process. When the feeding capacity of the roasting furnace is less than the feeding amount of the decomposition process, the feeding amount is reduced to a level that matches the feeding capacity of the roasting furnace. The required amount of mother liquor and the target mother liquor α kThe value is input into the evaporation linkage model for calculation to obtain the evaporation adjustment amount; When the actual value fed back from the evaporation process deviates from the target value by more than the limit, the temperature of the first tank is adjusted using the set temperature coordination adjustment formula.

[0013] Preferably, the temperature coordinated adjustment formula is: ;in, The adjustment amount for the setpoint of the first tank temperature; This is the temperature conversion coefficient; The actual mother liquor α fed back from the evaporation process k value; To meet the target mother liquor α requirements of the decomposition process k value.

[0014] A production control system for an alumina decomposition process includes: The data acquisition module is used to collect real-time operating parameters of the decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data. The rolling optimization module is used to perform collaborative prediction and rolling optimization of the first tank temperature and the last tank temperature based on the real-time process data using a pre-built state-space model and a multivariate predictive control algorithm, so as to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. The speed regulation module is used to set the target addition amount of crystallization aid according to the feed flow rate, and to regulate the speed of the metering pump according to the target addition amount of crystallization aid using closed-loop control. The opening adjustment module is used to adaptively adjust the opening of the lifting air valve based on the liquid level difference between the decomposition tank and the lifting tank using incremental PID control. The process parameter optimization module is used to optimize parameters such as the concentration of the decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. The parameter update module is used to update the optimal control parameter sequence and the optimized process parameter combination to the equipment of the decomposed process node; The linkage control module is used to adjust the upstream evaporation process and the downstream roasting process respectively using the roasting linkage model and the evaporation linkage model according to the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination; The closed-loop control module is used to extract predicted deviations based on the actual operating data and production indicators of each process, use the feedback correction mechanism to update the processing model of each process, and return the data to the data acquisition module.

[0015] The present invention discloses the following technical effects: This invention provides a production control method and system for an alumina decomposition process. Through collaborative prediction and rolling optimization, it solves the problems of slow response and large temperature differences in manual adjustment, achieving precise collaborative control of the temperatures of the first and last tanks. Through closed-loop speed control, it solves the problems of large metering errors and inability to dynamically adjust manually added additives, achieving precise control of the additive dosage. Through incremental PID control of the air valve opening, it solves the problems of slow response, overflow, or empty tanks in manual adjustment, achieving stable material feeding and air pressure protection. Through BP neural network prediction and genetic algorithm optimization, it solves the problem of relying on experience for adjustment, achieving prediction of the decomposition rate and multi-parameter collaborative optimization. Through bidirectional interaction and linkage models between upstream and downstream processes, it solves the problems of information silos between processes and delays in manual communication, achieving material and energy matching between decomposition, roasting, and evaporation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the production control process for the alumina decomposition step provided in an embodiment of the present invention; Figure 2 A system layered architecture diagram provided for embodiments of the present invention; Figure 3 This is a schematic diagram of multivariate predictive control of the first and last tank temperatures provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the dynamic prediction and optimization of decomposition rate provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of cross-process collaborative control provided in an embodiment of the present invention. Detailed Implementation

[0018] 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.

[0019] The purpose of this invention is to provide a production control method and system for the alumina decomposition process, which solves the problems existing in the current alumina decomposition process, such as low temperature control accuracy, large error in crystallization aid addition, lag in solid content detection, untimely adjustment of feed rate, lack of basis for decomposition rate optimization, and poor cross-process coordination.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] Figure 1 This is a schematic diagram of the production control process for the alumina decomposition step provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the present invention provides a production control method for an alumina decomposition process, comprising: Step 100: Collect real-time operating parameters of the decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data; Step 200: Based on the real-time process data, the temperature of the first tank and the temperature of the last tank are jointly predicted and rolled optimized using a pre-built state-space model and a multivariate predictive control algorithm to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. Step 300: Set the target amount of crystallization aid according to the feed flow rate, and adjust the speed of the metering pump according to the target amount of crystallization aid using closed-loop control; Step 400: Based on the liquid level difference between the decomposition tank and the feeding tank, the opening of the feeding air valve is adaptively adjusted using incremental PID control. Step 500: Adjust the concentration of the original decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-mother liquor. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. Step 600: Update the optimal control parameter sequence and the optimized process parameter combination to the equipment of the decomposed process node; Step 700: Based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination, adjust the upstream evaporation process and the downstream roasting process respectively using the roasting linkage model and the evaporation linkage model; Step 800: Based on the actual operating data and production indicators of each process, extract the predicted deviation, use the feedback correction mechanism to update the processing model of each process, and return to step "Collect the real-time operating parameters of the decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data".

