An artificial intelligence-based feedstock process efficiency optimization and real-time temperature control method and system

Abstract

The application discloses a kind of based on artificial intelligence's refining material process efficiency optimization and real-time temperature control method and system, and realizes refining material process high-efficiency optimization and accurate temperature control by four-step innovation: first step breaks through the limitation of traditional single-point monitoring, according to "three-dimensional coverage" deployment multimodal sensing array and feed inlet near-infrared spectrum sensor, real-time collection process-raw material two-dimensional data and association, solve data island problem;Second step innovation constructs "process-raw material" two-dimensional feature vector, adopts XGBoost model to predict yield, combines threshold value to judge dynamic power demand, replaces traditional experience decision-making;Third step is based on the value output by DQN model, calculates optimal dynamic temperature curve in stages, and then through the accurate power instruction generated, break through the fixed curve rigid problem;Fourth step is the first to create three-level time limit mechanism, corresponding "model reinforcement-local compensation-cross furnace repair" differentiated scheme, form effect closed loop.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of refining process, in particular to a refining process efficiency optimization and real-time temperature control method and system based on artificial intelligence. BACKGROUND

[0002] With the increasing demand for high-purity metals in new energy and high-end manufacturing fields, the limitations of traditional technical systems are becoming increasingly prominent: on the one hand, artificial experience cannot cope with complex working conditions with multiple parameter couplings; on the other hand, simple models and fixed curves cannot meet the comprehensive needs of "upgrading, increasing efficiency, and reducing costs". The industry urgently needs a new technical method that can realize "multi-source data collaboration, AI precise decision-making, and hierarchical compensation closed loop" to break through the bottleneck of traditional technology, so a refining process efficiency optimization and real-time temperature control method and system based on artificial intelligence is born.

[0003] The prior art such as the invention patent application with publication number CN102222128B discloses a method for optimizing waste plastic oil refining combustion. The existing method mainly relies on the experience of workers. The method of the present application establishes a model based on data mining technology by data collection of the main process of waste plastic oil refining, aiming at the mutual influence relationship between the axial temperature distribution of the reaction kettle and the operation parameters of each burner, the feed quantity and the product quantity, and applies parallel optimization algorithm and other means to establish a method for optimizing waste plastic oil refining combustion. The method of the present application can effectively control the temperature distribution in the reaction kettle during waste plastic oil refining, improve the reaction efficiency and product quality, and can implement offline optimization and online real-time combustion optimization.

[0004] For the above-mentioned scheme, the present application applicant found that the above-mentioned technology at least has the following technical problems: 1. The traditional technology adopts a "single-point monitoring + offline analysis" mode, the sensor deployment only covers a few key nodes such as furnace wall temperature and flue gas, lacks core data such as local temperature distribution of the molten pool and real-time changes of raw material composition, and the process dimension and raw material dimension data are stored in independent systems, the time stamps are not unified, the formats are not compatible, forming a "data island". At the same time, the system only has basic storage and alarm functions, has no data cleaning and standardization processing capability, the offline raw material detection result feedback is lagging, so that the process cannot be adjusted in time when the raw material composition fluctuates, and the historical data cannot be reused as optimization basis due to no structured archiving.

[0005] 2、Traditional yield prediction is only based on two process parameters of heating power and melting time to construct a linear regression model, without fusing key features such as raw material composition and furnace gas concentration, resulting in high prediction error and unable to provide reliable reference for power adjustment. Heating power adjustment is mainly based on the operator's experience observation of the instrument panel, without quantitative calculation logic, which is easy to cause "over-adjustment" or "insufficient adjustment" problem, damaging the stability of product quality, causing energy waste, and difficult to adapt to the demand for decision-making accuracy in large-scale production.

[0006] 3、The traditional smelting furnace adopts a fixed process curve operation, without considering the difference in raw material composition, resulting in incomplete separation of impurities during smelting of B-grade raw materials. When the temperature is not up to standard, a "one-size-fits-all" compensation strategy is adopted, without distinguishing fault types, adjusting data is not associated with working conditions for archiving, and similar problems occur repeatedly, increasing the probability of unplanned downtime, and failing to meet the requirements of high-quality production for temperature control flexibility and compensation specificity. SUMMARY

[0007] In view of the above technical deficiencies, the purpose of the present application is to provide a smelting process efficiency optimization and real-time temperature control method and system based on artificial intelligence.

[0008] To solve the above technical problems, the technical scheme adopted by the present application is as follows: In a first aspect, the present application provides a smelting process efficiency optimization and real-time temperature control method based on artificial intelligence, comprising: Step 1, real-time collection of smelting multi-source data and composition: deploying a multi-modal sensing array at each smelting furnace in the target factory, and simultaneously adding a near-infrared spectrum sensor at the inlet of each smelting furnace, so as to collect the process dimension data corresponding to each smelting furnace at each monitoring time point, and associate the raw material dimension data corresponding to each smelting furnace.

[0009] Step 2, cooperative evaluation of smelting process and composition: according to the process dimension data and raw material dimension data corresponding to each smelting furnace at each monitoring time point, the yield prediction value corresponding to each smelting furnace at each monitoring time point is predicted, and it is evaluated whether the heating power of each smelting furnace at each monitoring time point needs to be dynamically adjusted.

[0010] Step 3, composition-oriented process parameter optimization: if the heating power of a certain smelting furnace at a certain monitoring time point needs to be dynamically adjusted, the optimal dynamic temperature curve corresponding to the smelting furnace at the monitoring time point is analyzed, and the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is analyzed.

[0011] Step 4, compensation scheme landing and effect closed loop: the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is sent to the smelting furnace actuator, and the temperature regression time corresponding to the smelting furnace is monitored in real time, and the compensation scheme corresponding to the smelting furnace is analyzed.

[0012] The application provides a method for optimizing and controlling the temperature of a smelting process in real time based on artificial intelligence.

[0013] The method comprises the following steps: a smelting multi-source data and ingredient real-time acquisition module is used to deploy a multi-modal sensing array at each smelting furnace in a target factory, and a near-infrared spectrum sensor is added at the feeding port of each smelting furnace, so that the process dimension data corresponding to each smelting furnace at each monitoring time point and the raw material dimension data corresponding to each smelting furnace are collected.

[0014] A smelting process and ingredient cooperative evaluation module is used to predict the yield prediction value of each smelting furnace at each monitoring time point according to the process dimension data and the raw material dimension data corresponding to each smelting furnace at each monitoring time point, and to evaluate whether the heating power of each smelting furnace at each monitoring time point needs to be dynamically adjusted.

