Lithium iron phosphate roasting synthesis control system and method based on intelligent learning and medium

Abstract

The invention discloses a lithium iron phosphate roasting synthesis control system and method based on intelligent learning and a medium, and relates to the field of lithium iron phosphate, the method comprises the following steps: obtaining material parameters of lithium iron phosphate entering a kiln, the material parameters comprising material water content and material mass; acquiring current working condition parameters of the kiln; based on the material parameters and the current working condition parameters, the temperature of the lithium iron phosphate at the future moment is predicted, and the predicted material temperature is obtained; determining the temperature deviation degree of the predicted material temperature relative to the target material temperature; and adjusting a temperature control device of the kiln according to the temperature deviation degree. By implementing the application, the temperature of the lithium iron phosphate at the future moment can be predicted through the material parameters of the lithium iron phosphate entering the kiln and the current working condition parameters of the kiln, and the temperature control is performed in advance based on the temperature deviation degree at the future moment, so that overlarge temperature deviation caused by temperature adjustment lag is avoided, and the service life of the kiln is prolonged. Therefore, the high quality of the roasted lithium iron phosphate material is ensured.

Description

Technical Field

[0001] This application relates to the field of lithium iron phosphate technology, specifically to a lithium iron phosphate roasting synthesis control system, method, and medium based on intelligent learning. Background Technology

[0002] Lithium iron phosphate (LiFePO4), as a core cathode material for power batteries, has key electrochemical properties such as crystal structure, particle uniformity, specific capacity, and cycle life that are decisively related to the precision of temperature control during the calcination process. The ideal temperature range for crystal phase formation is typically very narrow. Too low a temperature can lead to incomplete reactions, forming impurity phases and affecting the purity and performance of the material; too high a temperature may cause lithium volatilization and iron reduction, producing inactive impurity phases, resulting in decreased battery capacity and reduced cycle life. Therefore, the calcination process of lithium iron phosphate has stringent temperature requirements.

[0003] In related technologies, temperature control during the calcination process of lithium iron phosphate is typically achieved using a PID (Proportional Integral Derivative) algorithm. However, as a lag-correction algorithm, the PID algorithm essentially corrects based on the current deviation, which can easily lead to temperature overshoot or undershoot, causing the temperature to deviate from the ideal crystal formation temperature range and making it difficult to meet the quality requirements of lithium iron phosphate materials. Summary of the Invention

[0004] This application provides a lithium iron phosphate roasting synthesis control system, method, and medium based on intelligent learning. It aims to achieve feedforward intelligent control of temperature during the roasting process by accurately predicting the future temperature of the material, thereby overcoming the problem of temperature control lag and improving product quality and production stability.

[0005] In a first aspect, this application provides a lithium iron phosphate roasting synthesis control method based on intelligent learning. The lithium iron phosphate roasting synthesis control method based on intelligent learning includes: acquiring material parameters of lithium iron phosphate entering the kiln, wherein the material parameters include material moisture content and material mass; acquiring the current operating parameters of the kiln; predicting the temperature of lithium iron phosphate at a future time based on the material parameters and the current operating parameters to obtain the predicted material temperature; determining the degree of temperature deviation of the predicted material temperature relative to the target material temperature; and adjusting the temperature control device of the kiln according to the degree of temperature deviation.

[0006] In the above embodiments, specific parameters of the material entering the kiln and the current operating conditions of the kiln are acquired. Based on these real-time and changing inputs, the material temperature at future moments is predicted. This process enables the intelligent learning-based lithium iron phosphate roasting synthesis control system to anticipate potential temperature deviations caused by changes in material characteristics or operating conditions, and to intervene and adjust in advance based on the predicted degree of deviation. This feedforward control mode overcomes the hysteresis defect of waiting for the actual occurrence of deviations before responding, effectively avoiding temperature overshoot or undershoot during the roasting process, suppressing temperature fluctuations within a minimal range, thereby improving the quality and batch consistency of lithium iron phosphate materials.

[0007] In conjunction with some embodiments of the first aspect, in some embodiments, after obtaining the material parameters of lithium iron phosphate entering the kiln, the method further includes: determining whether the material parameters meet preset cold lumps material determination conditions, wherein the cold lumps material determination conditions include at least one of material moisture content being greater than a preset moisture content threshold and material mass being greater than a preset mass threshold; if the material parameters meet the cold lumps material determination conditions, then the step of obtaining the current operating parameters of the kiln is performed.

[0008] In the above embodiments, the interference sources that have the greatest impact on the roasting temperature are specifically identified, namely, cold lumps of material with high moisture content or large mass. By setting preset judgment conditions, these batches of materials that cause drastic temperature drops can be processed first, focusing limited computing resources and control capabilities on the most critical moments. This avoids making complex predictions indiscriminate for all materials, making the control strategy more accurate and efficient.

[0009] In conjunction with some embodiments of the first aspect, in some embodiments, the temperature of lithium iron phosphate at a future time is predicted based on material parameters and current operating condition parameters to obtain a predicted material temperature. This includes: using a temperature prediction model based on material parameters and current operating condition parameters to predict the temperature change curve of lithium iron phosphate after it enters the kiln, thereby obtaining a predicted temperature change curve. The temperature prediction model is learned based on historical experience data, which includes at least historical operating condition parameters of the kiln; determining the difference points between the predicted temperature change curve and the preset roasting temperature curve; and determining the predicted material temperature based on the difference points.

[0010] In the above embodiments, a temperature prediction model learned from historical experience data is used to predict the complete temperature change trend curve of the material throughout the roasting process. The predicted temperature change curve is compared with the ideal roasting temperature curve to gain a more comprehensive understanding of the magnitude and duration of future temperature deviations, identify the most critical differences, and thus formulate a more forward-looking and holistic optimization control strategy.

[0011] In conjunction with some embodiments of the first aspect, in some embodiments, determining the predicted material temperature based on the difference point includes: obtaining the temperature difference between the corresponding temperature of the difference point in the predicted temperature change curve and the corresponding temperature in the calcination temperature curve; if the temperature difference is greater than a preset temperature difference threshold, then the corresponding temperature of the difference point in the predicted temperature change curve is taken as the predicted material temperature, wherein the corresponding temperature of the difference point in the calcination temperature curve is the target material temperature.

[0012] In the above embodiments, a temperature difference threshold judgment mechanism is introduced, which effectively filters out the tiny fluctuations or noises that may exist in the predicted material temperature and have a negligible impact on the actual roasting process. It only focuses on those temperature deviations that exceed the threshold and have a significant impact, avoiding over-adjustment of unnecessary tiny deviations, thereby enhancing the stability and robustness of the entire intelligent learning-based lithium iron phosphate roasting synthesis control system.

[0013] In conjunction with some embodiments of the first aspect, in some embodiments, adjusting the temperature control device of the kiln according to the degree of temperature deviation includes: determining a target adjustment time and a target parameter adjustment amount of the temperature control device based on the degree of temperature deviation; and adjusting the temperature control device according to the target parameter adjustment amount at the target adjustment time.

[0014] In the above embodiments, the control strategy is concretized into precise timing and adjustment amounts, that is, the optimal timing and intensity of the adjustment action to counteract temperature deviations are precisely calculated. This precise "timed and quantitative" intervention fully considers the thermal inertia and response delay of the kiln, ensuring that the adjustment action takes effect precisely when the temperature deviation is about to occur, minimizing the fluctuation range of the actual temperature, and achieving efficient and precise feedforward control.