[0022] refer to Figure 2This embodiment adopts a layered architecture combined with closed-loop control design. The system is divided into four layers from top to bottom: perception layer, control layer, model layer, and collaborative optimization layer. 1) The perception layer is responsible for real-time collection of all key process parameters and equipment operating status data for the decomposition process. Parameters such as temperature, flow rate, liquid level, and pressure are collected at high frequencies, while parameters such as solid content are collected periodically. The data is aggregated by the control layer and forwarded to the model layer as needed. Specifically: Flow detection unit: Electromagnetic flow meters are installed in the inlet and outlet pipes of the plate cooling water; electromagnetic flow meters are also installed in the feed pipe of the decomposed raw liquid and the mother liquor pipe.

[0023] Liquid level detection unit: A radar level gauge is installed at the center of the top of the decomposition tank (avoiding the feed inlet and agitator); radar level gauges are also installed on the top of the crystallization aid storage tank, the feed tank, the filtrate storage tank, and the mother liquor storage tank.

[0024] Solid content and composition detection unit: A horizontal telescopic solid content detection device is installed on the sidewall of the first, middle, and last decomposition tanks of each sequence. This device consists of an electric horizontal telescopic mechanism, a gamma-ray concentration detection probe, and a high-pressure water automatic cleaning unit. The probe extends horizontally into the tank to a specified depth for detection, and the built-in calculation module directly outputs the solid content value. An online particle size analyzer is installed on the decomposition tank discharge pipe; an online sodium aluminate solution concentration analyzer is installed on the decomposition concentrate feed pipe; and an online mother liquor α-ray analyzer is installed on the mother liquor feed pipe. k Online analyzer.

[0025] Temperature and pressure detection unit: Multiple platinum resistance temperature sensors are installed in layers at different heights on the walls of the first and last tanks; pressure sensors are installed on the material lifting duct.

[0026] 2) The control layer is responsible for receiving control commands from the model layer and driving the actuators to move. It also undertakes the initial aggregation and forwarding of data from the perception layer, including: DCS system: It enables data interaction with the perception layer, model layer and collaborative optimization layer through authorized programming, and has functions such as data storage, real-time display, audible and visual alarms and one-click switching between manual and automatic modes.

[0027] Actuators: Hardwired connection to the DCS system, including: frequency conversion metering pump (for adding crystallization aids); servo motor driven electric regulating air valve (for adjusting material feeding rate); electric regulating valve (for adjusting plate cooling water flow rate and mother liquor flow rate).

[0028] 3) Model Layer: Deployed on an independent APC (Advanced Process Control) platform, responsible for data preprocessing, model calculation, and control instruction generation. Data preprocessing module: Performs moving average filtering, noise reduction, and 3σ criterion outlier removal on the raw data collected by the perception layer.

[0029] Algorithm module: Includes multivariate prediction and control module for first and last tank temperatures, decomposition rate prediction and optimization module, incremental PID closed-loop control module for feed rate, and crystallization aid ratio control module.

[0030] Human-Machine Interface (HMI) module: Through a visual operating interface, it displays real-time process parameter curves, equipment operating status, control suggestions and alarm information, and supports operators to modify parameters, query historical data and switch control modes.