[0015] An ingredient-oriented process parameter optimization module is used to analyze the optimal dynamic temperature curve of a smelting furnace at a monitoring time point and the heating power adjustment instruction of the smelting furnace at the monitoring time point if the heating power of the smelting furnace at the monitoring time point needs to be dynamically adjusted.

[0016] A compensation scheme landing and effect closed loop module is used to send the heating power adjustment instruction of a smelting furnace at a monitoring time point to a smelting furnace actuator, to monitor the temperature regression time length of the smelting furnace in real time, and to analyze the compensation scheme of the smelting furnace.

[0017] The application has the following beneficial effects: 1. The embodiment of the application realizes the real-time correlation collection of process dimension data and raw material dimension data for the first time through the cooperative deployment of a three-dimensional multi-modal sensing array and a feeding port near-infrared spectrum sensor, the sampling frequency of the sensor is increased to 1 Hz, the analysis of the composition of raw materials is shortened from 3-4 hours offline to real time, and the traditional “data fragmentation” problem is completely solved. At the same time, the edge computing node standardizes the processing of four types of core data, forms a four-dimensional state vector input DQN model, and archives historical data in a “furnace-parameters-effect” structure, so that the yield fluctuation is greatly reduced, and high-value data support is provided for subsequent process optimization, and the data reuse rate is improved.

[0017] 2. The embodiment of the application constructs a multi-layer decision system of “XGBoost yield prediction + DQN temperature curve optimization + PID-AI power adjustment”, the yield prediction input is 9-dimensional process-raw material characteristics, the prediction error is reduced, and a scientific basis is provided for power adjustment; the optimal dynamic temperature curve is calculated according to a phased formula, which is suitable for different raw material compositions and equipment states, avoids the purity problem caused by the traditional fixed curve, and the single adjustment amplitude of the heating power adjustment is accurately controlled within 10% of the current power through “PID basic calculation + W value correction + safety constraint verification”, the single-furnace power fluctuation is reduced from ±12% to ±3%, the energy consumption is reduced by 8%-10%, and the energy waste problem of “excessive adjustment” or “insufficient adjustment” is solved.

[0018] 3、The embodiment of the present application breaks through the traditional "one-size-fits-all" compensation mode, divides three time-effect standard types based on temperature regression duration, 5 minutes for the first level, 5-10 minutes for the second level, and 10 minutes for the third level, and corresponding differential compensation schemes are executed, the first level standard is passed through model positive reinforcement to accumulate optimal parameters, the second level standard is passed through local accurate compensation to eliminate temperature unevenness, and the third level standard is passed through cross-furnace parameter migration and root repair to solve complex faults, thereby greatly reducing the unplanned downtime. At the same time, the compensation effect data and the previous step data form a full-link closed-loop archive, reducing the recurrence rate of similar faults, ensuring high-quality production, and further reducing long-term production costs through continuous optimization. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0020] Figure 1 The method embodiment of the present application is a flow chart of the implementation steps.

[0021] Figure 2 The system module connection diagram of the present application is shown. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0023] The embodiment of the present application is shown as follows: Figure 1 A method for optimizing and real-time temperature control of a refining process based on artificial intelligence is shown, comprising: step one, real-time collection of refining multi-source data and components: deploying a multi-modal sensing array at each melting furnace in the target factory, synchronously adding a near-infrared spectrum sensor at the inlet of each melting furnace, thereby collecting process dimension data corresponding to each melting furnace at each monitoring time point, and associating raw material dimension data corresponding to each melting furnace.

[0024] In a specific embodiment, the multi-modal sensor array is deployed at each smelting furnace in the target factory, and a near-infrared spectrum sensor is added synchronously at the inlet of each smelting furnace. The specific deployment process is as follows: the multi-modal sensor array deployed at each smelting furnace in the target factory is deployed according to the three-dimensional coverage principle: a K-type thermocouple is arranged every 1 m along the vertical direction of the furnace wall, 2 groups of dual-band infrared thermal imagers are symmetrically deployed 2 m above the molten pool, and a laser gas analyzer is installed on the side wall of the flue outlet pipeline; a power sensor is embedded in the furnace body distribution box; at the same time, the near-infrared spectrum sensor is installed above the smelting furnace feeding conveyor belt, 30 cm away from the raw material conveying surface, and all sensors are connected to the edge computing node through shielded cables, and the node is connected to the cloud platform through a 5G industrial module.

[0025] It should be noted that the K-type thermocouple is selected to have a ceramic protection tube with a temperature resistance of 1800°C or above to ensure stable operation in a high-temperature environment of the furnace wall; the dual-band infrared thermal imager is selected to have a resolution of 640x512 and a temperature measurement range of 0-2000°C, the lens is equipped with a dust cover and an automatic blowing device to avoid the influence of molten pool dust coverage on temperature measurement accuracy; the laser gas analyzer is selected to have a strong anti-interference ability of TDLAS technology, the sampling probe is installed at an angle of 45° to the flue gas flow direction to reduce probe damage caused by airflow impact; the power sensor is selected to have a Hall current sensor with an accuracy of 0.5 or below, and a safety distance of ≥10 cm is maintained between the sensor and the high-voltage line when embedded in the distribution box to prevent electromagnetic interference; the near-infrared spectrum sensor is selected to have a wavelength range of 900-1700 nm and a resolution of ≤2 nm, and a high-temperature insulation layer is added to the sensor housing to avoid the influence of material heat dissipation of the conveying belt on the detection result. All shielded cables are selected to have flame-retardant twisted shielded cables, which are laid through galvanized steel pipes and are laid separately from the power cable to reduce electromagnetic interference; the edge computing node is selected to have an industrial-grade fanless host with a built-in data encryption module, and the 5G industrial module transmits data through a VPN virtual private network to prevent data leakage or tampering. At the same time, a regular sensor maintenance plan is developed: the K-type thermocouple is calibrated once a month, the infrared thermal imager lens and blowing device are cleaned every quarter, the laser gas analyzer is calibrated every month, and the near-infrared spectrum sensor is calibrated every week with standard samples to ensure the accuracy and reliability of the data collected by each sensor and provide a high-quality data basis for subsequent process analysis and optimization.

[0026] Step two, process and ingredient coordination evaluation: based on the process dimension data and raw material dimension data of each smelting furnace at each monitoring time point, the product yield prediction value of each smelting furnace at each monitoring time point is predicted, and it is evaluated whether each smelting furnace at each monitoring time point needs to dynamically adjust the heating power.