[0015] In conjunction with some embodiments of the first aspect, in some embodiments, determining the target adjustment time and target parameter adjustment amount of the temperature control device based on the degree of temperature deviation includes: acquiring a prediction model of the adjustment strategy of the temperature control device, wherein the prediction model of the adjustment strategy is used to reduce the degree of temperature deviation; and using the prediction model of the adjustment strategy to determine the target adjustment strategy of the temperature control device based on the degree of temperature deviation, wherein the target adjustment strategy includes the target adjustment time and the target parameter adjustment amount.

[0016] In the above embodiments, the complex decision-making process of how to formulate the adjustment strategy is entrusted to a dedicated adjustment strategy prediction model. Based on the current predicted temperature deviation, the optimal adjustment strategy is generated intelligently and automatically, enabling the control strategy to have self-learning and self-adaptive capabilities, cope with more complex operating condition changes, and improve the intelligence level of the lithium iron phosphate roasting synthesis control system based on intelligent learning.

[0017] In conjunction with some embodiments of the first aspect, in some embodiments, the temperature control device includes at least one of a natural gas valve, a rotary motor, a preheating fan, and a cooling water pump. The target parameter adjustment amount includes at least one of a valve opening adjustment amount, a motor frequency adjustment amount, a fan frequency adjustment amount, and a water pump frequency adjustment amount. Adjusting the temperature control device according to the target parameter adjustment amount includes at least one of the following steps: adjusting the opening of the natural gas valve according to the valve opening adjustment amount; adjusting the frequency of the rotary motor according to the motor frequency adjustment amount; adjusting the frequency of the preheating fan according to the fan frequency adjustment amount; and adjusting the frequency of the cooling water pump according to the water pump frequency adjustment amount.

[0018] In the above embodiments, coordinated control of multiple key actuators in the kiln is achieved. Faced with a predicted temperature deviation, instead of simply adjusting a single temperature control device, a comprehensive assessment and coordinated adjustment of multiple dimensions, including heat supply, material residence time, kiln atmosphere, and cooling rate, is implemented. This multi-variable coordinated optimization approach can more efficiently and comprehensively address complex temperature fluctuations, resolving the potential side effects or ineffectiveness of single adjustments, and achieving globally optimal control of the roasting process.

[0019] In conjunction with some embodiments of the first aspect, in some embodiments, after obtaining the material parameters of lithium iron phosphate entering the kiln, the method further includes: determining the feeding interval of the kiln; if the feeding interval is longer than the preset interval, reducing the frequency of the preheating fan of the kiln to reduce the flue gas reuse flow rate of the kiln.

[0020] In the above embodiments, compensation control was implemented for the special operating condition of the feeding interval. When the feeding pause time is too long, the kiln will experience a tendency for the background temperature to rise due to the lack of cold material entering. The lithium iron phosphate roasting synthesis control system based on intelligent learning monitors the feeding interval and actively reduces the frequency of the preheating fan and the flue gas reuse flow when the feeding interval is too long. In this way, excess flue gas can be discharged to maintain the stability of the kiln temperature.

[0021] Secondly, embodiments of this application provide a lithium iron phosphate roasting synthesis control system based on intelligent learning. The intelligent learning-based lithium iron phosphate roasting synthesis control system includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the intelligent learning-based lithium iron phosphate roasting synthesis control system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a lithium iron phosphate roasting synthesis control system based on intelligent learning, cause the lithium iron phosphate roasting synthesis control system based on intelligent learning to execute the method described in the first aspect and any possible implementation thereof.

[0023] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a lithium iron phosphate roasting synthesis control system based on intelligent learning, cause the intelligent learning-based lithium iron phosphate roasting synthesis control system to perform the method described in the first aspect and any possible implementation thereof.

[0024] Understandably, the intelligent learning-based lithium iron phosphate roasting synthesis control system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0025] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: by using the material parameters of lithium iron phosphate entering the kiln and the current operating parameters of the kiln, the temperature of lithium iron phosphate at future moments can be predicted, and based on the degree of temperature deviation at future moments, temperature control can be performed in advance to avoid excessive temperature deviation caused by temperature adjustment lag, so as to ensure the high quality of the roasted lithium iron phosphate material. Attached Figure Description

[0026] Figure 1 This is a schematic diagram showing the connection relationship of at least some structures in the lithium iron phosphate roasting synthesis control system based on intelligent learning in the embodiments of this application; Figure 2 This is a schematic flowchart of a lithium iron phosphate roasting synthesis control method based on intelligent learning in an embodiment of this application; Figure 3 This is another schematic diagram of the lithium iron phosphate roasting synthesis control method based on intelligent learning in the embodiments of this application; Figure 4 This is a schematic diagram of at least part of the physical device structure of the lithium iron phosphate roasting synthesis control system based on intelligent learning in the embodiments of this application.

[0027] in, Figure 1 The image is in color so that different objects can be distinguished using different colors. Figure 1 This is the image after being rotated 90 degrees counterclockwise. Detailed Implementation

[0028] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Figure 1 This is a schematic diagram of the connection relationship of at least some structures in the lithium iron phosphate roasting synthesis control system based on intelligent learning disclosed in the embodiments of this application. Through the connection relationship of at least some structures in the lithium iron phosphate roasting synthesis control system based on intelligent learning, at least part of the process flow of the lithium iron phosphate roasting production line is formed.

[0031] Reference Figure 1 The intelligent learning-based lithium iron phosphate roasting synthesis control system can be divided into a material conveying and processing unit, a rotary kiln roasting unit, a combustion and flue gas treatment unit, and a material cooling and collection unit. All units interact with the central control unit of the intelligent learning-based lithium iron phosphate roasting synthesis control system through sensors and actuators, enabling intelligent and automated control of the entire roasting process. The following will provide a detailed description of each unit and its internal components.

[0032] I. Material conveying and processing unit.

[0033] The material conveying and processing unit is responsible for accurately and stably feeding the roasting raw materials (such as lithium iron phosphate) into the kiln (such as a rotary kiln), and may include: Electronic unloading valve: Located at the top of the material conveying pipeline, powered by a corresponding motor ( Figure 1The electronic unloading valve (marked as M) is an actuator that receives switching commands from the central control unit to control the feeding of the roasting raw material from the upper silo to the downstream feed silo, serving as the starting point of the entire feeding process. Its operating frequency is one of the current operating parameters that can be adjusted by the intelligent learning-based lithium iron phosphate roasting synthesis control system in this embodiment.

[0034] Feed hopper ( Figure 1 The conical container located below the electronic discharge valve: used for temporary storage of the roasting raw materials to be roasted. This feed hopper is typically equipped with a level sensor (not shown in the figure) to monitor the material level and may be injected with nitrogen (indicated by the green arrow "nitrogen" in the figure) to protect the material from oxidation before it enters the kiln. The feed hopper's outlet is connected to the roasting feed piston.

[0035] The roasting feed piston, located below the feed hopper, is a key actuator that directly pushes the roasting raw materials into the rotary kiln and is driven by a motor. The reciprocating frequency and stroke of this feed piston together determine the feed rate per unit time. The operating current and operating frequency of this feed piston are key operating parameters that the central control unit monitors and regulates in real time to precisely control the feed rate to match the requirements of the roasting process.

[0036] Feed hood: This is a sealed structure connecting the roasting feed piston to the rotary kiln inlet. The pressure in the feed hood dust removal pipeline and the temperature in the feed hood flue gas pipeline are two parameters used to monitor the sealing status of the feed inlet and whether there is any high-temperature flue gas backflow. These are crucial monitoring points to ensure safe production and environmental quality.

[0037] II. Rotary Kiln Firing Unit.