[0031] 4) The collaborative optimization layer enables cross-system data interaction and collaborative control between the decomposition process and the upstream evaporation and downstream roasting processes, breaking down information silos between processes. Based on the data from cross-process interactions, the collaborative control module automatically generates process parameter adjustment instructions for the decomposition process, ensuring material and energy balance throughout the entire process.

[0032] Preferably, real-time data acquisition and preprocessing are performed throughout the entire process. Each sensor in the sensing layer and the online detection device acquires raw data at a set frequency: rapidly changing parameters such as temperature, flow rate, liquid level, and pressure are acquired at high frequency; slowly changing parameters such as solid content are acquired periodically; and particle size, solution concentration, and mother liquor α are acquired at regular intervals. k Parameters are collected at a frequency of minutes. All raw data are first transmitted to the control layer DCS system, and then forwarded to the data preprocessing module of the model layer. Moving average filtering, 3σ criterion outlier removal and data normalization are performed in sequence to remove noise and outliers. After removing noise and outliers, the data is synchronously stored in the historical database of the DCS system and input into the various algorithm modules of the model layer.

[0033] refer to Figure 3 Multivariable predictive control (MPC) for first and last tank temperatures: The control is executed by the multivariate predictive control module for the first and last tank temperatures in the model layer, with the control objective being to maintain the temperature difference between the first and last tanks. (in, , (These correspond to the temperatures of the first and last tanks, respectively, and are within the range required by the process, while ensuring the stability of the temperature in the last tank.) The module uses the pretreated mother liquor flow rate... Plate cooling water flow rate Filtrate flow rate Decomposition temperature of the original solution Ambient temperature As input, the output after calculation is the mother liquor flow rate adjustment amount. And plate cooling water flow rate adjustment The control is transmitted to the corresponding electric regulating valve via the DCS system. The flow rate of filtrate from the upstream filtration process into the decomposition system is collected in real time by an electromagnetic flow meter installed in the filtrate pipeline; The temperature of the decomposition concentrate entering the first tank is collected in real time by a temperature sensor in the concentrate feed pipe. The ambient temperature of the workshop where the decomposition process is located is collected in real time by ambient temperature sensors in the tank area. The specific steps are as follows: S1: Based on sufficient historical production data, the correlation coefficients between each factor and the temperatures of the first and last tanks were calculated using Pearson correlation analysis. Cooling water flow rate showed a strong correlation with temperature, mother liquor flow rate and decomposition concentrate temperature showed a moderate correlation, and filtrate flow rate and ambient temperature showed a weak correlation. The above correlation analysis results are used to help understand the degree of influence of each variable. The actual dynamic transmission characteristics of the perturbation variables are precisely described by the coefficient matrix in the state-space model of S2.

[0034] S2: A second-order discrete state-space prediction model for the temperatures of the first and last tanks is constructed using a data-based system identification method. The first tank temperature is selected. and the temperature of the last tank As a state variable ; based on mother liquor flow rate and plate cooling water flow For manipulators ; based on filtrate flow rate Decomposition temperature of the original solution and ambient temperature For measurable disturbances ; based on the temperature of the first tank and the temperature of the last tank For controlled output The model expression is as follows: .

[0035] .

[0036] in, , , , The coefficient matrix is ​​to be identified.

[0037] Coefficient determination process: Under steady-state conditions, the mother liquor flow rate was determined separately. and plate cooling water flow A small step signal is applied, and the temperature response curves of the first and last tanks are recorded with a high-frequency sampling period until the temperature reaches a steady state again. Using the collected input and output data, a subspace identification algorithm, such as the N4SID algorithm, is employed to solve for the coefficient matrix. , , , The least squares optimal estimate is obtained, and the resulting model is downsampled to the target control period.

[0038] S3: Rolling Optimization and Feedback Correction: In each control cycle, based on the above state-space model, predict the temperature output trajectory over a future period (predicting P cycles in the time domain), and perform rolling optimization with the goal of minimizing the following performance indicators: .