[0027] In a specific embodiment, the predicted yield rate prediction value corresponding to each smelting furnace at each monitoring time point is specifically predicted as follows: a gradient boosting decision tree regression model is used to construct a yield rate prediction model, the training samples of the yield rate prediction model are derived from the historical associated data set of the target plant in the past three years, the number of training samples is ≥5000 groups, and the training samples cover different raw material types of A, B and C levels, and various working conditions of normal production, parameter fluctuation and equipment maintenance; a two-dimensional feature vector of process-raw material is constructed by input features: process dimension data includes 4 types of features such as standard deviation of furnace wall temperature, average temperature of molten pool, average heating power and CO concentration; raw material dimension data includes 5 types of features such as copper content, nickel content and impurity proportion analyzed by near-infrared spectrum sensor, and raw material weight and raw material grade code, a total of 9 input features.

[0028] It should be noted that the standard deviation of the furnace wall temperature: based on the K-type thermocouple deployed on the furnace wall according to the three-dimensional covering principle, the real-time temperature of each thermocouple is collected once every 1 minute, such as 6 measurement point temperatures of 1450℃, 1452℃, 1448℃, 1451℃, 1449℃, 1453℃, first calculate the average value of the 6 measurement points in this minute ((1450+1452+1448+1451+1449+1453) / 6=1450.5℃), and then calculate the deviation of each measurement point temperature from the average value by the standard deviation formula, such as standard deviation=[((1450-1450.5) 2 +(1452-1450.5) 2 +...+(1453-1450.5) 2 ) / 6]≈1.87℃, and the average value of the standard deviation every 5 minutes is taken as the characteristic value of the furnace wall temperature standard deviation at the monitoring time point.

[0029] The average temperature of the molten pool: 2 groups of double-band infrared thermal imagers are symmetrically arranged 2m above the molten pool, and 1 molten pool temperature field thermal map is generated every 30 seconds, the temperature data of 100 uniformly distributed sampling points in the thermal map is extracted by the edge computing node, such as the sampling point temperature range 1445-1455℃, the arithmetic average of the 100 sampling points is calculated, such as 1450.2℃, and the average value every 2 minutes is taken as the characteristic value of the average temperature of the molten pool at the monitoring time point.

[0030] The average heating power: the current and voltage data of the heating loop are collected in real time by the 0.5 level precision Hall current sensor embedded in the furnace body distribution box, the instantaneous power is calculated according to "power=voltage×current×power factor (0.92)", the average value of the instantaneous power in this minute is calculated every 1 minute, such as the instantaneous power fluctuates between 270-280kW in a certain minute, the average value is 275.3kW, and the average value every 5 minutes is taken as the characteristic value of the average heating power at the monitoring time point.

[0031] CO concentration: CO concentration detection value in the flue every 1 minute was output by TDLAS technology laser gas analyzer installed on the side wall of the flue outlet pipe, and the real-time detection value at the monitoring time point was directly taken as the CO concentration characteristic value. If data fluctuation occurred, such as the difference between adjacent two detection values exceeding 20 ppm, the median of the three detection values was taken as the final value.

[0032] Raw material dimension 5 feature acquisition: copper content, nickel content, impurity proportion: near-infrared spectrum sensor was installed 30 cm above the feeding belt of the smelting furnace, and the sensor scanned the surface of the raw material every 2 seconds when the raw material moved with the feeding belt. Through the preset near-infrared spectrum quantitative analysis model, the characteristic values of the corresponding features at the monitoring time point were calibrated based on the offline analysis data of the raw material of the target factory in the past three years. For example, the linear relationship between copper content and absorbance at a specific wavelength of 1200 nm was analyzed to obtain the copper content of 99.2%, the nickel content of 0.5%, and the impurity proportion of 0.3% in a single scan. The average value of the scanning data of each batch of raw material was taken as the corresponding characteristic value at the monitoring time point.

[0033] Raw material weight: a weighing sensor was installed in the middle of the feeding belt to collect the raw material weight per unit length of the feeding belt in real time. Combined with the running speed of the feeding belt, the raw material feeding weight per 1 minute was calculated according to “instantaneous weight = unit length weight × speed × time”, and the total weight of the feeding in the past 30 minutes at the monitoring time point was accumulated as the raw material weight characteristic value.

[0034] Raw material grade coding: based on the batch information of the raw material management system of the target factory, the A, B, and C grades of raw material were respectively coded, such as A grade = 1, B grade = 2, and C grade = 3. When the raw material was fed, the RFID card reader beside the feeding belt read the batch label information of the raw material, and automatically matched the corresponding grade code, such as the current feeding was A grade, and the code was 1, which was directly taken as the raw material grade coding characteristic value to ensure the format adaptation with the process-raw material two-dimensional feature vector.

[0035] During prediction execution, the process-raw material two-dimensional feature vector corresponding to each smelting furnace at each monitoring time point was input into the trained XGBoost model, and the model output the product yield prediction value corresponding to each smelting furnace at each monitoring time point.

[0036] It should be noted that firstly, the process-raw material two-dimensional feature vector collected at the current monitoring time point is preprocessed to ensure consistency with the feature format during model training: Process dimension feature processing: furnace wall temperature standard deviation is kept to 2 decimal places, such as "2.35℃", the average temperature of the molten pool is the average of 100 sampling points within 5 minutes of the infrared thermal imager, rounded to an integer, such as "1352℃", the average heating power is calculated as the average of 12 5-minute power data within the last 1 hour, in kW, rounded to 1 decimal place, such as "285.3kW", and the CO concentration is kept to 3 decimal places, such as "0.215%"; Raw material dimension feature processing: copper content, nickel content, and impurity percentage are kept to 2 decimal places, such as "98.52%" "0.85%" "0.63%", the raw material weight is accurate to 0.1 kg, such as "800.5kg", and the raw material grade code is converted to an integer according to "A grade = 3, B grade = 2, C grade = 1"; The 9-dimensional feature vector after preprocessing is arranged in the order of "process dimension (4 items) → raw material dimension (5 items)", forming a standardized input array, such as "[2.35, 1352, 285.3, 0.215, 98.52, 0.85, 0.63, 800.5, 3]".