[0038] The rotary kiln calcination unit is the core equipment for the solid-state reaction and crystal growth of lithium iron phosphate materials, and may include: Rotary kiln body ( Figure 1 (A long, cylindrical device with a transverse orientation): The main body of the rotary kiln rotates slowly around its central axis during the roasting process to promote the tumbling, mixing, and uniform heating of the internal materials. The main body of the rotary kiln is driven by a rotary motor (not shown in detail in the figure), and its rotation frequency is one of the key parameters controlled by the intelligent learning-based lithium iron phosphate roasting synthesis control system in this application embodiment, used to control the residence time of the materials in the kiln.

[0039] Temperature zones 1 to 10: These indicate that the rotary kiln is divided into ten independent heating and control zones along its axial direction. Each zone is equipped with an independent heating device (such as...). Figure 1The diagram shows burners A# and B#, and a temperature detection device. Each temperature zone corresponds to the opening degree of the heating device (e.g., a natural gas valve), the target set temperature for that zone, and the actual measured temperature for that zone. Burners A# and B# include modules A# and B#, respectively. Module A# supplies air (i.e., combustion air), and module B# supplies natural gas.

[0040] Wireless Temperature Receiver: The wireless temperature receiver is a wireless signal receiving device that receives real-time temperature data from multiple wireless temperature sensors (sensing terminals, not explicitly shown in the diagram) installed inside the rotary kiln and rotating with the kiln body. This data accurately reflects the actual temperature of the material at different locations within the kiln and is the most critical high-fidelity data source for temperature prediction and precise control in the intelligent learning-based lithium iron phosphate roasting and synthesis control system.

[0041] Burners A# and B#: Arranged in pairs below each temperature zone, they are the direct devices providing heat to the rotary kiln. They typically use natural gas as fuel, and the heating power of each zone is precisely controlled by adjusting the opening of the natural gas valve and the combustion air volume to maintain the set temperature of each zone. Their operation is directly regulated by the central control unit.

[0042] III. Combustion and Flue Gas Treatment Unit.

[0043] The combustion and flue gas treatment unit is responsible for supplying fuel and combustion air to the burner, and for treating and reusing the generated flue gas, and may include: Natural gas pipeline and shut-off valves 1 and 2: supply fuel to the burner. Shut-off valves 1 and 2 are safety devices used to quickly cut off the fuel supply in an emergency.

[0044] Combustion fan for roasting: Driven by an electric motor, it draws in and pressurizes air to provide the oxygen needed for natural gas combustion. Its outlet pressure is a crucial monitoring parameter to ensure a stable supply of combustion air. The frequency of the combustion fan is one of the adjustable parameters in the intelligent learning-based lithium iron phosphate roasting synthesis control system.

[0045] Flue gas recirculation heat exchanger: The flue gas recirculation heat exchanger is a key component for energy saving and process optimization. On one hand, it utilizes the high-temperature flue gas discharged from the rotary kiln to preheat the combustion air entering the burner through heat exchange, thereby improving combustion efficiency and saving fuel. The preheating fan is a key auxiliary device in the intelligent learning-based lithium iron phosphate roasting synthesis control system. One of its main functions is to control the flow rate of a portion of the high-temperature flue gas discharged from the end of the rotary kiln body to be redirected to the front end or preheating section of the rotary kiln body by adjusting its own operating frequency; this process is known as flue gas recirculation.

[0046] Exhaust fan: Located downstream of the combustion and flue gas treatment unit, it is driven by an electric motor. Its main function is to generate negative pressure throughout the kiln and combustion and flue gas treatment unit, guiding the flue gas to flow along a predetermined path and discharging the finally treated flue gas. Its operating frequency is the core means of regulating the pressure inside the kiln and the flue gas velocity.

[0047] The exhaust pipe of the roasting furnace cavity connects the exhaust end of the rotary kiln to the subsequent combustion and flue gas treatment unit. The pressure and temperature of the exhaust pipe of the roasting furnace cavity are important parameters for monitoring the roasting atmosphere and heat loss in the kiln.

[0048] Baghouse dust collection system: Includes a meter for calculating differential pressure, baghouse discharge valves, etc. Dust entrained in the roasting raw materials is efficiently captured here. Parameters such as inlet / outlet pipe pressure and flue gas temperature are used to monitor the operating status and efficiency of the baghouse dust collection system. The captured dust is discharged through the baghouse discharge valve (driven by a motor).

[0049] IV. Material Cooling and Collection Unit.

[0050] The material cooling and collection unit is responsible for rapidly and uniformly cooling the roasted, high-temperature material to a safe temperature and collecting it. It may include: Integrated roasting and cooling kiln: This is a device directly connected to the roasting section of a rotary kiln, used for cooling high-temperature materials. The integrated roasting and cooling kiln typically has a jacket or spray system, using circulating cooling water for heat exchange. Its operation is also driven by an electric motor.

[0051] The cooling water circulation system includes circulating cooling water pipelines, cooling water pumps (motor-driven), a cooling water reuse network, and related valves (such as pneumatic butterfly valves for the circulating cooling water pipelines). This system provides a continuous cooling medium to the integrated roasting and cooling kiln. The cooling water flow rate (adjusted by regulating the frequency of the cooling water pumps) is a key parameter predicted and adjusted by the intelligent learning-based lithium iron phosphate roasting synthesis control system based on the material discharge temperature, enabling precise control of the cooling rate. The kiln discharge temperature is used to detect the final temperature of the material after cooling and is a crucial feedback signal for judging the cooling effect, ensuring product quality, and ensuring safe discharge.

[0052] Finished product: After cooling, qualified lithium iron phosphate products are discharged from the end of the roasting and cooling integrated kiln and enter the subsequent collection and packaging process.

[0053] In summary, Figure 1The lithium iron phosphate roasting synthesis control system shown is a highly integrated and complex mechatronic system, with each unit tightly coupled through material flow, airflow, and energy flow. The central control unit acquires massive amounts of real-time operating data through sensors located throughout the system, and uses the lithium iron phosphate roasting synthesis control method based on intelligent learning described in this application to coordinate and precisely dynamically regulate multiple actuators, including electronic unloading valves, roasting feed pistons, rotary kiln main body rotary motors, burners in each temperature zone, roasting combustion fans, induced draft fans, and cooling water pumps, thereby achieving intelligent closed-loop control of the entire roasting process.

[0054] In the implementation scenario of this application, the lithium iron phosphate roasting process is typically carried out in a large rotary kiln, a complex thermal device with significant thermal inertia, multivariable coupling, and nonlinear characteristics. Material parameters, particularly moisture content, are key physical properties of the lithium iron phosphate precursor powder before entering the kiln. They directly determine the amount of heat required to evaporate moisture during the preheating stage and are one of the main sources of temperature fluctuations within the kiln. The current operating parameters of the kiln are a set of parameters, including real-time data describing the kiln's current operating status, such as natural gas flow rate, roasting combustion fan frequency, induced draft fan frequency, flue gas temperature at various points, and kiln operation (kiln rotation speed, i.e., the frequency of the rotary motor). The predicted material temperature is not an actual measurement value, but rather an estimate of the temperature the material will reach at a future point in time (e.g., when entering the high-temperature zone) based on a mathematical model by the intelligent learning-based lithium iron phosphate roasting synthesis control system. Temperature control device is a general term; it is not a single device but consists of multiple adjustable actuators, such as natural gas valves that control heat input, rotary motors that control material residence time, preheating fans that affect the kiln atmosphere and heat recovery, and cooling water pumps that control the cooling rate. The core logic of this application lies in accurately predicting the impact of an impending disturbance in material parameters, combined with current operating parameters, on the future temperature of the material, and proactively and collaboratively adjusting multiple temperature control devices to achieve active, feedforward, and precise control of the roasting process.

[0055] The following describes the process of the method provided in this implementation. Please refer to [link / reference]. Figure 2 This is a flowchart illustrating a lithium iron phosphate roasting synthesis control method based on intelligent learning in an embodiment of this application.