[0039] in, This is a temperature setpoint vector, set to maintain the temperature difference between the first and last tanks. Within the limits of process requirements; To output the error weight matrix, a larger weight is given to the last tank to emphasize the stable control of the temperature in the last tank, while also taking into account the temperature difference regulation. To control the incremental weight matrix; For control in the time domain, the current optimal control increment is obtained by solving. After limiting, the output is the mother liquor flow rate adjustment amount. And plate cooling water flow rate adjustment The control system sends the commands to the electric regulating valve for execution. In the next control cycle, the model state is updated using real-time temperature feedback, and the above optimization process is repeated.

[0040] Specifically, the precise and automatic addition control of the crystallization aid is executed by the crystallization aid ratio control module in the model layer. The feed flow rate of the pretreated decomposition tank... Enter the proportional control model. The total slurry flow rate entering the decomposition tank, i.e., the mixed feed flow rate of the decomposition raw liquid and the circulating mother liquor, is monitored in real time by an electromagnetic flowmeter in the feed pipeline. A linear proportional relationship is established between the amount of crystallization aid added and the feed flow rate. .

[0041] in, The target addition amount of crystallization aid. To add a scaling factor, The feed flow rate to the decomposition tank is then used. Subsequently, based on the metering pump's calibration curve, the target feed rate is converted into the metering pump's target rotational speed. .

[0042] in, For the volumetric efficiency of the metering pump, The displacement per revolution of the metering pump is given. A PI algorithm is then used for closed-loop control of the metering pump speed. Based on the deviation between the target speed and the actual speed, a speed adjustment is output, which drives the frequency converter via the DCS system to control the operation of the metering pump. The sensing layer collects the actual speed of the metering pump in real time and feeds it back to the PI controller, forming a speed closed loop.

[0043] The radar level gauge in the crystallization aid storage tank of the sensing layer collects the liquid level in real time and sets a primary alarm threshold and a secondary alarm threshold (primary for warning and secondary for danger). When the liquid level reaches the primary threshold, the DCS system issues an audible and visual alarm; when the liquid level reaches the secondary threshold, the system automatically outputs a stop command and cuts off the power to the metering pump.

[0044] Optionally, online monitoring and automatic maintenance of slurry solids content are performed collaboratively by a horizontally telescopic solids content detection device in the sensing layer, along with the control and model layers. The device detects slurry solids content based on the principle of gamma-ray attenuation. The detection probe, driven by an electrically operated horizontal telescopic mechanism, extends horizontally into the decomposition tank to a specified depth for online detection. The gamma-ray detection signal is processed by the device's built-in calculation module and directly outputs the solids content value, which is then transmitted to the DCS system for display and storage. The workflow is as follows: Detection is performed according to a set cycle. The electrically operated horizontal telescopic mechanism drives the detection probe to extend into the tank at a low speed to a specified depth. After completing the detection, the probe retracts quickly into the tank wall protective sleeve. The high-pressure water automatic cleaning unit then starts to rinse the surface. A comprehensive deep cleaning is automatically performed at regular intervals. When the deviation of the solids content value obtained from multiple consecutive detections exceeds the preset allowable range, the DCS system issues an equipment malfunction alarm, prompting the operator to perform calibration or maintenance.

[0045] Furthermore, the adaptive closed-loop adjustment of the feed rate is executed by the incremental PID closed-loop control module of the model layer. This module receives the pre-processed liquid level data of the decomposition tank and the feed tank, calculates the liquid level difference between the two in real time, and sets the target liquid level difference. The controller input is a normalized liquid level difference deviation signal, which is then used to calculate the liquid level deviation. The output is the valve opening command. The incremental PID control algorithm is as follows: .

[0046] in, , , These are the proportional coefficient, integral coefficient, and derivative coefficient, respectively. These parameters are determined through on-site debugging. Deviation signal. The definition of is: .

[0047] in, For the target liquid level difference, This represents the actual liquid level difference between the decomposition tank and the feeding tank at the current moment. For the liquid level difference measurement range, the actual liquid level difference deviation is normalized to the standard signal range.

[0048] The calculated opening adjustment amount Δθ(k) is added to the opening at the previous moment to form the current valve opening command, which is then sent to the servo motor to drive the electric valve for execution via the DCS system. The sensing layer collects the actual liquid levels in the decomposition tank and the feeding tank in real time and feeds them back to the PID controller to form a liquid level closed loop.