[0037] The trained XGBoost model is called through the model service interface of the cloud AI platform, and the specific calling process is as follows: interface parameter setting: the input parameters include "smelting furnace ID", "monitoring timestamp", "standardized feature array", and the model version is specified, the default is to call the latest iteration version, the version number format is "V+date+batch number", such as "V2025092501"; Model calculation process: the model first loads the pre-trained decision tree parameters, including node splitting threshold and leaf node weight, and traverses the input feature array dimension by dimension: 1. According to the decision tree level, it is judged whether the feature value meets the splitting condition, such as "molten pool average temperature > 1350℃", then enter the left subtree, otherwise enter the right subtree; 2. After traversing all 100 decision trees, the output prediction value of each tree is summarized, and the preliminary prediction result is obtained by weighted summation according to the sub-sample proportion (0.8); 3. The preliminary prediction result is processed by activation function, and the Sigmoid function is mapped to the range of 0-100%, and the final output is the yield prediction value, kept to 2 decimal places, such as "96.85%"; Time-consuming control: through cloud GPU acceleration calculation, ensure that the time-consuming of single prediction is ≤500ms, meet the real-time demand, the monitoring time interval is 5 minutes, and sufficient processing time is reserved.

[0038] After completing the prediction result verification, the following operations are performed: data storage: "smelting furnace ID, monitoring timestamp, standardized feature array, prediction value, deviation mark, verification status" are packaged as JSON format data and stored in time series database such as InfluxDB, while the original process data and raw material data of the monitoring time point are associated to form a complete traceability chain.

[0039] In a specific embodiment, the evaluation of whether the smelting furnace at each monitoring time point needs dynamic adjustment of heating power is specifically evaluated as follows: the product yield prediction value and the process dimension data of the smelting furnace at each monitoring time point are compared with the set product yield prediction value threshold and process dimension data threshold respectively, if the product yield prediction value is less than 92% or any one of the process dimension data is greater than the set threshold, it is evaluated that the smelting furnace at the monitoring time point needs dynamic adjustment of heating power, if the product yield prediction value is greater than or equal to 92%, and at the same time, all process dimension data are less than or equal to the set threshold, it is evaluated that the smelting furnace at the monitoring time point does not need dynamic adjustment of heating power.

[0040] Step three, component-oriented process parameter optimization: if the smelting furnace at a certain monitoring time point needs dynamic adjustment of heating power, the optimal dynamic temperature curve corresponding to the smelting furnace at the monitoring time point is analyzed, and the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is analyzed.

[0041] In a specific embodiment, the analysis of the optimal dynamic temperature curve corresponding to the smelting furnace at the monitoring time point is specifically analyzed as follows: A1, obtaining the molten pool temperature Q value, the raw material component deviation W value, the heating power E value and the smelting time R value corresponding to the state vector of each dimension of the smelting furnace at the monitoring time point.

[0042] A2, if the smelting furnace at the monitoring time point is in the heating stage, the heating stage curve corresponding to the smelting furnace at the monitoring time point is: the heating target temperature is calculated based on the molten pool temperature Q value, according to the target temperature = the basic melting temperature 1450 + (Q value-7.0) x 10; the heating rate is calculated based on the heating power E value, according to the rate = the basic rate 4 + (E value-6.5) x 0.8; the remaining heating time is calculated based on the smelting time R value, the current actual temperature is 1380℃, the remaining heating temperature difference = the heating target temperature - the current actual temperature, the remaining heating time = the remaining heating temperature difference ÷ 4.5, that is, the heating stage will continue to: the smelting time R value + the remaining heating time, taking the actual time corresponding to the smelting time R value as the current starting point, and the actual temperature at the starting point as the starting temperature, the temperature curve of temperature increasing with time is drawn minute by minute according to the heating rate, forming a continuous heating stage optimal dynamic temperature curve.

[0043] A2, if the smelting furnace is in the holding stage at the monitoring time point, the corresponding temperature rising stage curve of the smelting furnace at the monitoring time point is: holding temperature reference: based on the molten pool temperature Q value, the holding reference temperature is calculated as holding reference temperature = temperature rising stage target temperature + (Q value - 7.5) x 5; temperature fluctuation range: based on the raw material composition deviation W value, the fluctuation range is calculated as fluctuation range = ±3 - (W value - 6.5) x 0.5; holding time length: based on the smelting time length R value, the total holding time length is calculated as holding stage total time length = 80 - (R value - 6.0) x 10; taking the actual time corresponding to the smelting time length R value as the starting point and the holding reference temperature as the center, a temperature curve with small fluctuations is drawn within a fluctuation range of ±2.8℃, combined with the power compensation demand, to form an optimal dynamic temperature curve of the holding stage.

[0044] A3, if the smelting furnace is in the cooling stage at the monitoring time point, the corresponding temperature rising stage curve of the smelting furnace at the monitoring time point is: cooling starting point temperature: based on the holding stage reference temperature, the cooling starting point is calculated as cooling starting point = holding reference temperature - (Q value - 7.5) x 2; cooling rate: based on the heating power E value, the rate is calculated as rate = 3 + (E value - 6.5) x 2; remaining cooling time length: based on the smelting time length R value, the cooling end point temperature, the remaining cooling temperature difference = cooling starting point temperature - cooling end point temperature, the remaining cooling time length = remaining cooling temperature difference ÷ 3.8; cooling stage end time: smelting time length R value + remaining cooling time length; taking the actual time corresponding to the smelting time length R value as the starting point and the cooling starting point temperature as the starting temperature, a curve is drawn with the temperature decreasing over time at a cooling rate, to form an optimal dynamic temperature curve of the cooling stage.

[0045] In a specific embodiment, the smelting furnace each dimension state vector corresponding molten pool temperature Q value, raw material composition deviation W value, heating power E value and smelting time length R value at the monitoring time point are obtained as follows: 4 types of core data corresponding to the smelting furnace at the monitoring time point are collected and standardized to form the state vector of the input model: standardized molten pool temperature, standardized raw material composition deviation, standardized heating power and standardized smelting time length; the standardized molten pool temperature, standardized raw material composition deviation, standardized heating power and standardized smelting time length data are integrated in the order of molten pool temperature, composition deviation, power and time length to form a 4-dimensional state vector corresponding to the smelting furnace at the monitoring time point; the 4-dimensional state vector corresponding to the smelting furnace at the monitoring time point is input into the DQN model to calculate the smelting furnace each dimension state vector corresponding molten pool temperature Q value, raw material composition deviation W value, heating power E value and smelting time length R value at the monitoring time point.

[0046] It should be noted that the calculation process of the molten pool temperature Q value, the raw material composition deviation W value, the heating power E value and the smelting time R value is as follows: the standardized molten pool temperature is S1: the average temperature of the molten pool infrared thermal imager within 5 minutes is collected, standardized according to "S1=(actual temperature- lowest melting temperature 1200℃) / (highest tolerance temperature 1800℃-1200℃)", and the value range is mapped to 0-1.