[0056] S201. Obtain the material parameters of lithium iron phosphate entering the kiln, including the material moisture content and material mass.

[0057] Among them, kiln usually refers to rotary kiln or roller kiln used for high-temperature roasting; lithium iron phosphate material parameters refer to the physical properties of a batch or unit mass of lithium iron phosphate precursor powder that is about to enter the kiln through the feeding device; material moisture content indicates the mass percentage of water contained in the material; material mass indicates the total mass (i.e., weight) of the batch of material.

[0058] Specifically, in the continuous or batch production process of lithium iron phosphate, before the material is fed into the kiln, the lithium iron phosphate roasting and synthesis control system based on intelligent learning acquires key parameters of each batch of material to be fed in real time through connected sensors or data interfaces. For example, an online moisture detector and weighing sensor are installed on the conveyor belt at the feed inlet. When a batch of material passes through, the lithium iron phosphate roasting and synthesis control system based on intelligent learning automatically records its moisture content and mass, and uses this data as input for temperature prediction.

[0059] In some embodiments, this step can be implemented in several ways: Optionally, a high-precision electronic scale can be installed below the feed hopper to weigh the incoming material in real time, while a near-infrared spectroscopy analyzer or microwave humidity sensor is used to perform non-contact scanning of the flowing material to obtain the average moisture content of the material; Optionally, for batch production, samples can be taken and sent to a laboratory for rapid testing before material loading, and then the operator can manually input the detected moisture content and total batch mass into the intelligent learning-based lithium iron phosphate roasting synthesis control system. It is understood that other methods can also be used to obtain material parameters, such as reading database records from the upstream drying process, etc., which are not limited here.

[0060] S202. Obtain the current operating parameters of the kiln.

[0061] Among them, the current operating parameters refer to a series of dynamic variables that reflect the kiln's operating status at the current moment, such as the actual temperature of each temperature zone, the operating frequency of each fan, the opening degree of each valve, and the rotation speed of the kiln body. These parameters together determine the real-time thermodynamic environment of the kiln.

[0062] S203. Based on material parameters and current operating parameters, the temperature of lithium iron phosphate at future times is predicted to obtain the predicted material temperature.

[0063] Temperature prediction is the core of this step and can be achieved through a corresponding temperature prediction model. A temperature prediction model is a mathematical model built through machine learning or mechanistic modeling. It can learn and express the complex nonlinear relationship between material parameters and operating parameters and how they jointly affect the heat transfer and reaction process of the material in the kiln. The future moment refers to a point in time with actual physical significance, such as the moment when the material is expected to reach a certain critical temperature zone in the kiln (such as the highest temperature roasting zone).

[0064] Specifically, once the material parameters of a newly added batch of lithium iron phosphate entering the kiln are acquired, the intelligent learning-based lithium iron phosphate roasting and synthesis control system immediately inputs these material parameters (material moisture content, material mass) along with the acquired current kiln operating parameters into a pre-trained temperature prediction model. This temperature prediction model simulates the complete process of temperature change over time and position as the material moves within the kiln after entering, under the current operating conditions, and outputs the predicted material temperature at key future moments.

[0065] The temperature prediction model is learned from historical experience data. Specifically, the material conditions of lithium iron phosphate entering the kiln are complex; therefore, the temperature prediction model is essentially a comprehensive prediction rule derived from learning from the entire historical experience data. That is, it takes into account all relevant parameters during lithium iron phosphate roasting, and in a relatively stable environment, integrates a large amount of experimental and system operation data to derive a general adjustment rule, which is the temperature prediction model. For example, the historical experience data must at least include the historical operating parameters of the kiln. Of course, the historical experience data may also include relevant parameters such as kiln size, processing precision, and weather conditions, which are not limited here.

[0066] In some embodiments, temperature prediction can be achieved in several ways: Optionally, a deep learning model based on Long Short-Term Memory (LSTM) networks can be used, employing historical sequences of multiple material parameters, operating parameters, and corresponding actual temperature sequences as training data. The model learns the temporal dependencies within these sequences, thereby predicting future temperatures based on new inputs. Optionally, a mechanistic model based on partial differential equations can be used, combined with computational fluid dynamics (CFD) simulations. The heat transfer, mass transfer, and chemical reaction kinetic equations of materials and gases are numerically solved to predict temperature distribution. It is understood that other methods can also be used for temperature prediction, such as mixture models and Gaussian process regression, which are not limited here.

[0067] S204. Determine the degree of temperature deviation between the predicted material temperature and the target material temperature.

[0068] The target material temperature refers to the ideal temperature value that the material should reach at the aforementioned future time according to the requirements of the lithium iron phosphate roasting process. This is a preset process parameter, which can be determined, for example, based on the ideal crystal phase formation temperature range. The degree of temperature deviation is a quantitative indicator used to represent the difference between the predicted value and the target value. It can be the difference between the two directly or a relative deviation percentage.

[0069] Specifically, the lithium iron phosphate roasting synthesis control system, based on intelligent learning, reads the target material temperature value corresponding to the aforementioned future time from the process parameter database. Then, it compares the predicted material temperature with the target material temperature and calculates the difference between the two. For example, if the predicted temperature when the material reaches the center of the high-temperature zone in 10 minutes is 742°C, while the target temperature at that location is 750°C, the temperature deviation is -8°C.

[0070] S205. Adjust the kiln temperature control device according to the degree of temperature deviation.

[0071] Among them, regulation refers to the intelligent learning-based lithium iron phosphate roasting synthesis control system generating a series of specific and quantitative control commands based on the calculated degree of temperature deviation, and sending them to various actuators in the kiln to change the thermal state of the kiln, thereby counteracting the impending temperature deviation.

[0072] Specifically, when a non-negligible temperature deviation is identified, the intelligent learning-based lithium iron phosphate roasting synthesis control system invokes an adjustment strategy module. This module calculates the necessary control actions based on the degree of deviation (e.g., -8°C) and the kiln's dynamic response model. For example, to compensate for the -8°C deviation, the intelligent learning-based lithium iron phosphate roasting synthesis control system might calculate that the natural gas valve opening needs to be increased by 5% within the next two minutes, while the frequency of the kiln's rotary motor needs to be reduced by 2% to increase heat input and extend the material's heating time. Subsequently, the intelligent learning-based lithium iron phosphate roasting synthesis control system sends these instructions to the corresponding natural gas valves and frequency converters via a PLC (Programmable Logic Controller).

[0073] As can be seen, the embodiments of this application achieve predictive adjustments. For example, if the moisture content of the material increases, the material will absorb too much heat after reaching the corresponding temperature zone, which will cause the temperature of the temperature zone to drop. Therefore, the temperature control device of the kiln can be adjusted (e.g., the natural gas valve can be controlled in advance) to increase the temperature of the temperature zone to counteract this situation and maintain a constant temperature in the temperature zone.

[0074] The following provides a more detailed description of the process of the method provided in this implementation. Please refer to [link / reference]. Figure 3 This is another flowchart illustrating the lithium iron phosphate roasting synthesis control method based on intelligent learning in this application embodiment.

[0075] S301. Obtain the material parameters of lithium iron phosphate entering the kiln, including the material moisture content and material mass.

[0076] Refer to step S201, which will not be repeated here.

[0077] S302. Determine the feeding interval of the kiln; The feeding interval refers to the time difference between the current feeding operation and the previous feeding operation. In continuous or semi-continuous production, this duration reflects changes in the production rhythm.

[0078] Specifically, the lithium iron phosphate roasting synthesis control system based on intelligent learning maintains a timer and records the precise timestamp of each feeding action (such as driving the roasting feed piston). When a new feeding action occurs, the feeding interval is determined by calculating the difference between the current timestamp and the previously recorded timestamp. This duration data can be used to determine whether production is stable or whether there are unplanned stoppages.