[0049] The sensing layer collects the pressure of the material feeding duct in real time. When the pressure exceeds the preset safety threshold, the system automatically limits the upper limit of the valve opening and issues an alarm for excessive pressure.

[0050] refer to Figure 4 The dynamic prediction and optimization of the decomposition rate is performed by the decomposition rate prediction and optimization module of the model layer, based on the concentration of the pretreated decomposition solution and the temperature of the first tank. Final tank temperature 1. Slurry solids content, seed crystal addition amount (or seed crystal solids content), mother liquor The value is used as input.

[0051] S1: A backpropagation (BP) neural network is used to construct a decomposition rate prediction model. The number of nodes in the input layer corresponds to the number of input parameters, and the output layer contains the predicted decomposition rate. Mother liquor This refers to the caustic ratio of the sodium aluminate solution, specifically the molar ratio of caustic sodium oxide to aluminum oxide in the solution. All input parameters are normalized to [value missing]. The intervals correspond to their actual production and operation ranges. The model is trained using the LM algorithm, with the training dataset consisting of historical production data over a sufficient period, proportionally divided into training, validation, and test sets. During runtime, preprocessed key parameters are input into the BP neural network, and the model outputs a predicted decomposition rate. .

[0052] S2: Genetic Algorithm for Process Parameter Optimization: Using the maximization of the decomposition rate output by the decomposition rate prediction model as the objective function, and the allowable fluctuation range of each process parameter as a constraint, a genetic algorithm is used for global optimization: Objective function: The constraints are the allowable adjustment ranges of each parameter around the current operating value. The adjustment ranges of each parameter are reasonably set based on process experience and equipment capabilities. The genetic algorithm parameters (population size, crossover probability, mutation probability, number of generations) are selected according to the size and complexity of the optimization problem.

[0053] S3: The optimal combination of process parameters obtained by the genetic algorithm is pushed to the HMI interface in the form of visual suggestions. After the operator confirms, the system automatically sends the optimal parameters to the corresponding control modules (first and last tank temperature MPC module, crystallization aid ratio control module, feed rate PID control module, etc.) for execution; or it can be set to automatic execution mode according to production needs, without manual confirmation.

[0054] refer to Figure 5Cross-process collaborative control is executed by the OPCUA communication interface and collaborative control module of the collaborative optimization layer. Data interaction using the OPCUA protocol enables low-latency and highly reliable data transmission. The decomposed process DCS system achieves bidirectional data interaction with upstream and downstream process control systems through the OPCUA interface.

[0055] 1) Integration with the roasting process: The decomposition process transmits the slurry moisture content to the roasting process in real time. And decompose the slurry particle size data. The roasting process adjusts the roasting furnace temperature according to the slurry moisture content: .

[0056] in, As the reference roasting temperature, As the baseline moisture content, This is the temperature regulation coefficient, dynamically determined based on the degree of moisture deviation. The roasting process feeds back the roasting furnace's feed capacity to the decomposition process via the OPCUA interface. This value is calculated online by the DCS system of the calcination process based on real-time thermal conditions and equipment constraints. The specific determination logic is as follows: .

[0057] .

[0058] .

[0059] .

[0060] in, The maximum throughput capacity of the machinery is determined by the maximum linear velocity of the feeding equipment in the calcining furnace. Cross-sectional area of ​​feeding equipment and slurry density The decision will be finalized during equipment commissioning; The maximum heat load processing capacity is determined by the current available heat power of the roasting furnace. Calcination temperature setting value Incoming material temperature Moisture content of slurry The latent heat of vaporization of water Calculated in real time; This is the specific heat capacity of water; The temperature of the slurry entering the roasting furnace is detected online by a temperature sensor in the furnace feed pipe. To achieve maximum ventilation capacity, the exhaust fan operates at its maximum airflow. and the coefficient of flue gas production per unit moisture It is calculated in real time.