[0047] The standardized raw material composition deviation is S2: based on the copper, nickel and impurity content analyzed by the near-infrared spectrum sensor, the comprehensive composition deviation is calculated, such as copper content deviation-0.2%, nickel content deviation+0.1%, impurity content deviation+0.3%, comprehensive deviation=(-0.2+0.1+0.3) / 3=0.067%), standardized according to "S2=(comprehensive deviation-minimum deviation-2%) / (maximum deviation 2%-(-2%))", and the value range is 0-1.

[0048] The standardized heating power is S3: the real-time output power of the power sensor is collected, standardized according to "S3=(actual power-minimum maintenance power 150kW) / (rated maximum power 500kW-150kW)", and the value range is 0-1.

[0049] The standardized smelting time is S4: standardized according to "S4=smelting time / total smelting time", and the value range is 0-1.

[0050] The final 4-dimensional standardized state vector is formed: [S1, S2, S3, S4], formatted as a TensorFlow-compatible float32 type array, as the input of the DQN model.

[0051] The DQN model adopts a "convolutional layer + fully connected layer" network structure, which maps the 4-dimensional state vector to Q value, W value, E value and R value through pre-trained weight parameters, with a value range of 0-10, which meets the process evaluation requirements. The specific steps are as follows: input layer: receiving 4-dimensional state vector, converting to "[1,4]" feature matrix through Reshape layer, adapting to the input format of fully connected layer.

[0052] Fully connected layer 1 (64 neurons): using ReLU activation function, the calculation formula is "F1=ReLU(W1×X+B1)", where W1 is a 64×4 weight matrix (such as a row weight [0.2, 0.5, 0.1, 0.2]), X is the input feature matrix, B1 is a 64-dimensional bias vector (such as [0.1, 0.05,..., 0.08]), and a 64-dimensional feature vector F1 is output.

[0053] Fully connected layer 2 (32 neurons): same ReLU activation function, formula: "F2 = ReLU(W2 x F1 + B2)", W2 is a 32 x 64 weight matrix, B2 is a 32-dimensional bias vector, output a 32-dimensional feature vector F2, complete feature compression and key information extraction.

[0054] The output layer includes 4 independent linear neurons, corresponding to Q value, W value, E value, R value respectively, through "linear mapping + range scaling" to convert the hidden layer output into 0-10 process evaluation value: all weight matrices (W1, W2, W_Q, W_W, W_E, W_R) and bias vectors (B1, B2, B_Q, B_W, B_E, B_R) are obtained by training the target factory 3-year historical data, the training process takes "maximum yield + minimum energy consumption" as the dual objective function, and the model loss rate (MSE) is stable below 0.02.

[0055] Melt pool temperature Q value calculation: the calculation formula of output layer neuron 1 is "Q_raw = W_Q x F2 + B_Q", wherein W_Q is a 1 x 32 weight vector (such as [0.3, 0.15,..., 0.2]), B_Q is a bias term (such as 0.2), and Q_raw (example: Q_raw = 0.78) is obtained; scaled to 0-10 range according to "Q value = Q_raw x 10" (example: Q = 0.78 x 10 = 7.8), corresponding to melt pool temperature adaptability evaluation (7.8 points, indicating that the temperature is close to the optimal interval).

[0056] Raw material composition deviation W value calculation: the calculation formula of output layer neuron 2 is "W_raw = W_W x F2 + B_W", W_W is a 1 x 32 weight vector, and B_W is a bias term (example: W_raw = 0.68); scaled according to "W value = W_raw x 10" (example: W = 0.68 x 10 = 6.8).

[0057] Heating power E value calculation: the calculation formula of output layer neuron 3 is "E_raw = W_E x F2 + B_E", W_E is a 1 x 32 weight vector, and B_E is a bias term (example: E_raw = 0.69); scaled according to "E value = E_raw x 10" (example: E = 0.69 x 10 = 6.9).

[0058] Melted time R value calculation: the calculation formula of output layer neuron 4 is "R_raw = W_R x F2 + B_R", W_R is a 1 x 32 weight vector, and B_R is a bias term (example: R_raw = 0.4); scaled according to "R value = R_raw x 10" (example: R = 0.4 x 10 = 4.0).

[0059] In a specific embodiment, the analysis of the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is specifically as follows: first, the current actual temperature of the smelting furnace at the monitoring time point and the target temperature of the corresponding time point of the optimal dynamic temperature curve are obtained, and the temperature deviation = target temperature - current actual temperature is calculated; the basic power adjustment amount is calculated by using a traditional PID controller, and the PID core parameters are set as: proportional coefficient Kp = 2.5, integral coefficient Ki = 0.1, and differential coefficient Kd = 0.5, and the calculation formula is: basic power adjustment amount = proportional coefficient Kp x temperature deviation + integral coefficient Ki x past 5-minute temperature deviation total value + differential coefficient Kd x (temperature deviation current value - temperature deviation 1-minute ago value).

[0060] Then, the raw material composition deviation W value of the smelting furnace at the monitoring time point is obtained, and the raw material composition deviation W value of the smelting furnace at the monitoring time point is compared with the set standard raw material composition deviation W value, if the raw material composition deviation W value of the smelting furnace at the monitoring time point is lower than the set standard raw material composition deviation W value by 0.1%, the basic adjustment amount is increased by 1.5%; if it is higher than the set standard raw material composition deviation W value by 0.1%, the basic adjustment amount is reduced by 1%, and the adjusted basic adjustment amount is recorded as the heating power adjustment value corresponding to the smelting furnace at the monitoring time point.

[0061] Finally, the heating power E value of the smelting furnace at the monitoring time point is obtained, and the power adjustment safety constraint is set: the single adjustment amplitude is not more than 10% of the heating power E value of the smelting furnace at the monitoring time point; the upper limit of the adjusted power is not more than 90% of the rated maximum power of the smelting furnace at the monitoring time point; the lower limit of the adjusted power is not less than the minimum power of the smelting furnace at the monitoring time point; and finally the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is generated.

[0062] Step four, compensation scheme landing and effect closed loop: the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point is sent to the smelting furnace actuator, and the temperature regression time corresponding to the smelting furnace is monitored in real time, and the compensation scheme corresponding to the smelting furnace is analyzed.