[0079] S303. If the feeding interval is longer than the preset interval, reduce the frequency of the preheating fan of the kiln to reduce the flow rate of flue gas reuse in the kiln.

[0080] The preset interval is a threshold set based on experience, such as 5 minutes. Exceeding this interval means that no new cold material enters the kiln for a period of time. The preheating fan is usually used to draw a portion of the high-temperature flue gas discharged from the kiln tail back to the kiln head or preheating section to preheat the newly fed material and combustion air, thereby reducing its frequency and thus reducing the amount of high-temperature flue gas reused.

[0081] Specifically, after determining the feeding interval, the lithium iron phosphate roasting and synthesis control system, based on intelligent learning, compares it with a preset interval threshold. If the feeding interval is found to be too long (e.g., 10 minutes), it means that the total amount of material in the kiln is reduced, the heat input is relatively excessive, and there is a risk of a continuous rise in kiln temperature. Therefore, the intelligent learning-based lithium iron phosphate roasting and synthesis control system automatically performs a compensation operation, that is, it sends a command to the frequency converter of the preheating fan via the PLC to reduce its operating frequency from, for example, 35Hz to 30Hz. This reduces the amount of recycled flue gas introduced, and the excess flue gas can be discharged outside the kiln by the induced draft fan to maintain the stability of the kiln temperature. In addition, if the feeding interval is longer than the preset interval, the opening of the natural gas valve can be reduced to decrease the heat supply to the kiln.

[0082] In some embodiments, this step can be implemented in several ways: Optionally, a simple step adjustment is used, where the frequency of the preheating fan is reduced by a fixed percentage (e.g., 10%) once the feeding interval exceeds a preset interval; alternatively, a linear adjustment proportional to the feeding interval is used, where the frequency of the preheating fan is reduced more significantly as the feeding interval increases, thus achieving finer compensation. When normal feeding resumes (i.e., the feeding interval is less than or equal to the preset interval), the frequency of the preheating fan automatically returns to its normal value before reduction.

[0083] S304. Determine whether the material parameters meet the preset cold material judgment conditions, wherein the cold material judgment conditions include at least one of the following: the material moisture content is greater than the preset moisture content threshold and the material mass is greater than the preset mass threshold.

[0084] Cold lumps refer to batches of materials that, due to excessive moisture content or excessive single-feed volume, would significantly impact the thermal balance within the kiln. The preset criteria for identifying cold lumps are thresholds derived from process experience and historical data statistics; for example, a moisture content threshold of 1.5% and a mass threshold of 50 kg. Materials meeting either or both of these criteria can be classified as cold lumps.

[0085] In some embodiments, this step can be implemented in several ways: Optionally, an "OR" logic can be used, that is, if either the moisture content or the mass of the material exceeds the standard, it is determined to be a cold lump, and the material parameters are determined to meet the preset cold lump material determination conditions; Optionally, a more complex weighted scoring mechanism can be used, which sets different weights for moisture content and mass, and calculates a comprehensive coldness score. Only when the comprehensive coldness score exceeds a certain threshold is it determined to be a cold lump, and the material parameters are determined to meet the preset cold lump material determination conditions. This method can more comprehensively reflect the potential impact of the material.

[0086] In some embodiments, the weighted scoring mechanism can further assess the potential impact of "cold masses" by quantifying the degree of exceedance of different parameters and assigning them different weights. First, the relative exceedance rate of each material parameter can be calculated. For example, the exceedance rate R_h of material moisture content can be defined as (h_actual - h_threshold) / h_threshold, where h_actual is the material moisture content and h_threshold is a preset moisture content threshold. Similarly, the exceedance rate R_m of material mass can be defined as (m_actual - m_threshold) / m_threshold, where m_actual is the material mass and m_threshold is a preset mass threshold.

[0087] Then, based on historical data analysis or expert experience, weights w_h and w_m are assigned to these two exceedance rates. These weights reflect the relative importance of changes in material moisture content and material mass on the kiln's thermal balance, and w_h + w_m = 1. For example, since moisture evaporation typically consumes a large amount of heat, the weight w_h for material moisture content can be greater than the weight w_m for material mass; for instance, w_h could be set to 0.7, while w_m could be set to 0.3.

[0088] The mathematical expression for the overall coldness score (Score) is: Score = w_h × R_h + w_m × R_m. The intelligent learning-based lithium iron phosphate roasting synthesis control system can preset a comprehensive scoring threshold (Score_threshold), for example, 0.5. Only when the calculated Score is greater than Score_threshold is the batch of material judged as a "cold lump," confirming that the material parameters meet the preset cold lump material judgment conditions. In this way, it is possible to more precisely identify "potential cold lumps" where individual exceedances are not severe, but the combined effect of multiple exceedances is significant, improving the accuracy and sensitivity of the judgment.

[0089] S305. If the material parameters meet the cold agglomerate material determination conditions, then execute the step of obtaining the current operating parameters of the kiln.

[0090] This step is a logical judgment and flow control step. It acts as a "filter" or "trigger," ensuring that subsequent complex prediction and control calculations are only activated when necessary (i.e., when the significant disturbance of the "cold cluster" is identified), thereby saving computational resources.

[0091] S306. Obtain the current operating parameters of the kiln.

[0092] Refer to step S202, which will not be repeated here.

[0093] S307. Based on material parameters and current operating condition parameters, a temperature prediction model is used to predict the temperature change curve of lithium iron phosphate after it enters the kiln, and the predicted temperature change curve is obtained. The temperature prediction model is learned based on historical experience data, which includes at least the historical operating condition parameters of the kiln.

[0094] The predicted temperature change curve is a continuous curve with time or the position of the lithium iron phosphate material entering the kiln as the x-axis and material temperature as the y-axis. It describes the entire temperature change process of the predicted batch of material from entering the kiln to leaving (or reaching a certain key point). The description of the temperature prediction model can be found in step 203. The temperature prediction model is based on a database formed through extensive experiments and actual system operation, generating the predicted temperature change curve through empirical learning. Thus, once the lithium iron phosphate material entering the kiln experiences changes in moisture content and / or mass, the temperature prediction model can predict the temperature change, thereby enabling predictive adjustment of the kiln's temperature control device.

[0095] Specifically, unlike step S203 which predicts a single temperature point, in this step, after the intelligent learning-based lithium iron phosphate roasting synthesis control system inputs the material parameters and current operating condition parameters of the cold-formed material into the temperature prediction model, the temperature prediction model is configured to output a time series. This time series contains the predicted temperature values ​​of the material at multiple consecutive future time points starting from the current moment. Connecting these points forms the predicted temperature change curve.

[0096] S308. Determine the difference between the predicted temperature change curve and the preset roasting temperature curve.

[0097] Among them, the preset roasting temperature curve is the ideal temperature curve of the material changing with time or position in the kiln as required by the process; the difference point refers to the point or section on which the predicted temperature change curve deviates significantly from the ideal curve.

[0098] Specifically, the lithium iron phosphate roasting synthesis control system based on intelligent learning loads a standard roasting temperature curve (which is also a time series) into memory. Then, the predicted temperature change curve is compared point by point or segment by segment with this standard roasting temperature curve in the same coordinate system to find the point where the "separation" between the two curves is most obvious, or the entire segment where the predicted temperature change curve is lower or higher than the standard roasting temperature curve by more than a certain range. This point or segment is the difference point.

[0099] S309. Obtain the temperature difference between the corresponding temperature of the difference point in the predicted temperature change curve and the corresponding temperature in the calcination temperature curve.

[0100] This step involves quantifying the degree of deviation after identifying the key points of difference.