[0061] when When the material feed rate is less than the actual feed rate of the decomposition process, the collaborative control module automatically generates an adjustment command to reduce the feed rate, adjusting the feed rate setpoint to... The data is then sent to the PID control module for feeding, which reduces the feeding rate to a level that matches the feeding capacity of the roasting furnace.

[0062] 2) Integration with the evaporation process: The decomposition process transmits the required amount of mother liquor and the target amount of mother liquor to the evaporation process in real time. Value. The evaporation rate is adjusted according to the required amount of mother liquor in the evaporation process: .

[0063] in, The mother liquor loss is dynamically determined based on the liquid level in the decomposition tank. The evaporation process feeds back the actual mother liquor loss to the decomposition process via the OPCUA interface. value When the actual mother liquor With target value When the deviation exceeds the preset threshold, the collaborative control module automatically generates a first-tank temperature adjustment command: .

[0064] in, Adjustment amount for the setpoint temperature of the first tank; The actual mother liquor fed back from the evaporation process value; To meet the target mother liquor requirements of the decomposition process value; for - Temperature conversion coefficient, i.e., mother liquor The amount of initial tank temperature adjustment required for each unit deviation from the target value. Increasing the initial tank temperature accelerates the decomposition rate and makes the mother liquor... The temperature of the mother liquor decreases, and vice versa; therefore, the temperature of the first tank can be adjusted to regulate the mother liquor in reverse. For small deviations, use a smaller adjustment coefficient to avoid over-adjustment; for large deviations, increase the adjustment coefficient to correct them quickly.

[0065] generated The command is sent to the MPC control module for the first and last tank temperatures and executed by superimposing the first tank temperature setpoint.

[0066] Optionally, the system includes fault handling and maintenance. When equipment malfunctions or key parameters deviate beyond the set range, the system automatically issues an audible and visual alarm and displays the fault location and cause on the HMI interface. Operators can switch to manual control mode with a single click for manual intervention. After the fault is cleared, the system automatically reverts to automatic control mode. The system automatically records the operating time and maintenance cycle of each device, prompting operators to perform maintenance when due. The horizontal telescopic solids content detection device automatically performs a comprehensive cleaning at a set cycle, and each sensor is calibrated periodically according to regulations.

[0067] As an optional implementation, this embodiment also provides a production control system for the alumina decomposition process, including: The data acquisition module is used to collect real-time operating parameters of the decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data. The rolling optimization module is used to perform collaborative prediction and rolling optimization of the first tank temperature and the last tank temperature based on the real-time process data using a pre-built state-space model and a multivariate predictive control algorithm, so as to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. The speed regulation module is used to set the target addition amount of crystallization aid according to the feed flow rate, and to regulate the speed of the metering pump according to the target addition amount of crystallization aid using closed-loop control. The opening adjustment module is used to adaptively adjust the opening of the lifting air valve based on the liquid level difference between the decomposition tank and the lifting tank using incremental PID control. The process parameter optimization module is used to optimize parameters such as the concentration of the decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. The parameter update module is used to update the optimal control parameter sequence and the optimized process parameter combination to the equipment of the decomposed process node; The linkage control module is used to adjust the upstream evaporation process and the downstream roasting process respectively using the roasting linkage model and the evaporation linkage model according to the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination; The closed-loop control module is used to extract predicted deviations based on the actual operating data and production indicators of each process, use the feedback correction mechanism to update the processing model of each process, and return the data to the data acquisition module.

[0068] The beneficial effects of this invention are as follows: This invention achieves precise coordinated control of the temperatures of the first and last tanks through collaborative prediction and rolling optimization, reducing temperature deviation and suppressing the impact of disturbances; it achieves precise control of the additive dosage through closed-loop speed control, eliminating the risk of overflow or empty tanks; it achieves stable feeding and air pressure protection through incremental PID adjustment of the air valve opening; it achieves prediction of decomposition rate and multi-parameter collaborative optimization through BP neural network prediction and genetic algorithm optimization, which can stably improve the decomposition rate; and it achieves material and energy matching between decomposition, roasting, and evaporation through bidirectional interaction and linkage model between upstream and downstream processes.