[0063] In a specific embodiment, the analysis of the compensation scheme corresponding to the smelting furnace is specifically as follows: B1, the time effectiveness standard type corresponding to the smelting furnace is obtained, the time effectiveness standard type includes first-level optimization standard, second-level compensation standard and third-level collaborative intervention, if the time effectiveness standard level corresponding to the smelting furnace is first-level optimization standard, the no additional compensation + model positive reinforcement scheme is executed.

[0064] It should be noted that the no additional compensation + model positive reinforcement scheme is executed when the temperature of the smelting furnace returns to the optimal curve range within 5 minutes. The core is to optimize the model through data sedimentation. When implementing, first, the edge computing node automatically extracts the full link data of the heat: including the 4-dimensional standardized state vector, the Q value output by the DQN model, the W value and other evaluation parameters, as well as the temperature regression process data and energy consumption change data, and marks these data as "high confidence positive samples"; then, through the 5G industrial module, the samples are encrypted and synchronized to the plant federal learning node. In the monthly model iteration, the sample is assigned a weight of 1.3 times, focusing on strengthening the state-action mapping relationship in the "Q value, W value" interval of the DQN model, while adjusting the feature weight of the XGBoost yield prediction model; finally, a structured "no compensation confirmation report" is generated, recording key associated information such as "A-grade copper raw material, heating rate 4.2°C / min, yield 94.5%", and storing it in the process knowledge graph "optimal case library". When calling parameters for smelting furnaces with the same working conditions in the future, the system can directly match the case library data, the decision-making time is shortened from the original 5 seconds to 2 seconds, and no additional power adjustment is required, avoiding energy waste.

[0065] B2, if the time compliance level of the smelting furnace is secondary compensation compliance, execute the local precise compensation + parameter fine-tuning solidification scheme.

[0066] It should be noted that the local precise compensation + parameter fine-tuning solidification scheme is executed when the temperature of the smelting furnace returns to the optimal curve range within 5-10 minutes. First, use the temperature field thermal map generated by the dual-band infrared thermal imager directly above the molten pool to locate the local low temperature area; second, according to the pre-set "local temperature deviation-power adjustment" mapping rule, increase the heating power of the corresponding area by 0.5% for every 0.1°C drop, calculate the local compensation amount: the current power of the right side heating unit of the furnace wall is 275kW, which needs to be increased by 0.5% x (0.8 / 0.1) = 4%, i.e. 11kW, the adjusted power is 286kW, generate local compensation instructions and send them to the smelting furnace partition power controller; third, after the compensation is executed, the local temperature data is collected by the furnace wall thermocouple every 10 seconds, and the temperature deviation is monitored for 3 minutes to ensure that the temperature deviation of the area is stable within ±0.5°C; after the compensation is completed, the parameters such as "right side of the furnace wall, W value, local power increase by 4%" and the correlation between the raw material composition and the furnace area are solidified to the process knowledge graph, and a "raw material composition deviation-local area-power fine-tuning amount" mapping table is established. In the future, when the left side of the furnace wall has a 0.6°C low temperature and the W value is 6.5, the system can directly call the parameters in the table without the need to recalculate.

[0067] B3, if the time compliance level of the smelting furnace is three-level collaborative intervention, execute the root cause repair compensation + cross-furnace parameter migration scheme.

[0068] It needs to be explained that the root cause repair compensation + cross furnace parameter migration scheme is executed when the temperature does not return to the optimal curve range within 10 minutes of the smelting furnace, and the core is to solve the root problem and reuse similar experience. When implementing, first, combine multi-modal sensing data and digital twin model to locate the fault root cause: if the power sensor monitors that the actual output power of the heating pipe is 260kW, which is 15kW lower than the rated 280kW, it is determined that it is a device problem, and then the power upper limit of the heating pipe is adjusted to 300kW, and a device maintenance work order is pushed, marked as "heating pipe usage time 8000 hours, recommend replacing within 72 hours"; if the near-infrared spectrum sensor detects that the content of a certain raw material suddenly decreases by 0.8%, it is determined that it is a raw material problem, then the raw material ratio of the feeding port is adjusted from 10% to 15%, and the holding time is extended by 5 minutes; the second step is to call the real-time desensitization data of the furnace F002 of the same type and the same raw material batch in the factory area, extract the effective compensation parameters under the "raw material content low by 0.7%" scene, and combine the current furnace aging state to fine-tune the holding temperature increase to 8℃, and generate the final compensation scheme; the third step is to monitor the temperature and yield change within 15 minutes after compensation execution to ensure that the temperature deviation is ≤±1℃; finally, the "device aging-power upper limit adjustment-cross furnace reference furnace F002" and other information are entered into the knowledge graph "fault response library", and the DQN model fault handling branch is updated, so that the self-compensation success rate of subsequent similar faults is increased from the original 75% to more than 90%.

[0069] In a specific embodiment, the time compliance type corresponding to the smelting furnace is obtained, and the specific obtaining process is as follows: the temperature return time corresponding to the smelting furnace is obtained, if the temperature of the smelting furnace returns to the optimal curve range within 5 minutes, it is recorded as first-level optimization compliance, if the temperature of the smelting furnace returns to the optimal curve range within 5-10 minutes, it is recorded as second-level compensation compliance, and if the temperature of the smelting furnace does not return to the optimal curve range within 10 minutes, it is recorded as third-level collaborative intervention.

[0070] The embodiment of the present application comprises Figure 2 As shown in the figure, an artificial intelligence-based smelting process efficiency optimization and real-time temperature control system comprises a smelting multi-source data and ingredient real-time acquisition module: a multi-modal sensing array is deployed at each smelting furnace in the target factory, and a near-infrared spectrum sensor is added at the feeding port of each smelting furnace, so that the process dimension data corresponding to each smelting furnace is collected at each monitoring time point, and the raw material dimension data corresponding to each smelting furnace is associated.

[0071] A smelting process and ingredient cooperative evaluation module: according to the process dimension data and raw material dimension data corresponding to each smelting furnace at each monitoring time point, the yield prediction value corresponding to each smelting furnace at each monitoring time point is predicted, and whether each smelting furnace needs to dynamically adjust the heating power at each monitoring time point is evaluated.

[0072] The component-oriented process parameter optimization module is configured to analyze the optimal dynamic temperature curve corresponding to the smelting furnace at the monitoring time point and analyze the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point when the smelting furnace needs to dynamically adjust the heating power at the monitoring time point.