[0101] Specifically, suppose the predicted temperature change curve shows the greatest difference from the calcination temperature curve at the "15-minute" point. The lithium iron phosphate calcination synthesis control system, based on intelligent learning, will then read the temperature value at the 15-minute mark on both the predicted temperature change curve (e.g., 738℃) and the calcination temperature curve (e.g., 750℃). It will then calculate the difference between these two temperatures: 738 - 750 = -12℃. This -12℃ is the specific temperature difference at that point.

[0102] S310. If the temperature difference is greater than the preset temperature difference threshold, the temperature corresponding to the difference point in the predicted temperature change curve shall be used as the predicted material temperature, wherein the temperature corresponding to the difference point in the calcination temperature curve shall be the target material temperature.

[0103] This step introduces a threshold to filter out minor, unimportant deviations and pass on the deviation information that truly needs attention to subsequent processing stages.

[0104] Specifically, after calculating the temperature difference (e.g., -12℃), the lithium iron phosphate roasting synthesis control system based on intelligent learning compares its absolute value with a preset temperature difference threshold (e.g., 0.5℃). Since |-12℃| > 0.5℃, this is confirmed as a significant deviation requiring control intervention. Therefore, the intelligent learning-based lithium iron phosphate roasting synthesis control system assigns two key temperature values ​​at this point of difference to two variables: the corresponding temperature on the predicted temperature change curve (738℃) is designated as the predicted material temperature, and the corresponding temperature on the roasting temperature curve (750℃) is designated as the target material temperature. These two values ​​will serve as the core inputs for subsequent calculations of the adjustment strategy.

[0105] In some embodiments, the temperature difference threshold can be a dynamic value. For example, in the critical high-temperature zone of roasting, the temperature difference threshold can be set smaller (e.g., 0.2°C), while in the non-critical preheating zone, the temperature difference threshold can be set larger (e.g., 1°C) to achieve refined management of different stages.

[0106] S311. Determine the degree of temperature deviation between the predicted material temperature and the target material temperature.

[0107] Refer to step S204, which will not be repeated here.

[0108] S312. Obtain the adjustment strategy prediction model of the temperature control device, wherein the adjustment strategy prediction model is used to reduce the degree of temperature deviation.

[0109] Among these, the regulation strategy prediction model is another core artificial intelligence model, which differs from the temperature prediction model. Its input is the "degree of temperature deviation," and its output is "how to adjust to eliminate this deviation," i.e., the specific control strategy. This regulation strategy prediction model can be data-driven or mechanism-based. Internally, the regulation strategy prediction model has learned a large number of historical cases regarding "how much temperature deviation occurred under certain operating conditions, and what adjustment operations were performed to successfully eliminate the deviation."

[0110] In some embodiments, the regulation policy prediction model is a core model for decision optimization. Taking a regulation policy prediction model based on deep reinforcement learning (DRL) as an example, its construction and use also include three core steps: model training, the model itself, and model usage.

[0111] The training phase of the regulation strategy prediction model first requires a high-fidelity kiln environment simulator, typically represented by the aforementioned "temperature prediction model" or a more sophisticated digital twin model. The DRL agent performs tens of thousands of "exploration and exploitation" interactions within this virtual environment. At each time step, the agent observes the current state, which includes the degree of temperature deviation and other relevant operating parameters. The agent then outputs an action—a set of specific parameter adjustments to the temperature control devices (such as gas valves or rotary motors). The environment simulator calculates the new state for the next time step based on this action and provides a reward. The design of the reward function is crucial for training; its goal is to penalize large temperature deviations and frequent adjustment actions while rewarding temperature stability. For example, the reward function can be set as R = -α × |T_actual - T_target| - β × Σ|Δu|, where R is the reward, |T_actual - T_target| is the penalty for temperature deviation, T_actual is the actual temperature, T_target is the target temperature, Σ|Δu| is the penalty for changes in control actions, and α and β are preset weight coefficients. The training criterion is to maximize the cumulative expected reward. By employing advanced DRL algorithms such as Proximal Policy Optimization (PPO), the parameters of the agent's policy network (usually a deep neural network) are continuously optimized until it can generate actions that yield the highest long-term reward for various states.

[0112] The regulation strategy prediction model itself, i.e. the trained policy network, is essentially a nonlinear function. Its input is a vector depicting the current system state (including the degree of temperature deviation), and its output is a multidimensional continuous or discrete value vector, which directly corresponds to the target parameter adjustment amount and execution timing of each temperature control device.

[0113] The application phase of the regulation strategy prediction model: During online control, once the lithium iron phosphate roasting synthesis control system based on intelligent learning determines the degree of temperature deviation, it packages this temperature deviation and related operating parameters into a state vector and inputs it into the pre-trained regulation strategy prediction model. The regulation strategy prediction model performs a forward calculation and instantly outputs an optimal target regulation strategy. This target regulation strategy includes the specific target regulation time and the target parameter regulation amount, thus realizing an intelligent decision-making closed loop from problem identification to optimal solution generation.

[0114] S313. Using the adjustment strategy prediction model, determine the target adjustment strategy of the temperature control device based on the degree of temperature deviation. The target adjustment strategy includes the target adjustment time and the target parameter adjustment amount.

[0115] This step involves using the regulation strategy prediction model to perform actual calculations and derive the final target regulation strategy. For details on the process, please refer to the description of the usage phase of the regulation strategy prediction model.

[0116] Specifically, the lithium iron phosphate roasting synthesis control system based on intelligent learning inputs the determined temperature deviation (e.g., -8°C), along with the current operating parameters, into the regulation strategy prediction model. After calculation, the regulation strategy prediction model outputs a structured regulation strategy. This strategy explicitly specifies at which time point (the target regulation time), for which temperature control device (e.g., a natural gas valve), and by what magnitude of regulation (the target parameter adjustment amount, such as increasing the opening by 5%). It may be a single action or a series of continuous or coordinated actions.

[0117] S314. At the target adjustment time, adjust the temperature control device according to the target parameter adjustment amount.

[0118] Referring to step S205, it will not be repeated here. After the aforementioned series of detailed prediction, judgment, and strategy generation steps, the adjustment action here is already a well-thought-out and highly optimized intelligent intervention.

[0119] In one embodiment, the temperature control device includes at least one of a natural gas valve, a rotary motor, a preheating fan, and a cooling water pump. The target parameter adjustment amount includes at least one of a valve opening adjustment amount, a motor frequency adjustment amount, a fan frequency adjustment amount, and a water pump frequency adjustment amount. Accordingly, adjusting the temperature control device according to the target parameter adjustment amount may include at least one of the following steps: adjusting the opening of the natural gas valve according to the valve opening adjustment amount; adjusting the frequency of the rotary motor according to the motor frequency adjustment amount; adjusting the frequency of the preheating fan according to the fan frequency adjustment amount; and adjusting the frequency of the cooling water pump according to the water pump frequency adjustment amount. By coordinating and adjusting multiple dimensions such as heat supply, material residence time, kiln atmosphere, and cooling rate, complex temperature fluctuations can be addressed more efficiently and comprehensively, achieving global optimal control of the roasting process.

[0120] As can be seen in this embodiment, the lithium iron phosphate roasting synthesis control system based on intelligent learning introduces real-time perception of material parameters (material moisture content, material mass) entering the kiln, and combines this with the current operating parameters of the kiln to construct an artificial intelligence model that can accurately predict future temperature changes of the material. Therefore, it fundamentally transforms the control mode from passive "lagging correction" to active "predictive intervention", thereby realizing feedforward, multivariable, and collaborative optimization intelligent control of the roasting process. It can eliminate potential temperature fluctuations in advance, stabilize the roasting temperature within an extremely narrow ideal window, and ultimately improve the batch consistency, electrochemical performance, and production yield of lithium iron phosphate products.