[0069] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0070] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A production control method for an alumina decomposition process, characterized in that, include: Collect real-time operating parameters of decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data; Based on the real-time process data, the temperature of the first tank and the temperature of the last tank are jointly predicted and rolled optimized using a pre-built state-space model and a multivariate predictive control algorithm to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. The target amount of crystallization aid is set according to the feed flow rate, and the metering pump speed is adjusted using closed-loop control according to the target amount of crystallization aid. The opening of the lifting air valve is adaptively adjusted using incremental PID control based on the liquid level difference between the decomposition tank and the lifting tank. The concentration of the original decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor were all considered. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. The optimal control parameter sequence and the optimized process parameter combination are updated to the equipment at the decomposed process node; Based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination, the upstream evaporation process and the downstream roasting process are adjusted using the roasting linkage model and the evaporation linkage model, respectively. Based on the actual operating data and production indicators of each process, the predicted deviation is extracted and the feedback correction mechanism is used to update the processing model of each process. Then, the process returns to the step of "collecting the real-time operating parameters of the decomposed process nodes, preprocessing the real-time operating parameters, and obtaining real-time process data".

2. The production control method for an alumina decomposition process according to claim 1, characterized in that, Based on the real-time process data, a pre-built state-space model and a multivariate predictive control algorithm are used to collaboratively predict and roll-optimize the temperatures of the first and last tanks, resulting in an optimal control parameter sequence that satisfies temperature difference constraints and temperature stability requirements. This sequence includes: Historical production data were analyzed using Pearson correlation analysis to identify key perturbation variables and key manipulated variables. The key perturbation variables included: filtrate flow rate, decomposition concentrate temperature, and ambient temperature. The key manipulated variables included: mother liquor flow rate and plate cooling water flow rate. A second-order discrete state-space prediction model with the first tank temperature and the last tank temperature as state variables is constructed, and the second-order discrete state-space prediction model is identified offline using a subspace identification algorithm. A minimization objective is set, and the second-order discrete state-space prediction model is rolled over based on the minimization objective to obtain the optimal control parameter sequence corresponding to the key manipulated variable.

3. The production control method for an alumina decomposition process according to claim 2, characterized in that, The minimization objective is: ;in, The objective function is... For prediction in the time domain; This is the output of the i-th future step predicted at time k; Set a value for the output of the i-th step in the future; This represents the output error weight matrix; Indicates control over the time domain; This represents the control increment predicted at time k for the j-th future step. To control the incremental weight matrix.

4. The production control method for an alumina decomposition process according to claim 1, characterized in that, The target addition amount of crystallization aid is set according to the feed flow rate, and the metering pump speed is adjusted using closed-loop control based on the target addition amount of crystallization aid, including: The feed flow rate is input into a proportional control model for calculation to obtain the target addition amount of the crystallization aid; the expression of the proportional control model is: ;in, The target addition amount of the crystallization aid; To add a scaling factor; The feed flow rate is [value missing]. The target addition amount of the crystallization aid is converted into the target rotational speed of the metering pump based on the calibration curve of the metering pump. The actual speed of the metering pump is adjusted in a closed loop using a PI control algorithm until the actual speed of the metering pump reaches the target speed and then the speed is maintained. Monitor the liquid level in the crystallization aid storage tank. When the liquid level in the storage tank reaches the first-level threshold, trigger an alarm. When the liquid level in the storage tank reaches the second-level threshold, cut off the power supply to the metering pump.

5. The production control method for an alumina decomposition process according to claim 1, characterized in that, The expression for the incremental PID control is: ;in, This represents the current adjustment amount of the damper opening. This is the proportionality coefficient; This is the normalized liquid level difference deviation signal at the current moment; This is the liquid level difference deviation signal from the previous moment; The integral coefficient; These are the differential coefficients; This represents the liquid level difference deviation signal between the first two time points.