[0073] The compensation scheme landing and effect closed loop module is configured to send the heating power adjustment instruction corresponding to the smelting furnace at the monitoring time point to the smelting furnace executor, monitor the temperature regression duration corresponding to the smelting furnace in real time, and analyze the compensation scheme corresponding to the smelting furnace.

[0074] The above is merely an example and description of the concept of the present application, and those skilled in the art can make various modifications or supplements to the described specific embodiments or replace them with similar ways, as long as they do not deviate from the concept of the present application or exceed the scope defined in the specification, which shall belong to the protection scope of the present application.

Claims

1. A method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence, characterized in that, include: Step 1: Real-time acquisition of multi-source data and composition of raw materials: Deploy multi-modal sensor arrays in each smelting furnace of the target plant, and simultaneously add near-infrared spectral sensors at the feed inlet of each smelting furnace to collect process dimension data corresponding to each smelting furnace at each monitoring time point, and associate it with raw material dimension data corresponding to each smelting furnace. Step 2: Synergistic evaluation of refining process and composition: Based on the process dimension data and raw material dimension data corresponding to each smelting furnace at each monitoring time point, predict the yield value corresponding to each smelting furnace at each monitoring time point, and evaluate whether the heating power of each smelting furnace at each monitoring time point needs to be dynamically adjusted. Step 3: Composition-oriented process parameter optimization: If a smelting furnace needs to dynamically adjust its heating power at a certain monitoring time point, analyze the optimal dynamic temperature curve corresponding to the smelting furnace at that monitoring time point, and analyze the heating power adjustment command corresponding to the smelting furnace at that monitoring time point. Step 4: Implementation and Effect Closure of Compensation Scheme: Send the heating power adjustment command corresponding to the smelting furnace at the monitoring time point to the smelting furnace actuator, monitor the temperature return time corresponding to the smelting furnace in real time, and analyze the compensation scheme corresponding to the smelting furnace.

2. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 1, characterized in that, The deployment of multimodal sensor arrays in each smelting furnace of the target plant, along with the simultaneous addition of near-infrared spectral sensors at the feed inlet of each furnace, is described in the following steps: The target factory deploys multimodal sensor arrays in each smelting furnace according to the principle of three-dimensional coverage: one K-type thermocouple is installed every 1m along the vertical direction of the furnace wall, two sets of dual-band infrared thermal imagers are symmetrically deployed 2m directly above the molten pool, and a laser gas analyzer is installed on the side wall of the flue outlet pipe; power sensors are embedded in the furnace body power distribution box; at the same time, near-infrared spectral sensors are installed directly above the smelting furnace feed conveyor belt, 30cm away from the raw material conveying surface. All sensors are connected to edge computing nodes through shielded cables, and the nodes are connected to the cloud platform through 5G industrial modules.

3. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 2, characterized in that, The predicted yield values ​​for each smelting furnace at each monitoring time point are determined as follows: A gradient boosting decision tree regression model was used to construct a yield prediction model. The training samples of the yield prediction model came from the target plant’s refining history association dataset over the past 3 years, with a training sample size of ≥5000 sets, covering different raw material types of Grade A, Grade B, and Grade C, as well as various operating conditions such as normal production, parameter fluctuations, and equipment maintenance. The input features construct a two-dimensional feature vector of process and raw materials: the process dimension data includes four types of features: standard deviation of furnace wall temperature, average temperature of molten pool, average heating power and CO concentration; the raw material dimension data includes five types of features: copper content, nickel content and impurity ratio as resolved by near-infrared spectroscopy sensor, as well as raw material weight and raw material grade code, for a total of nine input features; During prediction, the process-raw material two-dimensional feature vectors corresponding to each smelting furnace at each monitoring time point are input into the well-trained XGBoost model, and the model outputs the predicted yield value corresponding to each smelting furnace at each monitoring time point.

4. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 3, characterized in that, The assessment process for determining whether each smelting furnace needs dynamic adjustment of its heating power at each monitoring time point is as follows: The predicted yield and process dimension data for each smelting furnace at each monitoring time point are compared with the set thresholds for predicted yield and process dimension data. If the predicted yield is less than 92% or any process dimension data is greater than the set threshold, the smelting furnace is assessed to need dynamic adjustment of heating power at that monitoring time point. If the predicted yield is greater than or equal to 92% and all process dimension data are less than or equal to the set threshold, the smelting furnace is assessed to not need dynamic adjustment of heating power at that monitoring time point.

5. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 4, characterized in that, The analysis of the optimal dynamic temperature curve of the smelting furnace at the monitoring time point is as follows: A1. Obtain the molten pool temperature Q, raw material composition deviation W, heating power E, and smelting time R corresponding to the state vector of the smelting furnace in each dimension at the monitoring time point; A2. If the smelting furnace is in the heating stage at the monitoring time point, then the heating stage curve corresponding to the smelting furnace at the monitoring time point is as follows: Target heating temperature: calculated based on the molten pool temperature Q value, according to the formula: Target temperature = Basic melting temperature 1450 + (Q value - 7.0) × 10; Heating rate: calculated based on the heating power E value, according to the formula: Rate = Basic rate 4 + (E value - 6.5) × 0.8; Remaining heating time: based on the smelting time R value, the current actual temperature is 1380℃, the remaining heating temperature difference = Target heating temperature - Current actual temperature, and the remaining heating time = Remaining heating temperature difference ÷ 4.

5. That is, the heating stage will continue until: The smelting time R value + the remaining heating time. Taking the actual time corresponding to the smelting time R value as the current starting point and the current actual temperature as the starting temperature, the temperature increases with time minute by minute according to the heating rate, forming a continuous optimal dynamic temperature curve for the heating stage. A2. If the smelting furnace is in the heat preservation stage at the monitoring time point, then the corresponding heating stage curve for the smelting furnace at the monitoring time point is as follows: Heat preservation temperature reference: calculated based on the molten pool temperature Q value, according to heat preservation reference temperature = heating stage target temperature + (Q value - 7.5) × 5; Temperature fluctuation range: calculated based on the raw material composition deviation W value, according to fluctuation range = ±3 - (W value - 6.5) × 0.5; Heat preservation time: Based on the melting time R value, it is calculated as follows: total heat preservation time = 80 - (R value - 6.0) × 10. Taking the actual time corresponding to the melting time R value as the starting point and the heat preservation reference temperature as the center, within the fluctuation range of ±2.8℃, a small fluctuation temperature curve is plotted in combination with the power compensation requirements to form the optimal dynamic temperature curve for the heat preservation stage. A3. If the smelting furnace is in the cooling stage at the monitoring time point, then the corresponding heating stage curve for the smelting furnace at the monitoring time point is as follows: Cooling start temperature: Based on the reference temperature of the holding stage, it is obtained by calculating the cooling start temperature = holding reference temperature - (Q value - 7.5) × 2; Cooling rate: Based on the heating power E value, it is calculated by calculating the rate = 3 + (E value - 6.5) × 2; Remaining cooling time: Based on the smelting time R value, the cooling end temperature, the remaining cooling temperature difference = cooling start temperature - cooling end temperature, the remaining cooling time = remaining cooling temperature difference ÷ 3.8; Cooling stage end time: The smelting time R value + the remaining cooling time, taking the actual time corresponding to the smelting time R value as the starting point, the cooling start temperature as the starting temperature, and plotting the temperature decreasing with time minute by minute according to the cooling rate, forming the optimal dynamic temperature curve of the cooling stage.

6. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 5, characterized in that, The specific process for obtaining the molten pool temperature Q, raw material composition deviation W, heating power E, and smelting time R corresponding to the state vectors of the smelting furnace at each monitoring time point is as follows: Four types of core data corresponding to the smelting furnace at the monitoring time point are collected and standardized to form a state vector for the input model: standardized molten pool temperature, standardized raw material composition deviation, standardized heating power, and standardized smelting time. The standardized molten pool temperature, standardized raw material composition deviation, standardized heating power, and standardized smelting time data are integrated in the order of molten pool temperature, composition deviation, power, and time to form a 4-dimensional state vector corresponding to the smelting furnace at the monitoring time point. The 4-dimensional state vector corresponding to the smelting furnace at the monitoring time point is input into the DQN model to calculate the molten pool temperature Q value, raw material composition deviation W value, heating power E value, and smelting time R value corresponding to each dimension of the smelting furnace at the monitoring time point.

7. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 6, characterized in that, The analysis of the heating power adjustment command corresponding to the smelting furnace at the monitoring time point is as follows: First, obtain the current actual temperature of the smelting furnace at the monitoring time point and the target temperature at the corresponding time point of the optimal dynamic temperature curve, and calculate the temperature deviation = target temperature - current actual temperature; use a traditional PID controller to calculate the basic power adjustment, and set the core PID parameters as: proportional coefficient Kp = 2.5, integral coefficient Ki = 0.1, and derivative coefficient Kd = 0.

5. The calculation formula is: basic power adjustment = proportional coefficient Kp × temperature deviation + integral coefficient Ki × total temperature deviation value of the past 5 minutes + derivative coefficient Kd × (current temperature deviation value - temperature deviation value of the previous minute). Then, the raw material composition deviation W value of the smelting furnace at the monitoring time point is obtained, and the raw material composition deviation W value of the smelting furnace at the monitoring time point is compared with the set standard raw material composition deviation W value. If the raw material composition deviation W value of the smelting furnace at the monitoring time point is 0.1% lower than the set standard raw material composition deviation W value, the basic adjustment amount is increased by 1.5%; if it is 0.1% higher, the basic adjustment amount is decreased by 1%. The adjusted basic adjustment amount is recorded as the heating power adjustment value of the smelting furnace at the monitoring time point. Finally, the heating power E value of the smelting furnace at the monitoring time point is obtained, and power adjustment safety constraints are set: single adjustment range: not exceeding 10% of the heating power E value of the smelting furnace at the monitoring time point; upper limit of adjusted power: not exceeding 90% of the rated maximum power of the smelting furnace at the monitoring time point; lower limit of adjusted power: not lower than the minimum power of the smelting furnace maintained at the monitoring time point; finally, the heating power adjustment command corresponding to the smelting furnace at the monitoring time point is generated.

8. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 7, characterized in that, The analysis of the compensation scheme corresponding to the smelting furnace is as follows: B1. Obtain the timeliness achievement type corresponding to the smelting furnace. The timeliness achievement type includes Level 1 optimization achievement, Level 2 compensation achievement and Level 3 collaborative intervention. If the timeliness achievement level corresponding to the smelting furnace is Level 1 optimization achievement, then execute the no additional compensation + model positive reinforcement scheme. B2. If the aging compliance level corresponding to the smelting furnace is Level II compensation compliance, then implement the local precise compensation + parameter fine-tuning solidification scheme. B3. If the aging compliance level corresponding to the smelting furnace is Level 3 collaborative intervention, then the root cause repair compensation + cross-furnace parameter migration scheme shall be implemented.

9. The method for optimizing refining process efficiency and real-time temperature control based on artificial intelligence as described in claim 8, characterized in that, The specific process for obtaining the aging compliance type corresponding to the smelting furnace is as follows: Obtain the temperature regression time corresponding to the smelting furnace. If the temperature of the smelting furnace returns to the optimal curve range within 5 minutes, it is recorded as Level 1 optimization achievement. If the temperature of the smelting furnace returns to the optimal curve range within 5 to 10 minutes, it is recorded as Level 2 compensation achievement. If the temperature of the smelting furnace does not return to the optimal curve range within 10 minutes, it is recorded as Level 3 collaborative intervention.

10. An AI-based refining process efficiency optimization and real-time temperature control system for implementing the AI-based refining process efficiency optimization and real-time temperature control method according to any one of claims 1-9, characterized in that, include: Real-time acquisition module for multi-source data and composition of raw materials: It is used to deploy multi-modal sensor arrays in each smelting furnace of the target plant and simultaneously add near-infrared spectral sensors at the feed inlet of each smelting furnace, so as to collect the corresponding process dimension data of each smelting furnace at each monitoring time point and associate it with the corresponding raw material dimension data of each smelting furnace. The refining process and composition co-evaluation module is used to predict the yield of each smelting furnace at each monitoring time point based on the process dimension data and raw material dimension data corresponding to each smelting furnace at each monitoring time point, and to evaluate whether the heating power of each smelting furnace at each monitoring time point needs to be dynamically adjusted. The composition-oriented process parameter optimization module is used to analyze the optimal dynamic temperature curve of the smelting furnace at a certain monitoring time point and the heating power adjustment command of the smelting furnace at that monitoring time point when the heating power of the smelting furnace needs to be dynamically adjusted at a certain monitoring time point. The compensation scheme implementation and effect closed-loop module is used to send the heating power adjustment command corresponding to the smelting furnace at the monitoring time point to the smelting furnace actuator, monitor the temperature return time corresponding to the smelting furnace in real time, and analyze the compensation scheme corresponding to the smelting furnace.

Images