[0121] In some embodiments, the determination of the degree of temperature deviation between the predicted material temperature and the target material temperature can be achieved using a risk quantification scheme based on a multi-dimensional dynamic health index assessment. Specifically, the lithium iron phosphate roasting synthesis control system based on intelligent learning no longer simply calculates the difference between the predicted material temperature and the target material temperature as the degree of deviation, but instead constructs a multi-dimensional dynamic health index assessment model to quantify the "risk" and "severity" of the temperature deviation, thereby determining the degree of temperature deviation. This multi-dimensional dynamic health index assessment model comprehensively considers the following factors: Temperature difference: The instantaneous absolute difference between the predicted material temperature and the target material temperature.

[0122] Deviation Duration: The predicted duration of the deviation. Short-term spikes and long-term gradual deviations pose different hazards to lithium iron phosphate materials entering the furnace.

[0123] Location of Deviation: Deviations occur at different logical stages of the preset calcination temperature curve (such as the preheating stage, high-temperature calcination stage, and cooling stage), and their impact varies. For example, deviations in the high-temperature calcination stage have a more severe impact on the quality of lithium iron phosphate products than deviations in the preheating stage.

[0124] Deviation rate of change: The rate of increase or decrease in temperature deviation reflects the dynamic trend of the deviation.

[0125] Material sensitivity: Depending on the characteristics of the lithium iron phosphate material entering the kiln, different materials have different sensitivities to temperature fluctuations. For example, highly active materials are more sensitive to high temperatures and may experience over-roasting.

[0126] All these dimensions of data are input into a fuzzy logic inference system or a Bayesian network system. Based on a pre-defined expert rule base and learned probability relationships, the fuzzy logic inference system or Bayesian network system fuses these multi-dimensional inputs to calculate a health risk index between 0 and 1 in real time, which serves as the predicted temperature deviation of the material temperature relative to the target material temperature. For example, a healthy roasting state corresponds to a health risk index close to 0, while a roasting state about to experience serious quality problems corresponds to a health risk index close to 1. Only when this health risk index exceeds a pre-defined high-risk threshold is it considered a temperature deviation requiring strong intervention. By calculating this health risk index, replacing a single temperature difference, the lithium iron phosphate roasting synthesis control system based on intelligent learning can more comprehensively and intelligently assess the true hazards of temperature deviations, avoiding over-control due to minor deviations, while promptly responding to key risks that may truly lead to product quality problems. The risk quantification scheme based on multi-dimensional dynamic health index assessment can specifically include the following stages: Phase 1: Extraction and quantification of multi-dimensional features; In embodiments of this application, the objective of this stage is to extract and quantify key features from the predicted temperature curve that can comprehensively reflect the risk of deviation.

[0127] For the "temperature difference: the instantaneous absolute difference between the predicted material temperature and the target material temperature", the lithium iron phosphate roasting synthesis control system based on intelligent learning compares the predicted temperature change curve and the preset roasting temperature curve point by point. At each predicted time point tᵢ, ΔT(tᵢ) = |T_predicted(tᵢ) - T_target(tᵢ)| is calculated, where ΔT(tᵢ) is the instantaneous absolute difference, T_predicted(tᵢ) is the predicted material temperature, and T_target(tᵢ) is the target material temperature.

[0128] Regarding "Duration of Deviation: the predicted duration of the deviation," a continuous period of time during which ΔT(tᵢ) exceeds a certain low threshold (e.g., 0.5℃) can be identified. Assuming that ΔT(tᵢ) continuously exceeds the threshold from a start time t_start to an end time t_end, then the duration D = t_end - t_start. This duration is an important indicator of the severity of the deviation, because short-term fluctuations usually have a small impact, while long-term deviations can lead to cumulative effects.

[0129] To address the issue of "deviation location: deviation occurring at different logical stages of the preset roasting temperature curve," the roasting process in the kiln can be divided into multiple logical stages, such as a preheating stage (0-T1), a high-temperature roasting stage (T1-T2), and a cooling stage (T2-T3), where T1, T2, and T3 are the temperature boundary values ​​for the corresponding logical stages. Each logical stage is assigned a different weighting factor W_position; for example, W_position for the high-temperature roasting stage might be 0.9, for the preheating stage 0.6, and for the cooling stage 0.7. When a deviation is detected at a certain time point tᵢ, the corresponding W_position can be extracted based on the stage tᵢ is in. This is done to differentiate the degree of impact of deviations at different logical stages on the quality of lithium iron phosphate products.

[0130] The "rate of change of deviation: the increase or decrease in temperature deviation" can be obtained by performing a first-order difference or a more complex curve fitting on ΔT(tᵢ). Mathematically, it can be expressed as Rate_change(tᵢ)=d(ΔT) / dt or Rate_change(tᵢ)=(ΔT(tᵢ)-ΔT(tᵢ) -1 )) / (tᵢ-tᵢ -1 ), where Rate_change(tᵢ) is the deviation rate. An extremely high deviation rate (whether it increases or decreases) indicates dynamic instability or sudden disturbances, requiring a rapid response from the lithium iron phosphate roasting synthesis control system based on intelligent learning.

[0131] To address material sensitivity, a pre-trained classification or regression model can be used to determine the intrinsic sensitivity (S_material) of the lithium iron phosphate material entering the kiln to temperature fluctuations, based on its crystal form, particle size distribution, activity, and other material characteristics. For example, highly pure materials with strict requirements for crystallinity will have a higher intrinsic sensitivity (S_material) value.

[0132] Phase Two: Multi-dimensional data fusion and calculation of health risk index.

[0133] All the quantified features extracted in the first stage, including temperature difference, deviation duration, deviation location, deviation rate of change, and material sensitivity, are input into the fuzzy logic inference system. This fuzzy logic inference system consists of three main parts: a fuzzifier, a fuzzy inference engine, and a defuzzifier.

[0134] The fuzzifier converts the quantized features extracted in the first stage into fuzzy sets. For example, "temperature difference" can be fuzzified into "small deviation," "medium deviation," and "large deviation." Each fuzzy set has a membership function that defines the degree to which the input value belongs to that set. For example, 5℃ has a membership degree of 0.8 for "medium deviation" and 0.2 for "large deviation."

[0135] The fuzzy inference engine contains a set of expert-defined fuzzy rules, presented in "IF-THEN" format, to simulate the decision-making process of human experts. For example: "IF temperature difference is large deviation AND deviation occurs at high-temperature roasting section AND deviation rate is high AND material sensitivity is high, THEN health risk index is extremely high." Each rule has an activation strength determined by the membership degree of its premise.

[0136] Defuzzifier: Converts the output of the fuzzy inference engine (a fuzzy set) back into a specific, executable numerical value, i.e., the final health risk index, for example, using the centroid method or area center method to calculate, and finally outputs a health risk index between 0 and 1.

[0137] In some embodiments of this application, as an alternative to fuzzy logic reasoning systems, the lithium iron phosphate roasting synthesis control system based on intelligent learning can also employ a Bayesian network system for multi-dimensional data fusion and risk quantification analysis. A Bayesian network system is a probabilistic graphical model that uses nodes to represent random variables (such as temperature difference, deviation duration, deviation location, etc.) and directed edges to represent the conditional dependencies between random variables. Each node is associated with a conditional probability distribution. By learning from historical data, the Bayesian network system can establish probabilistic relationships between various factors and the final health risk index. When new multi-dimensional features are input, the Bayesian network system can be used to calculate the posterior probability that the health risk index is in a high-risk state under the current observation conditions. This posterior probability can be directly used as the health risk index. The advantage of using a Bayesian network system is that it can intuitively represent the causal relationships between random variables and can handle uncertain data, enabling it to provide robust risk assessment even when some sensor data is missing or there is significant noise.