6. The production control method for an alumina decomposition process according to claim 1, characterized in that, The concentration of the original decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor were all considered. k The input values ​​are used to a BP neural network to predict the decomposition rate, and the process parameters are optimized with the goal of maximizing the decomposition rate, resulting in an optimal combination of process parameters, including: The normalized concentration of the decomposition stock solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the amount of mother liquor α are used to determine the optimal concentration of the stock solution. k The input values ​​are used to train a three-layer BP neural network to obtain the predicted decomposition rate. The objective function is to maximize the predicted decomposition rate. With the allowable fluctuation range of each input parameter as a constraint, a genetic algorithm is used to perform global optimization based on the objective function to obtain the optimized combination of process parameters.

7. The production control method for an alumina decomposition process according to claim 1, characterized in that, The expression for the roasting linkage model is: ; The expression for the evaporation linkage model is: ;in, The adjusted temperature setpoint for the roasting furnace; The reference calcination temperature; This is the temperature regulation coefficient; Real-time slurry moisture content; Based on the moisture content; The amount of evaporation that needs to be adjusted for the evaporation process; This represents the amount of mother liquor required for the decomposition process; This represents the amount of mother liquor lost.

8. The production control method for an alumina decomposition process according to claim 1, characterized in that, Based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination, the upstream evaporation process and the downstream roasting process are adjusted using the roasting linkage model and the evaporation linkage model, respectively, including: The moisture content of the slurry and the particle size of the decomposed slurry are input into the roasting linkage model for calculation to obtain the roasting furnace adjustment temperature. The feeding capacity of the roasting furnace is determined based on the real-time thermal state and equipment constraints of the roasting process. When the feeding capacity of the roasting furnace is less than the feeding amount of the decomposition process, the feeding amount is reduced to a level that matches the feeding capacity of the roasting furnace. The required amount of mother liquor and the target mother liquor α k The value is input into the evaporation linkage model for calculation to obtain the evaporation adjustment amount; When the actual value fed back from the evaporation process deviates from the target value by more than the limit, the temperature of the first tank is adjusted using the set temperature coordination adjustment formula.

9. The production control method for an alumina decomposition process according to claim 8, characterized in that, The temperature coordination adjustment formula is as follows: ;in, The adjustment amount for the initial tank temperature setpoint; This is the temperature conversion coefficient; The actual mother liquor α fed back from the evaporation process k value; To meet the target mother liquor α requirements of the decomposition process k value.

10. A production control system for an alumina decomposition process, characterized in that, A production control method for implementing the alumina decomposition process as described in claim 1 includes: The data acquisition module is used to collect real-time operating parameters of the decomposed process nodes, preprocess the real-time operating parameters, and obtain real-time process data. The rolling optimization module is used to perform collaborative prediction and rolling optimization of the first tank temperature and the last tank temperature based on the real-time process data using a pre-built state-space model and a multivariate predictive control algorithm, so as to obtain the optimal control parameter sequence that meets the requirements of temperature difference constraint and temperature stability. The speed regulation module is used to set the target addition amount of crystallization aid according to the feed flow rate, and to regulate the speed of the metering pump according to the target addition amount of crystallization aid using closed-loop control. The opening adjustment module is used to adaptively adjust the opening of the lifting air valve based on the liquid level difference between the decomposition tank and the lifting tank using incremental PID control. The process parameter optimization module is used to optimize parameters such as the concentration of the decomposition solution, the temperature of the first and last tanks, the solid content of the slurry, the amount of seed crystals added, and the α-coated mother liquor. k The input values ​​are used to predict the decomposition rate through a BP neural network, and the process parameters are optimized with the goal of maximizing the decomposition rate to obtain the optimal combination of process parameters. The parameter update module is used to update the optimal control parameter sequence and the optimized process parameter combination to the equipment of the decomposed process node; The linkage control module is used to adjust the upstream evaporation process and the downstream roasting process respectively using the roasting linkage model and the evaporation linkage model based on the optimal control parameter sequence and the cross-process linkage parameters in the optimized process parameter combination; The closed-loop control module is used to extract predicted deviations based on the actual operating data and production indicators of each process, use the feedback correction mechanism to update the processing model of each process, and return the data to the data acquisition module.

Images