[0138] As can be seen, the embodiments of this application elevate the determination of temperature deviation from a simple comparison of a single physical quantity to a comprehensive consideration of multiple dimensions such as time, space, dynamic trends, and the inherent sensitivity of materials. Through a fusion algorithm, a quantitative health risk index with risk assessment significance is output. This enables the intelligent learning-based lithium iron phosphate roasting synthesis control system to no longer merely judge temperature fluctuations superficially, but to deeply assess their real harm to the quality of the final lithium iron phosphate product. This improves the effectiveness, pertinence, and robustness of the lithium iron phosphate roasting synthesis control strategy, ensures more precise quality control, and optimizes energy consumption and wear and tear on related equipment.

[0139] The following describes the intelligent learning-based lithium iron phosphate roasting synthesis control system in this application embodiment from a hardware processing perspective. Please refer to [link to relevant documentation]. Figure 4 This is a schematic diagram of at least part of the physical device structure of the lithium iron phosphate roasting synthesis control system based on intelligent learning in the embodiments of this application.

[0140] It should be noted that, Figure 4 The structure of the lithium iron phosphate roasting synthesis control system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0141] like Figure 4 As shown, the lithium iron phosphate roasting synthesis control system based on intelligent learning includes a CPU 401, which can perform various appropriate actions and processes according to a program stored in ROM 402 or a program loaded from storage section 408 into RAM 403, such as executing the methods described in the above embodiments. RAM 403 also stores various programs and data required for system operation. The CPU 401, ROM 402, and RAM 403 are interconnected via bus 404. I / O interface 405 is also connected to bus 404.

[0142] The following components are connected to I / O interface 405: input section 406 including audio input devices, push-button switches, etc.; output section 407 including liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 408 including hard disks, etc.; and communication section 409 including network interface cards such as LAN (Local Area Network) cards, modems, etc. Communication section 409 performs communication processing via a network such as the Internet. Drive 410 is also connected to I / O interface 405 as needed. Removable media 411, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on drive 410 as needed so that computer programs read from them can be installed into storage section 408 as needed.

[0143] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by CPU 401, it performs the various functions defined in this application.

[0144] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0145] Specifically, the lithium iron phosphate roasting synthesis control system based on intelligent learning in this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the lithium iron phosphate roasting synthesis control method based on intelligent learning provided in the above embodiment.

[0146] In another aspect, this application also provides a computer-readable storage medium, which may be included in the intelligent learning-based lithium iron phosphate roasting synthesis control system described in the above embodiments; or it may exist independently and not assembled into the intelligent learning-based lithium iron phosphate roasting synthesis control system. The storage medium carries one or more computer programs, which, when executed by a processor of the intelligent learning-based lithium iron phosphate roasting synthesis control system, cause the intelligent learning-based lithium iron phosphate roasting synthesis control system to implement the intelligent learning-based lithium iron phosphate roasting synthesis control method provided in the above embodiments.

[0147] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for controlling the roasting synthesis of lithium iron phosphate based on intelligent learning, characterized in that, The intelligent learning-based method for controlling the roasting synthesis of lithium iron phosphate includes: Obtain the material parameters of lithium iron phosphate entering the kiln, wherein the material parameters include the material moisture content and the material mass; Obtain the current operating parameters of the kiln; Based on the material parameters and the current operating condition parameters, the temperature of the lithium iron phosphate at future times is predicted to obtain the predicted material temperature; Determine the degree of temperature deviation of the predicted material temperature relative to the target material temperature; The temperature control device of the kiln is adjusted according to the degree of temperature deviation.

2. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 1, characterized in that, After obtaining the material parameters of the lithium iron phosphate entering the kiln, the process also includes: Determine whether the material parameters meet the preset cold lump material judgment conditions, wherein the cold lump material judgment conditions include at least one of the material moisture content being greater than a preset moisture content threshold and the material mass being greater than a preset mass threshold; If the material parameters meet the cold agglomerate material determination conditions, then the step of obtaining the current operating parameters of the kiln is executed.

3. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 1, characterized in that, The step of predicting the temperature of the lithium iron phosphate at future times based on the material parameters and the current operating condition parameters to obtain the predicted material temperature includes: Based on the material parameters and the current operating condition parameters, a temperature prediction model is used to predict the temperature change curve of the lithium iron phosphate after it enters the kiln, and a predicted temperature change curve is obtained. The temperature prediction model is learned based on historical experience data, and the historical experience data includes at least the historical operating condition parameters of the kiln. Determine the points of difference between the predicted temperature change curve and the preset roasting temperature curve; Based on the points of difference, the predicted material temperature is determined.

4. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 3, characterized in that, Determining the predicted material temperature based on the difference points includes: Obtain the temperature difference between the corresponding temperature of the difference point in the predicted temperature change curve and the corresponding temperature in the calcination temperature curve; If the temperature difference is greater than a preset temperature difference threshold, then the temperature corresponding to the difference point in the predicted temperature change curve is taken as the predicted material temperature, wherein the temperature corresponding to the difference point in the calcination temperature curve is the target material temperature.

5. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 1, characterized in that, The temperature control device for adjusting the kiln according to the degree of temperature deviation includes: Based on the degree of temperature deviation, the target adjustment time and target parameter adjustment amount of the temperature control device are determined; At the target adjustment time, the temperature control device is adjusted according to the target parameter adjustment amount.

6. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 5, characterized in that, The step of determining the target adjustment time and target parameter adjustment amount of the temperature control device based on the degree of temperature deviation includes: Obtain the adjustment strategy prediction model of the temperature control device, wherein the adjustment strategy prediction model is used to reduce the degree of temperature deviation; Using the aforementioned adjustment strategy prediction model, a target adjustment strategy for the temperature control device is determined based on the degree of temperature deviation. The target adjustment strategy includes the target adjustment time and the target parameter adjustment amount.

7. The lithium iron phosphate roasting synthesis control method based on intelligent learning as described in claim 5, characterized in that, The temperature control device includes at least one of a natural gas valve, a rotary motor, a preheating fan, and a cooling water pump. The target parameter adjustment amount includes at least one of a valve opening adjustment amount, a motor frequency adjustment amount, a fan frequency adjustment amount, and a water pump frequency adjustment amount. Adjusting the temperature control device according to the target parameter adjustment amount includes at least one of the following steps: Adjust the opening of the natural gas valve according to the valve opening adjustment amount; Adjust the frequency of the rotary motor according to the motor frequency adjustment amount; Adjust the frequency of the preheating fan according to the fan frequency adjustment amount; Adjust the frequency of the cooling water pump according to the specified pump frequency adjustment amount.

8. The method for controlling the roasting synthesis of lithium iron phosphate based on intelligent learning as described in claim 1, characterized in that, After obtaining the material parameters of the lithium iron phosphate entering the kiln, the process also includes: Determine the feeding interval of the kiln; If the feeding interval is longer than the preset interval, the frequency of the preheating fan of the kiln is reduced to reduce the flue gas reuse flow rate of the kiln.

9. A lithium iron phosphate roasting synthesis control system based on intelligent learning, characterized in that, The lithium iron phosphate roasting synthesis control system based on intelligent learning includes: One or more processors and memory; The memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the intelligent learning-based lithium iron phosphate roasting synthesis control system to execute the intelligent learning-based lithium iron phosphate roasting synthesis control method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes instructions that, when executed on the intelligent learning-based lithium iron phosphate roasting synthesis control system, cause the intelligent learning-based lithium iron phosphate roasting synthesis control system to perform the intelligent learning-based lithium iron phosphate roasting synthesis control method according to any one of claims 1 to 8.

Images